'N`/hYDQ$G%G%((c*c*,\,..0h022-3T -3]6 ]6S70 S7<X <W?l W?AACDCF*F]H]HW  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=Woperation.0exchange : zbreasonOperatorInputsinvocations1T   $$,,4,44<<DDLLTLTT\\ddlltltt||   $$,,4,44<<DDLLTLTT\\ddlltltt|| bJwoperation.0copy_u645:5R929929"9B9Z=bAAI*IRMBreasonLocalCopyvalue22bytes512patternContiguousinvocations1uniformInvocationstruearguments[10, 4] b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations15e6VVvev bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements96input0Elements96input1Elements288usefulCycles72physicalWorkCycles72workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 96] bJ7Moperation.1exchange :z2breasonOperatorInputsvalue14invocations1$+33;57;AAEEMMQS[TV bwoperation.1static_copy_u325:5R9299299B9Z=bAAI*IRM"reasonLocalCopyvalue15bytes32patternContiguousinvocations1uniformInvocationstruearguments[8] boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36]  broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4]  bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`8 !)'/<>>@@HJR[c  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7operation.2exchange :z2breasonOperatorInputsvalue19invocations1  bzoperation.2reduce_sum_f16\" b ))*r+:,- - . Z/ "0 01z2B3 445b6*7789J:;;<j=2>> ?!@"RA#B$B%C&rD':E(F)F*G+ZH,"I-I.J/zK0BL1 M2M3N4bO5*P6P7Q8R9JS:T;T<U=jV>2W?W@XAvBRwC DҕEFbG*HIJKJLMNOZP"QRSzTBU VWXb Y* Z [ \ ]J ^_`ajb2cdefRghijrk:lmnoZp"qrsztBu v w!xb"y*#z#{$|%}J&~''(JFG"ffgzhBi jjkbl*mmnoJpqqrjs2ttuvRwxxyrz:{||}Z~"zB ҃b*Jڊj2ŽRr:ʕb*Jj2Rr:Z"zB b*Jj2Rr:Z"z!"8"   z B       b *Jj2Rr: !" #Z!$""%"&#'z$(B%) &*&+',b(-*).)/*0+1J,2-3-4.5j/620708192:R3;4<4=5>r6?:7@8AzUBBVCsDtEZuF"vGvHwIzxJByK zLzM{Nb|O*}P}Q~RSJTUځVWjX2YZ…[\R]^_`ra:bcʌdeZf"ghizjBk lғmnbo*pqrsJtuښvwjx2yzž{|R}~r*Jj2Rr:Z"zB b*Jj2Rr:Z "   z B  :,-jJ2KKLMRNOOPrQ:RSSTZU"VVWzXBY ZZ[b\*]]^_J`aabjc2ddefRghhirj:kllmZn"oopzqBr sstbu*vvwxJyzzBr:ʸ Z "   zB ҿb*Jj2 R!"#$r%:&'()Z*"+,-z.B/ 012b3*4567J89:;j<2=>?@RABC D!Ez"FB#G $H$I%Jb&K*'L'M(N)OJ*P+Q+R,Sj-T2.U.V/W0XR1Y2Z2[3\r4]:5^6_6`7aZ8b"9c9d:ez;fB<g =h=i>jb?k*@l@mAnBoJCpDqDrEsjFt2GuGvHwIxRJyKzK{L|rM}:N~OOPmnp:Z "!!"z#B$ %%&b'*(()*J+,,-j.2//01R2334r5:6778Z9"::;z<B= >>?b@*AABCJDEEFjG2HHIJRKLiJjzB Ҋb*Jڑj2•Rr:ʜZ"zB ңb*Jڪj2®Rr:ʵZzb*  J   j2Rr:Z" z !B " # $ %b &*'()*J+,-.j/20123R4567r8:9:;<Z=" > ?!@z"A?B@C]D^E_FJ`GaHaIbJjcK2dLdMeNfORgPhQhRiSrjT:kUlVlWmXZnY"oZo[p\zq]Br^ s_s`tabub*vcvdwexfJygzhzi{jj|k2}l}m~noRpqrsrt:uvʅwxZy"z{|z}B~ ҌJj2Rr:Z"zB b*Jj2Rr:Z"zB %.ZM"VVWzXBY ZZ[b\*]]^_J`aabjc2ddefRghhirj:kllmZn"oopzqBr sstbu*vvwxJyzz{j|2}}~Rr:ʍZ"zB   Ҭ  b *Jڳj2·Rr: !ʾ"#Z$"%&'z(B) *+,b-*./01J2345j62789:R;<=>r?:@ABCZD"EFGzHBI JKLbM*NOPQJRSTUjV2WX Y ZR [ \ ] ^r_:`abcZd"efgzhBi jklbm*nopqJrstuj v2!w!x"y#zR${%|%}&~r':()1:ZC"LLMzNBO PPQbR*SSTUJVWWXjY2ZZ[\R]^^_r`:abbcZd"eefzgBh iijbk*llmnJoppqjr2sstuRvwwxry:z{ʃZ"zB Ңb*Jکj2­Rr:ʴZ"zB һb*Jj2Rr:Z"zB     b *Jj2Rr: !"#Z$" % & 'z (B ) * +,b-*./01J2345j62789:R;<=>r?:@A'B0CZ9D"BEBFCGzDHBEI FJFKGLbHM*INIOJPKQJLRMSMTNUjOV2PWPXQYRZRS[T\T]U^rV_:W`XaXbYcZZd"[e[f\gz]hB^i _j_k`lbam*bnbocpdqJerfsftgujhv2iwixjykzRl{m|m}n~ro:pqyZ"zB Ҙb*Jڟj2£Rr:ʪZ"zB ұb*Jڸj2¼Rr:85U9IIYUuYiq 0]:]za2aaZeqzuuzyyz}"}j"" b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements64input0Elements64input1Elements448usefulCycles112physicalWorkCycles112workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 64]: bJ7Qoperation.2exchange :2b!reasonLayoutRearrangementvalue20invocations15==EEMMUU]]eemmuu}}  %%--55==EEMMUU]]eemmuu}}  %%--5 gcopy_strided_u64-:-R121"1B1zUbYYa*aRereasonLocalCopybytes128patternStrided { rows: 4, row_bytes: 32, source_stride: 32, destination_stride: 128 }invocations1uniformInvocationstruearguments[4, 4, 32, 128] b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2̓ b*'5operation.3idle : j2reasonInactiveTilevalue218C9h9((k+k+..0022I5I5?6?6<<W? W?A\ AC0 CEp E#  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L%5 -5MUmu -5MUmu b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations15e6VVvev bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements96input0Elements96input1Elements288usefulCycles72physicalWorkCycles72workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 96] bJ7Moperation.1exchange :z2breasonOperatorInputsvalue14invocations1(!))13;5=;CCIIQNTVXjn boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`A (0=??AAIKSX\\d  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations185U8HHXUuYip~  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j"" b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements64input0Elements64input1Elements448usefulCycles112physicalWorkCycles112workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 64] : bJ7QGoperation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1%5=MU%-  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2̓ b*'5operation.3idle : j2reasonInactiveTilevalue21̓ b*'5operation.3idle : j2reasonInactiveTilevalue21̓ b*'5operation.3idle : j2reasonInactiveTilevalue21̓ b*'5operation.3idle : j2reasonInactiveTilevalue21̓ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue218C9h9((k+k+..0022C5C59696<<W? W?Ad AC8 CEx E) broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L+7 ,4LTlt ,4LTlt b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations15U7gUugu bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements96input0Elements96input1Elements288usefulCycles72physicalWorkCycles72workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 96] bJ7Moperation.1exchange :z2breasonOperatorInputsvalue14invocations1,#+-3397?9AAIIQQSSWag boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`F '/48<>>@@HJR[chn  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations187W9IIYWwYiq!  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j"" b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements64input0Elements64input1Elements448usefulCycles112physicalWorkCycles112workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 64] : bJ7QGoperation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1#+;CS[+3  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2̓ b*'5operation.3idle : j2reasonInactiveTilevalue218C9h9((k+k+..00228888<<W? W?A\ AC0 CEp E/  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L37 -5MUmu -5MUmu b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations14T7gTtgt bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements96input0Elements96input1Elements288usefulCycles72physicalWorkCycles72workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 96] bJ7Moperation.1exchange :z2breasonOperatorInputsvalue14invocations1(""((0087?8@@HHJLT`b boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb2R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements20input0Elements20usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations2uniformInvocationsfalsearguments[20] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`= (0=??AAIKS\d  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations187W8HHXWwXhp   bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j"" b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements64input0Elements64input1Elements448usefulCycles112physicalWorkCycles112workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 64] : bJ7QGoperation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1+3CK[c3;  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2̓ b*'5operation.3idle : j2reasonInactiveTilevalue21̓ b*'5operation.3idle : j2reasonInactiveTilevalue21̓ b*'5operation.3idle : j2reasonInactiveTilevalue21̓ b*'5operation.3idle : j2reasonInactiveTilevalue21̓ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue218C9h9((k+k+..0022C5C59696<<3? 3?Ad AB8 BEx E5  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L9 ,4LTlt ,4LTlt b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations15U9iUuiu bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements96input0Elements96input1Elements288usefulCycles72physicalWorkCycles72workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 96] bJ7Moperation.1exchange :z2breasonOperatorInputsvalue14invocations1,%--57;9A;CCKKMMQS[U[bd boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`9"$399AKQQSSUU]]egooww  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations1899IEUM]Ueaqqr  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1ai9A  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2̓ b*'5operation.3idle : j2reasonInactiveTilevalue21<C9p9((k+k+..00 2 222 64 66\6< <9? 9?A AB BE$EA broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L 9 -5MUmu -5MUmu b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations14T9iTtit bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements96input0Elements96input1Elements288usefulCycles72physicalWorkCycles72workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 96] bJ7Moperation.1exchange :z2breasonOperatorInputsvalue14invocations1$##++3399==C>FFHX` bwoperation.1static_copy_u325:5R9299299B9Z=bAAI*IRM"reasonLocalCopyvalue15bytes16patternContiguousinvocations1uniformInvocationstruearguments[4] boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4]  bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`A#'4::BLRRTTVV^^ffnnvv~~  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations1889IDTM]Tdbrpr  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1iqAI b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2̓ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21<C9p9((k+k+..0022-3-3]6<]6S7dS7< <9? 9?A AB BE,EG broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L&; ,4LTlt ,4LTlt b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations15U;kUuku bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements96input0Elements96input1Elements288usefulCycles72physicalWorkCycles72workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 96] bJ7Moperation.1exchange :z2breasonOperatorInputsvalue14invocations1("*+3377==AACCIIKOUUY bwoperation.1static_copy_u325:5R9299299B9Z=bAAI*IRM"reasonLocalCopyvalue15bytes32patternContiguousinvocations1uniformInvocationstruearguments[8] boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4]  bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`F399AKQQSSUU]]egooww  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations189;KEUO_Uecsqs  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1owGO b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2̓ b*'5operation.3idle : j2reasonInactiveTilevalue21<C9p9((q+q+..0022227<78d8< <9? 9?A AB BE,EM broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L.; -5MUmu -5MUmu b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations14T;kTtkt bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements96input0Elements96input1Elements288usefulCycles72physicalWorkCycles72workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 96] bJ7Moperation.1exchange :z2breasonOperatorInputsvalue14invocations1( !))117799==CCGLT^f bwoperation.1static_copy_u325:5R9299299B9Z=bAAI*IRM"reasonLocalCopyvalue15bytes8patternContiguousinvocations1uniformInvocationstruearguments[2] boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb2R"reasonOperatorKernelvalue15kernelSpecGeluoutputElements4input0Elements4usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations2uniformInvocationsfalsearguments[4]  broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4]  bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`=  4::BLRRTTVV^^ffnnvv~~  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations188;KDTO_Tdcsps  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1wOW b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2̓ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue218C9h9((}+}+..0022I5pI5?6?6<<Q? Q?1AL 1AB BF` FS  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L=$<D\d|$<D\d| b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations1E=]Nnen bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements64input0Elements64input1Elements192usefulCycles48physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Moperation.1exchange :z2breasonOperatorInputsvalue14invocations1 (05==E>DDLJRLNgi boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`= !)&.)119BJX`dlw  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18)1A=MIYQa[kiyl|y~  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1}U]  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2̓ b*'5operation.3idle : j2reasonInactiveTilevalue218C9h9((}+}+..00+2+2[5x[5Q6Q6<<'? '?A ABB F F_ broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L=%=E]e}%=E]e} b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations1D=]Nnen bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements64input0Elements64input1Elements192usefulCycles48physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Moperation.1exchange :z2breasonOperatorInputsvalue14invocations1$,4488<=?@HAGGORVrx boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`C!"**22:?CCKPXYackpvv~~  b%operation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r18 & bJoperation.2exchange̓r0]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[]7 :z2breasonOperatorInputsvalue19invocations18(0@=MHXQaYiiyk{y|  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1]e  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue218C9h9((}+}+..00229p9::<<E? E?1AL 1AB BF` Fe  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L%?$<D\d|$<D\d| b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations1E?_Mmgm bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements64input0Elements64input1Elements192usefulCycles48physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Moperation.1exchange :z2breasonOperatorInputsvalue14invocations1 )119=E?GEIJNSY[_ boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb2R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements24input0Elements24usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations2uniformInvocationsfalsearguments[24] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`D !))119BJPXX`dlqww  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18)1A?OIYSc[kk{l|{  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1ck  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2̓ b*'5operation.3idle : j2reasonInactiveTilevalue218C9h9((++..0022O5xO5E6E6<<'? '?%AT %AB( BFh Fk  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L-?%=E]e}%=E]e} b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations1D?_Llgl bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements64input0Elements64input1Elements192usefulCycles48physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Moperation.1exchange :z2breasonOperatorInputsvalue14invocations1$-55==ABJCIIMMO\dmo boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`?!"**22:3;CKYackv~~  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18(0@?OHXScYiiyk{{}  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1ks  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2̓ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue218I?h?((++..0022I5pI5?6?6<<3? 3?7AL 7AB B#F` #Fq  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1LA$<D\d|$<D\d| b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations1EAaMmim bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements64input0Elements64input1Elements192usefulCycles48physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Moperation.1exchange :z2breasonOperatorInputsvalue14invocations1 &./77==EAIEGRZbd boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`A  247;>@FNPXX`ailtt||  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18))9AQIYQa[{m}{~  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1qy  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue218I?h?((++..00 2 2=5p=53636<<-? -?1AL 1AB B)F` )F}  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1LA%=E]e}%=E]e} b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations1DAaLlil bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements64input0Elements64input1Elements192usefulCycles48physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Moperation.1exchange :z2breasonOperatorInputsvalue14invocations1 *23;;CAICINPPVX\ boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`C  !*08<?AGOOWW_`hmuu}}  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18((8AQHXQaYym}y  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1y  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue218I?h?((++..00229h9::<<3? 3?1AD 1AB B/FX /F  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L C$<D\d|$<D\d| b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations1ECcMmkm bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements64input0Elements64input1Elements192usefulCycles48physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mwoperation.1exchange :z2breasonOperatorInputsvalue14invocations1'//77??GCK[aae boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb2R"reasonOperatorKernelvalue15kernelSpecGeluoutputElements8input0Elements8usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations2uniformInvocationsfalsearguments[8] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`D  %+7;>@FNPXX`ailtt||  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18))9CSIYSc[{o{  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue218I?h?((++..00%2%2U5hU5K6K6<<3? 3?1AD 1AB B;FX ;F  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L(C%=E]e}%=E]e} b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations1DCcLlkl bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements64input0Elements64input1Elements192usefulCycles48physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mwoperation.1exchange :z2breasonOperatorInputsvalue14invocations1+33;;CCKDLdloq boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`C!-/8<?AGOOWW_`hmuu}}  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18((8CSHXScYyoy  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue218I?h?))++..0022O5hO5E6E6<<{? {?CAD CAB BAFX AF  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1LE4<T\t|4<T\t| b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations15=mEeno bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements64input0Elements64input1Elements192usefulCycles48physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mwoperation.1exchange :z2breasonOperatorInputsvalue14invocations1199AAIGOIQ_ggi boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`E $&$&&.NRpt  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18))99IEUUeYieuiyw  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue218I?h?))++..0022'9h'9::<<{? {?7AD 7AB BGFX GF  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1LE5=U]u}5=U]u} b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations14<lEelo bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements64input0Elements64input1Elements192usefulCycles48physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mwoperation.1exchange :z2breasonOperatorInputsvalue14invocations15==EEMHPMUV\fj boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb2R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements28input0Elements28usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations2uniformInvocationsfalsearguments[28] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`E%%''/OSqu  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18((89IEUTdYidtiyu  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue218I?h?))++..0022O5hO5E6E6<<{? {?7AD 7AB BMFX MF  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1LG4<T\t|4<T\t| b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations15=mGgmo bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements64input0Elements64input1Elements192usefulCycles48physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mwoperation.1exchange :z2breasonOperatorInputsvalue14invocations1.66>>FGMHPY]ai boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`G $$&&.NRpt  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18))99IGWUe[keuk{  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue218I?h?))++..00%2%2U5hU5K6K6<<{? {?7AD 7AB BYFX YF  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L'G5=U]u}5=U]u} b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations14<lGglo bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements64input0Elements64input1Elements192usefulCycles48physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mwoperation.1exchange :z2breasonOperatorInputsvalue14invocations12::BBJJRKS[cln boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`G%%''/OSqu  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18((88HGWTd[kdtk{  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue218I?h?))++..0022I5hI5?6?6<<W? W?7AD 7AB B_FX _F  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L I4<T\t|4<T\t| b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations15=mIimq bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements64input0Elements64input1Elements192usefulCycles48physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mwoperation.1exchange :z2breasonOperatorInputsvalue14invocations1/77??GGOIQUWZb boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`I06?CFNNTTVVXpxx  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18))IIYM]]m]mm}q}  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue218I?h?))++. .0J02 292 9::<<W?BW?7AJ7AB BeFeF broperation.0fill_zero_u64%Ӄ(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1LI5=U]u}5=U]u} b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations14<lIilq bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements64input0Elements64input1Elements192usefulCycles48physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mwoperation.1exchange :z2breasonOperatorInputsvalue14invocations13;;CCKIQKSUY\b boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb2R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements12input0Elements12usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations2uniformInvocationsfalsearguments[12] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`I  17@DGOOUUWWYqyy  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18((HIYL\\l]mm}p}  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue218I?h?))++..00%2%2U5hU5K6K6<<W? W?=AD =AB BkFX kF  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1LK4<T\t|4<T\t| b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations15=mKkms bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements64input0Elements64input1Elements192usefulCycles48physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mwoperation.1exchange :z2breasonOperatorInputsvalue14invocations1,44<<DDLKSX\ek boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`K06?CFNNTTVVXpxx  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18))IK[M]]m_ooq  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue218I?h?))++..00+2+2[5h[5Q6Q6<<W? W?IAD IAB BwFX wFŅ  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L"K5=U]u}5=U]u} b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations14<lKkls bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements64input0Elements64input1Elements192usefulCycles48physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mwoperation.1exchange :z2breasonOperatorInputsvalue14invocations1088@@HHPKS`hpr boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`K  17@DGOOUUWWYqyy  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18((HK[L\\l_oop  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2%Ӄ b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue218OEhE ) )++..00%2%2/7`/7%8%8<<o? o?=A< =AB B}FP }F˅  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L M $,DLdl $,DLdl b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations155UNnen bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements48input0Elements48input1Elements192usefulCycles36physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mgoperation.1exchange :z2breasonOperatorInputsvalue14invocations1'//77?MUailn boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb2R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations2uniformInvocationsfalsearguments[32] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`M &&.46_c  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18))99IM]N^aqiyq}  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2+ك b*'5operation.3idle : j2reasonInactiveTilevalue218OEhE ) )++..00+2+2[5`[5Q6Q6<<o? o?IA< IAB BFP Fׅ  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1LM %-EMem %-EMem b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations144TOodo bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements48input0Elements48input1Elements192usefulCycles36physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mgoperation.1exchange :z2breasonOperatorInputsvalue14invocations1#++33;MU_emq boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`M''/57`d  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18((88HM]N^aqiyq|  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue218OEhE ) )++..00%2%2U5`U5K6K6<<o? o?IA< IAB BFP F݅  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1LO $,DLdl $,DLdl b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations155UOoeo bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements48input0Elements48input1Elements192usefulCycles36physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mgoperation.1exchange :z2breasonOperatorInputsvalue14invocations1&..66>OW[agk boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`O&&.46_c  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18))99IM]O_csk{s}  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2+ك b*'5operation.3idle : j2reasonInactiveTilevalue218OEhE ) )++..00%2%2U5`U5K6K6<<u? u?IA< IAB BFP Fㅵ  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1L!O %-EMem %-EMem b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations144TOodo bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements48input0Elements48input1Elements192usefulCycles36physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mgoperation.1exchange :z2breasonOperatorInputsvalue14invocations1"**22:OW^fjl boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`O''/57`d  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18((88HL\O_csiys|  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements32input0Elements32input1Elements224usefulCycles56physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Q'operation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1   b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue21+ك b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue218OEhE ) )++..00%2%28`8 9 9<<c? c?UA< UA C  CFP F酵  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1LQ $,DLdl $,DLdl b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations155UQqeq bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements48input0Elements48input1Elements192usefulCycles36physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mgoperation.1exchange :z2breasonOperatorInputsvalue14invocations1!))119QY]_dl boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb2R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements16input0Elements16usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations2uniformInvocationsfalsearguments[16] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`Q %%''))1157;rzz  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18))99IK[Qaaqfvx{  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements16input0Elements16input1Elements224usefulCycles28physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Qoperation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 O%&8%&..+2+2I5\I5<<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1]eiq boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 jr b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations28OEhE ) )++..00+2+2[5`[5Q6Q6<<]? ]?OA< OAC CFP F  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1LQ %-EMem %-EMem b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations144TQqdq bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements48input0Elements48input1Elements192usefulCycles36physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mgoperation.1exchange :z2breasonOperatorInputsvalue14invocations1%%--5T\bhlp boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`Q &&((**2268<s{{  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18((88HJZQaaqdtuz  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements16input0Elements16input1Elements224usefulCycles28physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Qoperation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 O%&8%&..+2+2I5\I5<<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1]eiq boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 dl b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations21߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue218OEhE ) )++..00+2+2[5`[5Q6Q6<<c? c?UA< UA C  CFP F  broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=7operation.0exchange : zbreasonOperatorInputsinvocations1LS $,DLdl $,DLdl b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations155USses bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements48input0Elements48input1Elements192usefulCycles36physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mgoperation.1exchange :z2breasonOperatorInputsvalue14invocations1 ((008S[_cgm boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`S %%''))1157;rzz  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18))99IK[Sccsfvw}  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements16input0Elements16input1Elements224usefulCycles28physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Qoperation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1  b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 O%&8%&..1212O5\O5<<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1_gks boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 EM b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2<OEpE ) )*x*!-,!-..08012x12a5 a5W64W6<\ <c?p c?UA UA C  CF Fl broperation.0fill_zero_u64(M:MU2UUZYqzu"ujyyrb*R:reasonLayoutRearrangementvalue10kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4] bJ'=operation.0exchange : zbreasonOperatorInputsinvocations1D %-EMem %-EMem bJwoperation.0copy_u645:5R929929"9B9Z=bAAI*IRMBreasonLocalCopyvalue22bytes512patternContiguousinvocations1uniformInvocationstruearguments[10, 4] b)aoperation.0gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue22kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] bJ7Mgoperation.0exchange :z2breasonOperatorInputsvalue12invocations144TSsds bz)operation.0reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue12kernelSpecReductionSum { partials: 4 }outputElements48input0Elements48input1Elements192usefulCycles36physicalWorkCycles48workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[3, 64] bJ7Mgoperation.1exchange :z2breasonOperatorInputsvalue14invocations1$$,,4V^kssu boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeeji"iq"qybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements36input0Elements36usefulCycles108physicalWorkCycles108workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[36] broperation.2fill_zero_u64(M:MzQ2QQZUmzq"qjuu}}rb*R:reasonOperatorInputsvalue17kernelSpecFillZero { offset: 544, bytes: 32, padding_only: true }outputElements272usefulCycles0physicalWorkCycles0workEstimateBasispadding initialization: no useful tensor workinvocations1uniformInvocationstruearguments[0, 4]  bJ'=operation.2exchange : zbreasonOperatorInputsinvocations1`S  &&((**2268<s{{  b)aoperation.2gemm_f16_init_small_rows_interleaved_k16_c16_r18_r180]:]za2aaZez"z"z"jb*RreasonOperatorKernelvalue23kernelSpecGemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 16, output_columns: 16 }outputElements272input0Elements272input1Elements256usefulCycles68physicalWorkCycles72workEstimateBasisAMP FLOPs / IPU21 peakinvocations1uniformInvocationstruearguments[] & bJ7Moperation.2exchange :z2breasonOperatorInputsvalue19invocations18((88HIYSccsdtu|  bz)operation.2reduce_sum_f160]:]za2aaZeqzuuzyyz}"}j b*RBreasonOperatorKernelvalue19kernelSpecReductionSum { partials: 8 }outputElements16input0Elements16input1Elements224usefulCycles28physicalWorkCycles56workEstimateBasisvector reduction add issue slotsinvocations1uniformInvocationstruearguments[7, 32] : bJ7Qoperation.2exchange :2b!reasonLayoutRearrangementvalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 O%&8%&..1212O5\O5i<i<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1_gks boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 *2T\ b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations21߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue21 O%&8%&..+2+2I5\I5c<c<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1U]go boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 OW b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 O%&8%&..%2%2C5\C5<<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1U]ai boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 dlv~ b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 O%&8%&..%2%2C5\C5E<E<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1X``h boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 4< b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 O%&8%&..%2%2C5\C5<<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1U]ai boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 ^fz b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations21߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21 O+&8+&..%2%2C5\C5<<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1W_ai boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 >Fnv b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 O+&8+&..+2+2I5\I5<<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1W_ck boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 9Au} b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 O+&8+&..+2+2I5\I5{<{<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1Zbfn boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 @Hhp b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 O+&8+&..+2+2I5\I5]<]<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1W_ck boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 #+JR b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations21߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21 U+&8+&..7272U5\U5i<i<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1Yamu boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1  S[ b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 U+&8+&..+2+2I5\I5<<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1Yaem boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 Zb b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 U+&8+&..7272U5\U5K<K<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1Yamu boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1  8@ b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 U+&8+&..+2+2I5\I5<<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1Yaem boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 T\| b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations21߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21 U+&8+&..7272U5\U5<<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1[cow boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 BJrz b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 U+&8+&..1212O5\O5<<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1[cgo boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 -5iq b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 U+&8+&..7272U5\U5<<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1[cow boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 DLlt b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 U+&8+&..1212O5\O5Q<Q<e@e@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1[cgo boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1  =E b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations21߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue211߃ b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21 U+&8+&..1212O5\O5o<o<k@k@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1]eiq boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 #W_ b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 U+&8+&..C2C2a5\a5{<{<k@k@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1ksw boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 T\fn b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 U+&8+&..1212O5\O5Q<Q<k@k@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1]eiq boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1  <D b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 U+&8+&..=2=2[5\[5<<k@k@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1iqu} boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 NVjr b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations27僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21 U+&8+&..7272U5\U5<<k@k@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1_gks boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 FNv~ b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 U+&8+&..=2=2[5\[5u<u<k@k@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1iqu} boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 !)]e b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 U+&8+&..7272U5\U5{<{<k@k@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1_gks boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 FNbj b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 U+&8+&..=2=2[5\[5K<K<k@k@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1gou} boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1  19 b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations27僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21 U+&8+&..C2C2a5\a5o<o<k@k@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1aiu} boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 !U] b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 U+&8+&..I2I2g5\g5<<k@k@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1go{ boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 JRz b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 U+&8+&..C2C2a5\a5W<W<k@k@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1aiv~ boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1 @H b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations2 U+&8+&..C2C2a5\a5<<k@k@DPD b*'9operation.0sync : breasonExchangeBarrierinvocations1 b*7Ioperation.0sync :2breasonExchangeBarriervalue12invocations1 bJ7M'operation.1exchange :z2breasonOperatorInputsvalue14invocations1emy boperation.1gelu_tanh_approx_f16,U:UzY2YYZ]*]zaazeejiiqqybb*R*reasonOperatorKernelvalue15kernelSpecGeluoutputElements32input0Elements32usefulCycles96physicalWorkCycles96workEstimateBasisGeLU arithmetic and optional FP8 conversioninvocations1uniformInvocationstruearguments[32] bJ'=7operation.2exchange : zbreasonOperatorInputsinvocations1  DLlt b*7Ioperation.2sync :2breasonExchangeBarriervalue19invocations1 b*7Ioperation.2sync :2breasonExchangeBarriervalue20invocations1 b7Moperation.3repeat-remainder ::2ZreasonRepeatvalue21iterations27僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue217僵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21U b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21O b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21I b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21C b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21=냵 b*'5operation.3idle : j2reasonInactiveTilevalue21