2026-09-23T08:16:31.230230Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_runtime: close time.busy=846µs time.idle=2.43µs
2026-09-23T08:17:01.895455Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:gemm_candidates: close time.busy=30.7s time.idle=5.38µs
2026-09-23T08:17:47.620357Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:gemm_candidates: close time.busy=45.7s time.idle=4.35µs
2026-09-23T08:17:54.430260Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid: close time.busy=83.2s time.idle=4.14µs
2026-09-23T08:17:54.521191Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: wrote planner memory profile path=artifacts/native-resident-20260923/fp16/fp16-b1-n1-default/memory/baseline-1185267-0000.json scope="baseline" peak_step=10 effective_peak_bytes=343040
2026-09-23T08:17:56.666748Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:group_exchanges: close time.busy=673µs time.idle=3.82µs
2026-09-23T08:17:56.675406Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:merge_copies: close time.busy=8.40ms time.idle=1.15µs
2026-09-23T08:17:56.731231Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:eliminate: eliminated padding clears using kernel access contracts removed=0
2026-09-23T08:17:56.732796Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:eliminate: close time.busy=57.4ms time.idle=1.41µs
2026-09-23T08:17:56.788847Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:place_provisional_storage: close time.busy=54.8ms time.idle=2.67µs
2026-09-23T08:17:57.223613Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=50461
2026-09-23T08:17:57.235571Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=18833
2026-09-23T08:17:57.414162Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=1 transfers=29652 maximum_identical_destinations=1
2026-09-23T08:17:57.414194Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=1 initial_horizon=7297 selected_horizon=7297 endpoint_lower_bound=6272 lower_bound_gap=1025 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T08:17:57.799228Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=8612
2026-09-23T08:17:57.860267Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=50461 elapsed_ms=636 success=true
2026-09-23T08:17:57.909183Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=18833 elapsed_ms=673 success=true
2026-09-23T08:17:58.519864Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=5 transfers=50461 maximum_identical_destinations=1
2026-09-23T08:17:58.519897Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=5 initial_horizon=838 selected_horizon=838 endpoint_lower_bound=288 lower_bound_gap=550 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T08:17:58.957523Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=88718
2026-09-23T08:17:59.022896Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: compared ordinary and paired exchange schedules phase=3 candidates=17370 ordinary_horizon=36470 paired_horizon=34539 use_paired=true
2026-09-23T08:17:59.059410Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=3 transfers=18833 maximum_identical_destinations=1
2026-09-23T08:17:59.059427Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=3 initial_horizon=34539 selected_horizon=34539 endpoint_lower_bound=29954 lower_bound_gap=4585 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T08:17:59.253219Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=8612 elapsed_ms=1453 success=true
2026-09-23T08:18:00.616116Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=8612
2026-09-23T08:18:01.183660Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=88718 elapsed_ms=2226 success=true
2026-09-23T08:18:02.058929Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=8612 elapsed_ms=1442 success=true
2026-09-23T08:18:02.707374Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: compared ordinary and paired exchange schedules phase=0 candidates=2808 ordinary_horizon=34415 paired_horizon=21684 use_paired=true
2026-09-23T08:18:02.788122Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=0 transfers=8612 maximum_identical_destinations=1
2026-09-23T08:18:02.788148Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=0 initial_horizon=21684 selected_horizon=21684 endpoint_lower_bound=16512 lower_bound_gap=5172 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T08:18:02.823965Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=2 transfers=88718 maximum_identical_destinations=1
2026-09-23T08:18:02.823989Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=2 initial_horizon=3698 selected_horizon=3698 endpoint_lower_bound=1152 lower_bound_gap=2546 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T08:18:07.532346Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=588528
2026-09-23T08:18:22.001979Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=588528 elapsed_ms=14469 success=true
2026-09-23T08:18:29.315340Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=4 transfers=588528 maximum_identical_destinations=1
2026-09-23T08:18:29.315371Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=4 initial_horizon=7783 selected_horizon=7783 endpoint_lower_bound=6624 lower_bound_gap=1159 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T08:18:30.149159Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: close time.busy=33.4s time.idle=5.20µs
2026-09-23T08:18:30.191551Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:plan_kernels: close time.busy=42.3ms time.idle=2.97µs
2026-09-23T08:18:30.192598Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: compiling kernel source=/srv/home/gc-sdk/ipu-stack/crates/ipu-tests/../../device/gemm_f16_amp.S name="gemm_f16_interleaved_packed0_k192_c64_r216" target="ipu21"
2026-09-23T08:18:30.288156Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/1b7ba824c31f543ea887056469477b7e16ff7bab6cae274567896dff60accd15/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/1b7ba824c31f543ea887056469477b7e16ff7bab6cae274567896dff60accd15/kernel.json
2026-09-23T08:18:30.288691Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: compiling kernel source=/srv/home/gc-sdk/ipu-stack/crates/ipu-tests/../../device/gemm_f16_amp.S name="gemm_f16_interleaved_packed0_k192_c64_r214" target="ipu21"
2026-09-23T08:18:30.334046Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/3b38fb60aa03cc8452676fdc4a772763cb22b0ed5659b5ee9aedaf9622589791/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/3b38fb60aa03cc8452676fdc4a772763cb22b0ed5659b5ee9aedaf9622589791/kernel.json
2026-09-23T08:18:30.334754Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: compiling kernel source=/srv/home/gc-sdk/ipu-stack/crates/ipu-tests/../../device/unpack_transposed_amp_f16.S name="unpack_amp_to_row_major_f16_o1_r64_p64_c36_p36" target="ipu21"
2026-09-23T08:18:30.378267Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/166210c505d1620d2097999874ed3b18c75180982a1b14e0a9eb3f6e23032c29/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/166210c505d1620d2097999874ed3b18c75180982a1b14e0a9eb3f6e23032c29/kernel.json
2026-09-23T08:18:30.378596Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: compiling kernel source=/srv/home/gc-sdk/ipu-stack/crates/ipu-tests/../../device/unpack_transposed_amp_f16.S name="unpack_amp_to_row_major_f16_o1_r48_p48_c36_p36" target="ipu21"
2026-09-23T08:18:30.424018Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/379baf7ac463bee4dd06c0a05b3585b9a4a4712a7e72f707d6fbe8263205aeaf/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/379baf7ac463bee4dd06c0a05b3585b9a4a4712a7e72f707d6fbe8263205aeaf/kernel.json
2026-09-23T08:18:30.424390Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: compiling kernel source=/srv/home/gc-sdk/ipu-stack/crates/ipu-tests/../../device/unpack_transposed_amp_f16.S name="unpack_amp_to_row_major_f16_o1_r41_p48_c36_p36" target="ipu21"
2026-09-23T08:18:30.467272Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/7c88f9daa0bdafeed534de4beba01021f88bb50c89f2d3eac316ece553efe4f7/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/7c88f9daa0bdafeed534de4beba01021f88bb50c89f2d3eac316ece553efe4f7/kernel.json
2026-09-23T08:18:30.467574Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: compiling kernel source=/srv/home/gc-sdk/ipu-stack/crates/ipu-tests/../../device/unpack_transposed_amp_f16.S name="unpack_amp_to_row_major_f16_o1_r64_p64_c34_p34" target="ipu21"
2026-09-23T08:18:30.510105Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/e5b699f35fe81ebaa051ee527861e6da64334321bafb46e349b53e51926221a4/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/e5b699f35fe81ebaa051ee527861e6da64334321bafb46e349b53e51926221a4/kernel.json
2026-09-23T08:18:30.510499Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: compiling kernel source=/srv/home/gc-sdk/ipu-stack/crates/ipu-tests/../../device/unpack_transposed_amp_f16.S name="unpack_amp_to_row_major_f16_o1_r48_p48_c34_p34" target="ipu21"
2026-09-23T08:18:30.550069Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/43f13d62669fa54e03beed241df80dc41eeae6e9d1183560718f74d5b6aba388/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/43f13d62669fa54e03beed241df80dc41eeae6e9d1183560718f74d5b6aba388/kernel.json
2026-09-23T08:18:30.550448Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: compiling kernel source=/srv/home/gc-sdk/ipu-stack/crates/ipu-tests/../../device/unpack_transposed_amp_f16.S name="unpack_amp_to_row_major_f16_o1_r41_p48_c34_p34" target="ipu21"
2026-09-23T08:18:30.595357Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/44fc526c54ccbe4d995603dd3f214af2107f7a966d1b30ece719ba8c8721db02/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/44fc526c54ccbe4d995603dd3f214af2107f7a966d1b30ece719ba8c8721db02/kernel.json
2026-09-23T08:18:30.596176Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: compiling kernel source=/srv/home/gc-sdk/ipu-stack/crates/ipu-tests/../../device/rearrange_amp_left_f16.S name="rearrange_row_major_to_amp_f16_o0_r46_p46_c192_p192" target="ipu21"
2026-09-23T08:18:30.638567Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/67bd535b96b02903a52f4f9723db6a23003f37c36434f30a48ba5f5d13ce7897/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/67bd535b96b02903a52f4f9723db6a23003f37c36434f30a48ba5f5d13ce7897/kernel.json
2026-09-23T08:18:30.638920Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: compiling kernel source=/srv/home/gc-sdk/ipu-stack/crates/ipu-tests/../../device/rearrange_amp_left_f16.S name="rearrange_row_major_to_amp_f16_o0_r46_p46_c80_p192" target="ipu21"
2026-09-23T08:18:30.681967Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/2cba73585001086a4a7ce4903148962a41835276d4662b2a30d08b3a5def1d26/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/2cba73585001086a4a7ce4903148962a41835276d4662b2a30d08b3a5def1d26/kernel.json
2026-09-23T08:18:30.682357Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: compiling kernel source=/srv/home/gc-sdk/ipu-stack/crates/ipu-tests/../../device/rearrange_amp_left_f16.S name="rearrange_row_major_to_amp_f16_o0_r45_p45_c192_p192" target="ipu21"
2026-09-23T08:18:30.727798Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/c76c4c80ce1a3cc43cb307db5691a3e8cca18f44f33ea4586e33228d4782377c/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/c76c4c80ce1a3cc43cb307db5691a3e8cca18f44f33ea4586e33228d4782377c/kernel.json
2026-09-23T08:18:30.728147Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: compiling kernel source=/srv/home/gc-sdk/ipu-stack/crates/ipu-tests/../../device/rearrange_amp_left_f16.S name="rearrange_row_major_to_amp_f16_o0_r45_p45_c80_p192" target="ipu21"
2026-09-23T08:18:30.770432Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/ca4d70417d39d598af909f6f78c483bd888f3606c5297bd1af417a8c1e9061b0/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/ca4d70417d39d598af909f6f78c483bd888f3606c5297bd1af417a8c1e9061b0/kernel.json
2026-09-23T08:18:30.770916Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: compiling kernel source=/srv/home/gc-sdk/ipu-stack/crates/ipu-tests/../../device/gemm_f16_amp.S name="gemm_f16_interleaved_packed0_k192_c288_r46" target="ipu21"
2026-09-23T08:18:30.817938Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/e13f0f529a10187b121cf50b9305cb240ca6c4112b45bfebaab04d484d0ceb79/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/e13f0f529a10187b121cf50b9305cb240ca6c4112b45bfebaab04d484d0ceb79/kernel.json
2026-09-23T08:18:30.818478Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: compiling kernel source=/srv/home/gc-sdk/ipu-stack/crates/ipu-tests/../../device/gemm_f16_amp.S name="gemm_f16_interleaved_packed0_k192_c288_r45" target="ipu21"
2026-09-23T08:18:30.862812Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/9a08cb0b4985c89ca3265315f391fa96d2b80628a38d6d11954b7289b18882ee/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/9a08cb0b4985c89ca3265315f391fa96d2b80628a38d6d11954b7289b18882ee/kernel.json
2026-09-23T08:18:30.862878Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: close time.busy=671ms time.idle=5.59µs
2026-09-23T08:18:30.863479Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:link_runtime: close time.busy=409µs time.idle=2.39µs
2026-09-23T08:18:31.040730Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: exchange table and repeat patch storage exchange_table_bytes=14324 maximum_uncompressed_repeat_patch_bytes=0 maximum_elided_arithmetic_patch_bytes=0
2026-09-23T08:18:31.361378Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:size_tile_code: close time.busy=22.8ms time.idle=3.99µs
2026-09-23T08:18:31.361407Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: placing generated tile programs linked_end=317384 host_code_base=376832 host_code_bytes=2324 generated_code_bytes=2988 exchange_table_bytes=14324 executable_ranges=[(379224, 524288)]
2026-09-23T08:18:31.361440Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: allocated package support memory linked_end=317384 profile_bytes=136 exchange_table_bytes=14324 host_code_bytes=2324 host_data_bytes=1792 generated_code_bytes=2988 code_address=379224 available_ranges=[(327680, 360448), (364000, 376832), (409600, 949168)]
2026-09-23T08:18:31.452445Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:place_storage: close time.busy=77.0ms time.idle=3.37µs
2026-09-23T08:18:31.710438Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=1
2026-09-23T08:18:31.719878Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=1 transfers=29652 maximum_identical_destinations=1
2026-09-23T08:18:31.719895Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=1 initial_horizon=7297 selected_horizon=7297 endpoint_lower_bound=6272 lower_bound_gap=1025 selected_kind="reused" neighborhood_improvements=0
2026-09-23T08:18:32.050510Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=3
2026-09-23T08:18:32.069047Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=3 transfers=18833 maximum_identical_destinations=1
2026-09-23T08:18:32.069064Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=3 initial_horizon=34539 selected_horizon=34539 endpoint_lower_bound=29954 lower_bound_gap=4585 selected_kind="reused" neighborhood_improvements=0
2026-09-23T08:18:32.635759Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=50461
2026-09-23T08:18:33.232173Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: finished incremental instruction-alignment validation transfers=50461 elapsed_ms=596 success=true
2026-09-23T08:18:33.797885Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=2
2026-09-23T08:18:33.829590Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=2 transfers=88718 maximum_identical_destinations=1
2026-09-23T08:18:33.829611Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=2 initial_horizon=3698 selected_horizon=3698 endpoint_lower_bound=1152 lower_bound_gap=2546 selected_kind="reused" neighborhood_improvements=0
2026-09-23T08:18:33.863452Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=5 transfers=50461 maximum_identical_destinations=1
2026-09-23T08:18:33.863475Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=5 initial_horizon=838 selected_horizon=838 endpoint_lower_bound=288 lower_bound_gap=550 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T08:18:36.457480Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=8612
2026-09-23T08:18:37.991239Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: finished incremental instruction-alignment validation transfers=8612 elapsed_ms=1533 success=true
2026-09-23T08:18:39.346420Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=8612
2026-09-23T08:18:40.763794Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: finished incremental instruction-alignment validation transfers=8612 elapsed_ms=1417 success=true
2026-09-23T08:18:41.418155Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: compared ordinary and paired exchange schedules phase=0 candidates=2808 ordinary_horizon=28101 paired_horizon=18780 use_paired=true
2026-09-23T08:18:41.509261Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=0 transfers=8612 maximum_identical_destinations=1
2026-09-23T08:18:41.509288Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=0 initial_horizon=18780 selected_horizon=18780 endpoint_lower_bound=16512 lower_bound_gap=2268 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T08:18:46.837738Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=4
2026-09-23T08:18:47.035454Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=4 transfers=588528 maximum_identical_destinations=1
2026-09-23T08:18:47.035477Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=4 initial_horizon=7783 selected_horizon=7783 endpoint_lower_bound=6624 lower_bound_gap=1159 selected_kind="reused" neighborhood_improvements=0
2026-09-23T08:18:48.370830Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: close time.busy=16.9s time.idle=5.68µs
2026-09-23T08:18:48.380618Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: costed final placed program final_cycles=283975 final_exchange=76535
2026-09-23T08:18:48.657477Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:prepare_tile_code: close time.busy=40.4ms time.idle=4.39µs
2026-09-23T08:18:48.681424Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:emit_tile_code: close time.busy=23.9ms time.idle=1.42µs
2026-09-23T08:18:48.833832Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:build_tile_images: close time.busy=138ms time.idle=1.94µs
2026-09-23T08:18:49.035709Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: wrote exact tile placement profile path=artifacts/native-resident-20260923/fp16/fp16-b1-n1-default/memory/placement-1185267.html
2026-09-23T08:18:49.204597Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: close time.busy=138s time.idle=5.44µs
2026-09-23T08:18:49.214166Z  INFO writing application package tiles=1472
2026-09-23T08:18:49.279687Z  INFO application package written
2026-09-23T08:18:49.279898Z  INFO reading application package
2026-09-23T08:18:49.427706Z  INFO application package read tiles=1472 compiler=0.1.0
2026-09-23T08:18:49.428345Z  INFO waiting for hardware ownership path=/tmp/ipu-stack-device.lock
2026-09-23T08:18:49.428357Z  INFO acquired hardware ownership path=/tmp/ipu-stack-device.lock
2026-09-23T08:18:49.428362Z  INFO opening IPU device path="/dev/ipu0"
2026-09-23T08:18:49.428393Z  INFO IPU device attached path=/dev/ipu0
2026-09-23T08:18:49.428395Z  INFO initializing IPU device
2026-09-23T08:18:49.436253Z  INFO IPU device initialized
2026-09-23T08:18:49.436297Z  INFO replaying device configuration register_writes=2003
2026-09-23T08:18:49.442153Z  INFO loading application tile_count=1472 final_mark=23
2026-09-23T08:18:49.630088Z  INFO application loaded and initial synchronization completed tile_count=1472
2026-09-23T08:18:50.108219Z  INFO prepared reference inputs elapsed_ms=476
2026-09-23T08:18:50.120228Z  INFO configured host reference BLAS threads=4
referencePrecision=device
2026-09-23T08:18:50.231075Z  INFO evaluated host reference elapsed_ms=122
2026-09-23T08:18:50.232490Z  INFO starting host exchange session startup_mark=23
2026-09-23T08:18:50.232501Z  INFO attaching host exchange pages pages=2
2026-09-23T08:18:50.233214Z  INFO host exchange session started
2026-09-23T08:18:50.233251Z  INFO invoking host exchange call call="initialize" command=0 phases=58
2026-09-23T08:18:50.246265Z  INFO invoking host exchange call call="run" command=0 phases=13
referenceInference=1 numericalTest=PASS
profile=artifacts/native-resident-20260923/fp16/fp16-b1-n1-default/profile.capnp tiles=1472 samples=41024
workload=siglip-mlp-f16-b1-t729-d1152-h4304-n1 benchmark=siglip-mlp-f16 batch=1 tokens=729 rows=729 dimension=1152 hiddenDimension=4304 blocks=1 biases=false activeTiles=1469 executionTiles=1472 inputBytes=1679616 weightBytes=20090880 cycles=282330 minimumTileCycles=274818 deviceMicroseconds=188.220 effectiveGemmTflops=76.815 peakTflops=282.624 efficiencyPercent=27.18 maximumAbsoluteError=0.000977
package=artifacts/native-resident-20260923/fp16/fp16-b1-n1-default/model.ipuexe tiles=1472 hardwareTest=PASS