2026-09-22T14:10:21.363291Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_runtime: close time.busy=310µs time.idle=2.45µs
2026-09-22T14:10:52.985698Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:gemm_candidates: close time.busy=31.6s time.idle=4.44µs
2026-09-22T14:11:39.759106Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:gemm_candidates: close time.busy=46.8s time.idle=3.68µs
2026-09-22T14:11:44.083743Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid: close time.busy=82.7s time.idle=3.15µs
2026-09-22T14:11:46.294577Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:group_exchanges: close time.busy=2.35ms time.idle=3.12µs
2026-09-22T14:11:46.328078Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:merge_copies: close time.busy=33.2ms time.idle=819ns
2026-09-22T14:11:46.411144Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:eliminate: close time.busy=82.5ms time.idle=960ns
2026-09-22T14:11:46.476722Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:place_provisional_storage: close time.busy=62.9ms time.idle=2.19µs
2026-09-22T14:11:46.707985Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=6 transfers=10871 maximum_identical_destinations=1
2026-09-22T14:11:46.708011Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=6 initial_horizon=1543 selected_horizon=1381 endpoint_lower_bound=1152 lower_bound_gap=229 selected_kind="matching-waves" neighborhood_improvements=0
2026-09-22T14:11:46.804021Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=37560
2026-09-22T14:11:47.241515Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=5 transfers=33488 maximum_identical_destinations=1
2026-09-22T14:11:47.241537Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=5 initial_horizon=29330 selected_horizon=29330 endpoint_lower_bound=26064 lower_bound_gap=3266 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-22T14:11:47.315344Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=37560 elapsed_ms=511 success=true
2026-09-22T14:11:47.689739Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=1 transfers=37560 maximum_identical_destinations=1
2026-09-22T14:11:47.689756Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=1 initial_horizon=21790 selected_horizon=21790 endpoint_lower_bound=13800 lower_bound_gap=7990 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-22T14:11:48.062230Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=2 transfers=51228 maximum_identical_destinations=1
2026-09-22T14:11:48.062251Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=2 initial_horizon=6215 selected_horizon=5961 endpoint_lower_bound=4440 lower_bound_gap=1521 selected_kind="critical-neighborhood" neighborhood_improvements=1
2026-09-22T14:11:48.672973Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=67068
2026-09-22T14:11:48.715850Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=18645
2026-09-22T14:11:51.555709Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=67068 elapsed_ms=2882 success=true
2026-09-22T14:11:51.573866Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=18645 elapsed_ms=2857 success=true
2026-09-22T14:11:52.572706Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=4 transfers=18645 maximum_identical_destinations=1
2026-09-22T14:11:52.572725Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=4 initial_horizon=14670 selected_horizon=14670 endpoint_lower_bound=13824 lower_bound_gap=846 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-22T14:11:53.827927Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=35229
2026-09-22T14:11:56.157069Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: compared ordinary and paired exchange schedules phase=3 candidates=64356 ordinary_horizon=18394 paired_horizon=20503 use_paired=false
2026-09-22T14:11:56.263469Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=3 transfers=67068 maximum_identical_destinations=1
2026-09-22T14:11:56.263483Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=3 initial_horizon=18394 selected_horizon=18394 endpoint_lower_bound=17568 lower_bound_gap=826 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-22T14:12:01.696447Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=35229 elapsed_ms=7868 success=true
2026-09-22T14:12:14.064436Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=35229
2026-09-22T14:12:23.074455Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=35229 elapsed_ms=9009 success=true
2026-09-22T14:12:25.490322Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: compared ordinary and paired exchange schedules phase=0 candidates=9120 ordinary_horizon=36265 paired_horizon=32361 use_paired=true
2026-09-22T14:12:25.758668Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=0 transfers=35229 maximum_identical_destinations=1
2026-09-22T14:12:25.758685Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=0 initial_horizon=32361 selected_horizon=32361 endpoint_lower_bound=25240 lower_bound_gap=7121 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-22T14:12:25.920096Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: close time.busy=39.4s time.idle=4.83µs
2026-09-22T14:12:26.017880Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:plan_kernels: close time.busy=97.7ms time.idle=5.24µs
2026-09-22T14:12:26.018087Z  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_f8.cpp name="rearrange_row_major_to_amp_f8_o0_r146_p146_c288_p288_codelet" target="ipu21"
2026-09-22T14:12:26.328457Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/ee65d716ca043a3d57e38b88a087dd2830c4dead8e6a41dd181907a3b225ee7a/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/ee65d716ca043a3d57e38b88a087dd2830c4dead8e6a41dd181907a3b225ee7a/kernel.json
2026-09-22T14:12:26.328792Z  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/worker_call.S name="rearrange_row_major_to_amp_f8_o0_r146_p146_c288_p288_wrapper" target="ipu21"
2026-09-22T14:12:26.373386Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/e328fceca8e34db9ec0d64ae64145091d4808fefe57b5e5e361c73ac2c76d43c/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/e328fceca8e34db9ec0d64ae64145091d4808fefe57b5e5e361c73ac2c76d43c/kernel.json
2026-09-22T14:12:26.373718Z  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_f8.cpp name="rearrange_row_major_to_amp_f8_o0_r145_p145_c288_p288_codelet" target="ipu21"
2026-09-22T14:12:26.673442Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/1985ca6dfdb5f3bd3244923f8d96c03c5da923fa7ae7adb47164bea757d9acc5/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/1985ca6dfdb5f3bd3244923f8d96c03c5da923fa7ae7adb47164bea757d9acc5/kernel.json
2026-09-22T14:12:26.673774Z  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/worker_call.S name="rearrange_row_major_to_amp_f8_o0_r145_p145_c288_p288_wrapper" target="ipu21"
2026-09-22T14:12:26.719559Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/834b3a0f0a9f62feaf027ba917cbffc5580a93c141aefcf369a9b4042dd70986/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/834b3a0f0a9f62feaf027ba917cbffc5580a93c141aefcf369a9b4042dd70986/kernel.json
2026-09-22T14:12:26.719931Z  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_block_major_f8.S name="rearrange_row_major_to_amp_f8_o2_r288_p288_c240_p240_b288x16" target="ipu21"
2026-09-22T14:12:26.767227Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/9ee86d064c8e378416e7c87acfcbba0c9c729d2f6824006622408b7ce8c40638/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/9ee86d064c8e378416e7c87acfcbba0c9c729d2f6824006622408b7ce8c40638/kernel.json
2026-09-22T14:12:26.767376Z  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_block_major_f8.S name="rearrange_row_major_to_amp_f8_o2_r288_p288_c224_p240_b288x16" target="ipu21"
2026-09-22T14:12:26.812079Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/d2644744df648a4f468701a6f2bd3d40acf79177c44f623f2c43b31f7ad70e56/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/d2644744df648a4f468701a6f2bd3d40acf79177c44f623f2c43b31f7ad70e56/kernel.json
2026-09-22T14:12:26.812421Z  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_f8_interleaved_packed0_k288_c240_r146" target="ipu21"
2026-09-22T14:12:26.855763Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/6a1e3b9e414e128cfc62b7adcecfa0587e9f558669ded6ec971420618d6a1330/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/6a1e3b9e414e128cfc62b7adcecfa0587e9f558669ded6ec971420618d6a1330/kernel.json
2026-09-22T14:12:26.855920Z  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_f8_interleaved_packed0_k288_c240_r145" target="ipu21"
2026-09-22T14:12:26.902158Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/68f4d577461933ab768cf53be8ec93fb59929a781e88804379634c4421e3499f/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/68f4d577461933ab768cf53be8ec93fb59929a781e88804379634c4421e3499f/kernel.json
2026-09-22T14:12:26.902895Z  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_r183_p183_c192_p192" target="ipu21"
2026-09-22T14:12:26.945297Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/567778df847f7b5875df7e879fd6c3f1a4ade3eefd8144db502b2ffee1b9fdd3/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/567778df847f7b5875df7e879fd6c3f1a4ade3eefd8144db502b2ffee1b9fdd3/kernel.json
2026-09-22T14:12:26.945512Z  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_r183_p183_c80_p192" target="ipu21"
2026-09-22T14:12:26.987983Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/ef962594a750dfad08ab168f62217318bc9dc1600a9db7167125664172385fb9/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/ef962594a750dfad08ab168f62217318bc9dc1600a9db7167125664172385fb9/kernel.json
2026-09-22T14:12:26.989570Z  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_block_major_f8.S name="rearrange_row_major_to_amp_f8_o2_r192_p192_c288_p288_b192x16" target="ipu21"
2026-09-22T14:12:27.032371Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/29b7a72c1314d5762742eb69779902e31802aca1705e1169b18b913aea42cc52/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/29b7a72c1314d5762742eb69779902e31802aca1705e1169b18b913aea42cc52/kernel.json
2026-09-22T14:12:27.032727Z  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_block_major_f8.S name="rearrange_row_major_to_amp_f8_o2_r80_p192_c288_p288_b192x16" target="ipu21"
2026-09-22T14:12:27.075675Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/6bf94629d13dd7f887c22a718d68458071d1f3a230a55f24ee922b0dabd4fb38/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/6bf94629d13dd7f887c22a718d68458071d1f3a230a55f24ee922b0dabd4fb38/kernel.json
2026-09-22T14:12:27.076122Z  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_f8_interleaved_packed0_k192_c288_r183" target="ipu21"
2026-09-22T14:12:27.117852Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/55bd7ba4fc1b7e2e0ed716dcf1e79b8ecd95488712a8985d034ee1011e911321/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/55bd7ba4fc1b7e2e0ed716dcf1e79b8ecd95488712a8985d034ee1011e911321/kernel.json
2026-09-22T14:12:27.118617Z  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_f8_interleaved_packed0_k192_c288_r182" target="ipu21"
2026-09-22T14:12:27.162791Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/9f830e295ae4d107cf59c7c0a6bc18e1e501843823387174d63f57138200cfb5/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/9f830e295ae4d107cf59c7c0a6bc18e1e501843823387174d63f57138200cfb5/kernel.json
2026-09-22T14:12:27.163008Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: close time.busy=1.15s time.idle=4.52µs
2026-09-22T14:12:27.163742Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:link_runtime: close time.busy=543µs time.idle=2.44µs
2026-09-22T14:12:27.332098Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: exchange table and repeat patch storage exchange_table_bytes=15124 maximum_uncompressed_repeat_patch_bytes=0 maximum_elided_arithmetic_patch_bytes=0
2026-09-22T14:12:27.688686Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:size_tile_code: close time.busy=18.2ms time.idle=3.88µs
2026-09-22T14:12:27.688707Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: placing generated tile programs linked_end=321964 host_code_base=376832 host_code_bytes=964 generated_code_bytes=10696 exchange_table_bytes=15124 executable_ranges=[(377864, 524288)]
2026-09-22T14:12:27.688740Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: allocated package support memory linked_end=321964 profile_bytes=128 exchange_table_bytes=15124 host_code_bytes=964 host_data_bytes=664 generated_code_bytes=10696 code_address=377864 available_ranges=[(327680, 360448), (362872, 376832), (409600, 949168)]
2026-09-22T14:12:27.830669Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:place_storage: close time.busy=126ms time.idle=3.48µs
2026-09-22T14:12:27.931701Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=6
2026-09-22T14:12:27.934689Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=6 transfers=10871 maximum_identical_destinations=1
2026-09-22T14:12:27.934698Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=6 initial_horizon=1381 selected_horizon=1381 endpoint_lower_bound=1152 lower_bound_gap=229 selected_kind="reused" neighborhood_improvements=0
2026-09-22T14:12:28.121683Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=5
2026-09-22T14:12:28.133310Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=5 transfers=33488 maximum_identical_destinations=1
2026-09-22T14:12:28.133319Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=5 initial_horizon=29330 selected_horizon=29330 endpoint_lower_bound=26064 lower_bound_gap=3266 selected_kind="reused" neighborhood_improvements=0
2026-09-22T14:12:28.247351Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=2
2026-09-22T14:12:28.264783Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=2 transfers=51228 maximum_identical_destinations=1
2026-09-22T14:12:28.264794Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=2 initial_horizon=5961 selected_horizon=5961 endpoint_lower_bound=4440 lower_bound_gap=1521 selected_kind="reused" neighborhood_improvements=0
2026-09-22T14:12:28.854773Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=37560
2026-09-22T14:12:29.343665Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: finished incremental instruction-alignment validation transfers=37560 elapsed_ms=488 success=true
2026-09-22T14:12:29.712380Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=1 transfers=37560 maximum_identical_destinations=1
2026-09-22T14:12:29.712404Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=1 initial_horizon=17914 selected_horizon=17914 endpoint_lower_bound=13800 lower_bound_gap=4114 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-22T14:12:30.937767Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=4
2026-09-22T14:12:31.006760Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=4 transfers=18645 maximum_identical_destinations=1
2026-09-22T14:12:31.006780Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=4 initial_horizon=14670 selected_horizon=14670 endpoint_lower_bound=13824 lower_bound_gap=846 selected_kind="reused" neighborhood_improvements=0
2026-09-22T14:12:31.415968Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=3
2026-09-22T14:12:31.499393Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=3 transfers=67068 maximum_identical_destinations=1
2026-09-22T14:12:31.499414Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=3 initial_horizon=18394 selected_horizon=18394 endpoint_lower_bound=17568 lower_bound_gap=826 selected_kind="reused" neighborhood_improvements=0
2026-09-22T14:12:35.300513Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=35229
2026-09-22T14:12:47.146801Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: finished incremental instruction-alignment validation transfers=35229 elapsed_ms=11846 success=true
2026-09-22T14:13:00.184003Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=35229
2026-09-22T14:13:12.780669Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: finished incremental instruction-alignment validation transfers=35229 elapsed_ms=12596 success=true
2026-09-22T14:13:16.354490Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: compared ordinary and paired exchange schedules phase=0 candidates=9118 ordinary_horizon=31703 paired_horizon=31102 use_paired=true
2026-09-22T14:13:16.860091Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=0 transfers=35229 maximum_identical_destinations=1
2026-09-22T14:13:16.860119Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=0 initial_horizon=31102 selected_horizon=31102 endpoint_lower_bound=25240 lower_bound_gap=5862 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-22T14:13:17.138167Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: close time.busy=49.3s time.idle=5.92µs
2026-09-22T14:13:17.199945Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: costed final placed program final_cycles=1213203 final_exchange=122952
2026-09-22T14:13:17.522814Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:prepare_tile_code: close time.busy=31.7ms time.idle=6.28µs
2026-09-22T14:13:17.600090Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:emit_tile_code: close time.busy=77.2ms time.idle=3.37µs
2026-09-22T14:13:17.864389Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:build_tile_images: close time.busy=200ms time.idle=3.59µs
2026-09-22T14:13:18.186684Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: close time.busy=177s time.idle=5.45µs
2026-09-22T14:13:18.193430Z  INFO writing application package tiles=1472
2026-09-22T14:13:18.400775Z  INFO application package written
2026-09-22T14:13:18.404530Z  INFO reading application package
2026-09-22T14:13:18.636023Z  INFO application package read tiles=1472 compiler=0.1.0
2026-09-22T14:13:18.643171Z  INFO waiting for hardware ownership path=/tmp/ipu-stack-device.lock
2026-09-22T14:13:18.643186Z  INFO acquired hardware ownership path=/tmp/ipu-stack-device.lock
2026-09-22T14:13:18.643192Z  INFO opening IPU device path="/dev/ipu0"
2026-09-22T14:13:18.643254Z  INFO IPU device attached path=/dev/ipu0
2026-09-22T14:13:18.643257Z  INFO initializing IPU device
2026-09-22T14:13:18.651240Z  INFO IPU device initialized
2026-09-22T14:13:18.651277Z  INFO replaying device configuration register_writes=2003
2026-09-22T14:13:18.659958Z  INFO loading application tile_count=1472 final_mark=23
2026-09-22T14:13:18.885618Z  INFO application loaded and initial synchronization completed tile_count=1472
2026-09-22T14:13:19.572258Z  INFO prepared reference inputs elapsed_ms=684
2026-09-22T14:13:20.049939Z  INFO configured host reference BLAS threads=4
referencePrecision=device
2026-09-22T14:13:21.529193Z  INFO evaluated host reference elapsed_ms=1956
2026-09-22T14:13:21.532714Z  INFO starting host exchange session startup_mark=23
2026-09-22T14:13:21.532728Z  INFO attaching host exchange pages pages=2
2026-09-22T14:13:21.533438Z  INFO host exchange session started
2026-09-22T14:13:21.533706Z  INFO invoking host exchange call call="initialize" command=0 phases=12
2026-09-22T14:13:21.545997Z  INFO invoking host exchange call call="run" command=0 phases=15
referenceInference=1 numericalTest=PASS
profile=artifacts/fp8-pack-check/mlp-asm.capnp tiles=1472 samples=35127
workload=siglip-mlp-f16-b4-t729-d1152-h4304-n1 benchmark=siglip-mlp-f16 batch=4 tokens=729 rows=2916 dimension=1152 hiddenDimension=4304 blocks=1 biases=false activeTiles=1458 executionTiles=1472 inputBytes=3359232 weightBytes=9916416 cycles=884784 minimumTileCycles=877692 deviceMicroseconds=589.856 effectiveGemmTflops=98.045 peakTflops=282.624 efficiencyPercent=34.69 maximumAbsoluteError=0.005859
package=artifacts/fp8-pack-check/mlp-asm.ipuexe tiles=1472 hardwareTest=PASS