2026-09-23T10:08:04.288990Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_runtime: close time.busy=284µs time.idle=4.74µs
2026-09-23T10:08:58.858343Z DEBUG GEMM model frontier assignments=990724 retained=188
2026-09-23T10:08:58.859970Z DEBUG GEMM model frontier assignments=990724 retained=188
2026-09-23T10:09:14.251912Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:gemm_candidates: close time.busy=70.0s time.idle=3.63µs
2026-09-23T10:10:14.988759Z DEBUG GEMM model frontier assignments=854348 retained=208
2026-09-23T10:10:14.990820Z DEBUG GEMM model frontier assignments=854348 retained=208
2026-09-23T10:10:14.991410Z DEBUG GEMM model frontier assignments=854348 retained=208
2026-09-23T10:10:14.991935Z DEBUG GEMM model frontier assignments=854348 retained=208
2026-09-23T10:10:28.466427Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:gemm_candidates: close time.busy=74.2s time.idle=5.19µs
2026-09-23T10:11:03.452699Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:group_exchanges: close time.busy=917µs time.idle=2.72µs
2026-09-23T10:11:03.466321Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:merge_copies: close time.busy=13.3ms time.idle=1.40µs
2026-09-23T10:11:03.535484Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:eliminate: eliminated padding clears using kernel access contracts removed=0
2026-09-23T10:11:03.536999Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:eliminate: close time.busy=70.7ms time.idle=775ns
2026-09-23T10:11:03.774274Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid: costed planning finalist rank=0 mid_cycles=722911 low_cycles=744606 elapsed_ms=1830
2026-09-23T10:11:05.641329Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:group_exchanges: close time.busy=1.01ms time.idle=3.00µs
2026-09-23T10:11:05.651717Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:merge_copies: close time.busy=10.1ms time.idle=1.55µs
2026-09-23T10:11:05.729887Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:eliminate: eliminated padding clears using kernel access contracts removed=0
2026-09-23T10:11:05.731368Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:eliminate: close time.busy=79.6ms time.idle=1.32µs
2026-09-23T10:11:05.948113Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid: costed planning finalist rank=1 mid_cycles=722911 low_cycles=744606 elapsed_ms=1946
2026-09-23T10:11:08.002524Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:group_exchanges: close time.busy=905µs time.idle=3.00µs
2026-09-23T10:11:08.011691Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:merge_copies: close time.busy=8.86ms time.idle=1.01µs
2026-09-23T10:11:08.094843Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:eliminate: eliminated padding clears using kernel access contracts removed=0
2026-09-23T10:11:08.096554Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:eliminate: close time.busy=84.8ms time.idle=1.43µs
2026-09-23T10:11:08.444218Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid: costed planning finalist rank=2 mid_cycles=723025 low_cycles=745458 elapsed_ms=2243
2026-09-23T10:11:10.431147Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:group_exchanges: close time.busy=1.02ms time.idle=3.21µs
2026-09-23T10:11:10.440576Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:merge_copies: close time.busy=9.13ms time.idle=1.16µs
2026-09-23T10:11:10.525317Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:eliminate: eliminated padding clears using kernel access contracts removed=0
2026-09-23T10:11:10.526875Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:eliminate: close time.busy=86.3ms time.idle=828ns
2026-09-23T10:11:10.813930Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid: costed planning finalist rank=3 mid_cycles=723025 low_cycles=745458 elapsed_ms=2094
2026-09-23T10:11:13.018888Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:group_exchanges: close time.busy=1.08ms time.idle=3.07µs
2026-09-23T10:11:13.032629Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:merge_copies: close time.busy=13.5ms time.idle=652ns
2026-09-23T10:11:13.119565Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:eliminate: eliminated padding clears using kernel access contracts removed=0
2026-09-23T10:11:13.121018Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:eliminate: close time.busy=88.4ms time.idle=713ns
2026-09-23T10:11:13.402539Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid: costed planning finalist rank=4 mid_cycles=723720 low_cycles=762036 elapsed_ms=2303
2026-09-23T10:11:15.460620Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:group_exchanges: close time.busy=1.11ms time.idle=2.71µs
2026-09-23T10:11:15.471122Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:merge_copies: close time.busy=10.3ms time.idle=760ns
2026-09-23T10:11:15.556003Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:eliminate: eliminated padding clears using kernel access contracts removed=0
2026-09-23T10:11:15.557501Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:eliminate: close time.busy=86.4ms time.idle=724ns
2026-09-23T10:11:15.799020Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid: costed planning finalist rank=5 mid_cycles=723720 low_cycles=762036 elapsed_ms=2118
2026-09-23T10:11:18.334823Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:group_exchanges: close time.busy=1.08ms time.idle=3.13µs
2026-09-23T10:11:18.346044Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:merge_copies: close time.busy=10.9ms time.idle=1.22µs
2026-09-23T10:11:18.442110Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:eliminate: eliminated padding clears using kernel access contracts removed=0
2026-09-23T10:11:18.443980Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:eliminate: close time.busy=97.9ms time.idle=788ns
2026-09-23T10:11:18.827300Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid: costed planning finalist rank=6 mid_cycles=723834 low_cycles=762888 elapsed_ms=2741
2026-09-23T10:11:21.356639Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:group_exchanges: close time.busy=1.08ms time.idle=3.10µs
2026-09-23T10:11:21.367220Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:merge_copies: close time.busy=10.3ms time.idle=1.20µs
2026-09-23T10:11:21.458443Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:eliminate: eliminated padding clears using kernel access contracts removed=0
2026-09-23T10:11:21.460080Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:eliminate: close time.busy=92.8ms time.idle=954ns
2026-09-23T10:11:21.767469Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid: costed planning finalist rank=7 mid_cycles=723834 low_cycles=762888 elapsed_ms=2602
2026-09-23T10:11:22.000577Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid: close time.busy=198s time.idle=3.01µs
2026-09-23T10:11:23.491504Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:group_exchanges: close time.busy=916µs time.idle=2.83µs
2026-09-23T10:11:23.500241Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:merge_copies: close time.busy=8.48ms time.idle=1.16µs
2026-09-23T10:11:23.562845Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:eliminate: eliminated padding clears using kernel access contracts removed=0
2026-09-23T10:11:23.564352Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:eliminate: close time.busy=64.1ms time.idle=1.00µs
2026-09-23T10:11:23.615739Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:place_provisional_storage: close time.busy=50.0ms time.idle=1.28µs
2026-09-23T10:11:23.992415Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=47344
2026-09-23T10:11:24.536672Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=47344 elapsed_ms=544 success=true
2026-09-23T10:11:25.157788Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=1 transfers=47344 maximum_identical_destinations=1
2026-09-23T10:11:25.157817Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=1 initial_horizon=15159 selected_horizon=15159 endpoint_lower_bound=13320 lower_bound_gap=1839 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T10:11:27.158645Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=132190
2026-09-23T10:11:27.275050Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=4 transfers=121620 maximum_identical_destinations=1
2026-09-23T10:11:27.275086Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=4 initial_horizon=19737 selected_horizon=19737 endpoint_lower_bound=18048 lower_bound_gap=1689 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T10:11:27.851568Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=32435
2026-09-23T10:11:31.767570Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=132190 elapsed_ms=4608 success=true
2026-09-23T10:11:34.351951Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=32435 elapsed_ms=6500 success=true
2026-09-23T10:11:35.056588Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=5 transfers=132190 maximum_identical_destinations=1
2026-09-23T10:11:35.056624Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=5 initial_horizon=1899 selected_horizon=1899 endpoint_lower_bound=1248 lower_bound_gap=651 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T10:11:35.476404Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: compared ordinary and paired exchange schedules phase=3 candidates=9684 ordinary_horizon=70656 paired_horizon=49404 use_paired=true
2026-09-23T10:11:35.678885Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=3 transfers=53424 maximum_identical_destinations=1
2026-09-23T10:11:35.678908Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=3 initial_horizon=49404 selected_horizon=49404 endpoint_lower_bound=41472 lower_bound_gap=7932 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T10:11:40.869862Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=32435
2026-09-23T10:11:43.532431Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=766680
2026-09-23T10:11:46.841020Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=32435 elapsed_ms=5971 success=true
2026-09-23T10:11:49.989387Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: compared ordinary and paired exchange schedules phase=0 candidates=10121 ordinary_horizon=61022 paired_horizon=50730 use_paired=true
2026-09-23T10:11:50.290509Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=0 transfers=32435 maximum_identical_destinations=1
2026-09-23T10:11:50.290541Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=0 initial_horizon=50730 selected_horizon=50730 endpoint_lower_bound=42874 lower_bound_gap=7856 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T10:12:16.213265Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=766680 elapsed_ms=32680 success=true
2026-09-23T10:12:35.071987Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=2 transfers=766680 maximum_identical_destinations=1
2026-09-23T10:12:35.072024Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=2 initial_horizon=6618 selected_horizon=6618 endpoint_lower_bound=4392 lower_bound_gap=2226 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T10:12:37.706464Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: close time.busy=74.1s time.idle=8.48µs
2026-09-23T10:12:37.734901Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:plan_kernels: close time.busy=28.4ms time.idle=2.70µs
2026-09-23T10:12:37.735576Z  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_k288_c144_r243" target="ipu21"
2026-09-23T10:12:37.786099Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/9e973f8dc473f792512a7ccdf408f0a6f7a74b42d5d67c6ce617524c92d2a105/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/9e973f8dc473f792512a7ccdf408f0a6f7a74b42d5d67c6ce617524c92d2a105/kernel.json
2026-09-23T10:12:37.786964Z  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_k288_c96_r384" target="ipu21"
2026-09-23T10:12:37.836900Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/7061c1e6224d7af20e153705fbc83f5ec1c9af29f91633c0def9fe75e06358b3/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/7061c1e6224d7af20e153705fbc83f5ec1c9af29f91633c0def9fe75e06358b3/kernel.json
2026-09-23T10:12:37.837061Z  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_r96_p96_c26_p26" target="ipu21"
2026-09-23T10:12:37.881670Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/8a617bc97d58600793e7dc5bc67994aca60e3d1e25961ee428052c48779db22f/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/8a617bc97d58600793e7dc5bc67994aca60e3d1e25961ee428052c48779db22f/kernel.json
2026-09-23T10:12:37.881961Z  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_r80_p80_c26_p26" target="ipu21"
2026-09-23T10:12:37.926174Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/b2df073dbc1f9fe92277494d3cd5ddeb4f11839cfe97236c32967fcb37412d90/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/b2df073dbc1f9fe92277494d3cd5ddeb4f11839cfe97236c32967fcb37412d90/kernel.json
2026-09-23T10:12:37.926415Z  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_r73_p80_c26_p26" target="ipu21"
2026-09-23T10:12:37.971984Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/7a848347911d6f0927fdcba8078c8e79de1522c38c4c16b00b9bcb6aad87922d/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/7a848347911d6f0927fdcba8078c8e79de1522c38c4c16b00b9bcb6aad87922d/kernel.json
2026-09-23T10:12:37.972354Z  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_r96_p96_c24_p24" target="ipu21"
2026-09-23T10:12:38.014009Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/d449db5cd7240552bb247d1aed4129fe7216ca170a5322dc070b47a92f6e05e4/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/d449db5cd7240552bb247d1aed4129fe7216ca170a5322dc070b47a92f6e05e4/kernel.json
2026-09-23T10:12:38.014406Z  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_r80_p80_c24_p24" target="ipu21"
2026-09-23T10:12:38.059655Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/a8e886137e2abb6896092d910e28a9275217f31a0049a9eda56e321d1b3a763d/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/a8e886137e2abb6896092d910e28a9275217f31a0049a9eda56e321d1b3a763d/kernel.json
2026-09-23T10:12:38.059914Z  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_r73_p80_c24_p24" target="ipu21"
2026-09-23T10:12:38.102330Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/6168044d3f520a3424685ed9056db5cc82851f9386f3e043052561c514054525/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/6168044d3f520a3424685ed9056db5cc82851f9386f3e043052561c514054525/kernel.json
2026-09-23T10:12:38.102373Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: close time.busy=367ms time.idle=3.36µs
2026-09-23T10:12:38.102680Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:link_runtime: close time.busy=197µs time.idle=1.95µs
2026-09-23T10:12:38.466909Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: exchange table and repeat patch storage exchange_table_bytes=21780 maximum_uncompressed_repeat_patch_bytes=0 maximum_elided_arithmetic_patch_bytes=0
2026-09-23T10:12:38.988128Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:size_tile_code: close time.busy=16.1ms time.idle=3.85µs
2026-09-23T10:12:38.988161Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: placing generated tile programs linked_end=316156 host_code_base=376832 host_code_bytes=956 generated_code_bytes=4188 exchange_table_bytes=21780 executable_ranges=[(377856, 524288)]
2026-09-23T10:12:38.988193Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: allocated package support memory linked_end=316156 profile_bytes=176 exchange_table_bytes=21780 host_code_bytes=956 host_data_bytes=692 generated_code_bytes=4188 code_address=377856 available_ranges=[(327680, 360448), (362900, 376832), (425984, 949168)]
2026-09-23T10:12:39.086995Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:place_storage: close time.busy=82.6ms time.idle=3.68µs
2026-09-23T10:12:39.804366Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=1
2026-09-23T10:12:39.815338Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=1 transfers=47344 maximum_identical_destinations=1
2026-09-23T10:12:39.815357Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=1 initial_horizon=15159 selected_horizon=15159 endpoint_lower_bound=13320 lower_bound_gap=1839 selected_kind="reused" neighborhood_improvements=0
2026-09-23T10:12:39.876228Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=4
2026-09-23T10:12:39.904418Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=4 transfers=121620 maximum_identical_destinations=1
2026-09-23T10:12:39.904434Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=4 initial_horizon=19737 selected_horizon=19737 endpoint_lower_bound=18048 lower_bound_gap=1689 selected_kind="reused" neighborhood_improvements=0
2026-09-23T10:12:43.675988Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=132190
2026-09-23T10:12:43.942668Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=32435
2026-09-23T10:12:47.158353Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: finished incremental instruction-alignment validation transfers=132190 elapsed_ms=3482 success=true
2026-09-23T10:12:50.028335Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=5 transfers=132190 maximum_identical_destinations=1
2026-09-23T10:12:50.028363Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=5 initial_horizon=1899 selected_horizon=1899 endpoint_lower_bound=1248 lower_bound_gap=651 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T10:12:50.866388Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: finished incremental instruction-alignment validation transfers=32435 elapsed_ms=6923 success=true
2026-09-23T10:12:51.652106Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: compared ordinary and paired exchange schedules phase=3 candidates=9006 ordinary_horizon=70894 paired_horizon=54519 use_paired=true
2026-09-23T10:12:51.864801Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=3 transfers=53424 maximum_identical_destinations=1
2026-09-23T10:12:51.864840Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=3 initial_horizon=54519 selected_horizon=54519 endpoint_lower_bound=50688 lower_bound_gap=3831 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T10:12:57.388425Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=32435
2026-09-23T10:13:01.135153Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=2
2026-09-23T10:13:01.494240Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=2 transfers=766680 maximum_identical_destinations=1
2026-09-23T10:13:01.494286Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=2 initial_horizon=6618 selected_horizon=6618 endpoint_lower_bound=4392 lower_bound_gap=2226 selected_kind="reused" neighborhood_improvements=0
2026-09-23T10:13:05.009160Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: finished incremental instruction-alignment validation transfers=32435 elapsed_ms=7620 success=true
2026-09-23T10:13:09.208154Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: compared ordinary and paired exchange schedules phase=0 candidates=9193 ordinary_horizon=57377 paired_horizon=51931 use_paired=true
2026-09-23T10:13:09.466602Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=0 transfers=32435 maximum_identical_destinations=1
2026-09-23T10:13:09.466651Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=0 initial_horizon=51931 selected_horizon=51931 endpoint_lower_bound=46830 lower_bound_gap=5101 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-23T10:13:09.636930Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: close time.busy=30.5s time.idle=5.46µs
2026-09-23T10:13:09.649433Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: costed final placed program final_cycles=732613 final_exchange=153463
2026-09-23T10:13:10.140097Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:prepare_tile_code: close time.busy=23.4ms time.idle=7.44µs
2026-09-23T10:13:10.179171Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:emit_tile_code: close time.busy=39.0ms time.idle=4.19µs
2026-09-23T10:13:10.326380Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:build_tile_images: close time.busy=120ms time.idle=4.14µs
2026-09-23T10:13:10.603092Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: close time.busy=306s time.idle=7.39µs
2026-09-23T10:13:10.611755Z  INFO writing application package tiles=1472
2026-09-23T10:13:11.118725Z  INFO application package written
2026-09-23T10:13:11.125339Z  INFO reading application package
2026-09-23T10:13:11.410241Z  INFO application package read tiles=1472 compiler=0.1.0
2026-09-23T10:13:11.420198Z  INFO waiting for hardware ownership path=/tmp/ipu-stack-device.lock
2026-09-23T10:13:11.420244Z  INFO acquired hardware ownership path=/tmp/ipu-stack-device.lock
2026-09-23T10:13:11.420250Z  INFO opening IPU device path="/dev/ipu0"
2026-09-23T10:13:11.420289Z  INFO IPU device attached path=/dev/ipu0
2026-09-23T10:13:11.420292Z  INFO initializing IPU device
2026-09-23T10:13:11.428246Z  INFO IPU device initialized
2026-09-23T10:13:11.428251Z  INFO replaying device configuration register_writes=2003
2026-09-23T10:13:11.437662Z  INFO loading application tile_count=1472 final_mark=23
2026-09-23T10:13:11.644075Z  INFO application loaded and initial synchronization completed tile_count=1472
2026-09-23T10:13:12.176796Z  INFO prepared reference inputs elapsed_ms=529
2026-09-23T10:13:12.205300Z  INFO configured host reference BLAS threads=4
referencePrecision=device
2026-09-23T10:13:12.646603Z  INFO evaluated host reference elapsed_ms=469
2026-09-23T10:13:12.648876Z  INFO starting host exchange session startup_mark=23
2026-09-23T10:13:12.648893Z  INFO attaching host exchange pages pages=2
2026-09-23T10:13:12.649587Z  INFO host exchange session started
2026-09-23T10:13:12.649629Z  INFO invoking host exchange call call="initialize" command=0 phases=8
2026-09-23T10:13:12.657891Z  INFO invoking host exchange call call="run" command=0 phases=17
referenceInference=1 numericalTest=PASS
profile=artifacts/gemm-solver-20260923/f16.capnp tiles=1472 samples=52908
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=1471 executionTiles=1472 inputBytes=6718464 weightBytes=19832832 cycles=732378 minimumTileCycles=718602 deviceMicroseconds=488.252 effectiveGemmTflops=118.448 peakTflops=282.624 efficiencyPercent=41.91 maximumAbsoluteError=0.001465
package=artifacts/gemm-solver-20260923/f16.ipuexe tiles=1472 hardwareTest=PASS