2026-09-22T14:53:06.858101Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_runtime: close time.busy=262µs time.idle=5.79µs
2026-09-22T14:53:59.916412Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:gemm_candidates: close time.busy=53.1s time.idle=2.94µs
2026-09-22T14:55:27.251717Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid:gemm_candidates: close time.busy=87.3s time.idle=3.89µs
2026-09-22T14:55:31.904708Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:construct_mid: close time.busy=145s time.idle=13.3µs
2026-09-22T14:55:31.994227Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: wrote planner memory profile path=artifacts/planner-memory-20260922/fp8-b2-n1-384k/memory/baseline-1171249-0000.json scope="baseline" peak_step=2 effective_peak_bytes=235008
2026-09-22T14:55:33.118189Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:group_exchanges: close time.busy=155µs time.idle=3.05µs
2026-09-22T14:55:33.120627Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:merge_copies: close time.busy=2.16ms time.idle=1.48µs
2026-09-22T14:55:33.153393Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:eliminate: eliminated padding clears using kernel access contracts removed=0
2026-09-22T14:55:33.154910Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:eliminate: close time.busy=34.3ms time.idle=967ns
2026-09-22T14:55:33.200281Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:place_provisional_storage: close time.busy=44.6ms time.idle=2.28µs
2026-09-22T14:55:33.524264Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=69984
2026-09-22T14:55:33.827040Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=4 transfers=3096 maximum_identical_destinations=1
2026-09-22T14:55:33.827069Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=4 initial_horizon=15897 selected_horizon=15868 endpoint_lower_bound=15360 lower_bound_gap=508 selected_kind="critical-neighborhood" neighborhood_improvements=1
2026-09-22T14:55:34.067749Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=69984 elapsed_ms=543 success=true
2026-09-22T14:55:34.244644Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=39366
2026-09-22T14:55:34.541800Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=16932
2026-09-22T14:55:34.631354Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=6 transfers=69984 maximum_identical_destinations=1
2026-09-22T14:55:34.631374Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=6 initial_horizon=554 selected_horizon=554 endpoint_lower_bound=384 lower_bound_gap=170 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-22T14:55:35.374532Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=1 transfers=91056 maximum_identical_destinations=1
2026-09-22T14:55:35.374773Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=1 initial_horizon=8422 selected_horizon=7576 endpoint_lower_bound=7056 lower_bound_gap=520 selected_kind="matching-waves" neighborhood_improvements=0
2026-09-22T14:55:35.765743Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=39366 elapsed_ms=1521 success=true
2026-09-22T14:55:36.970265Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=16932 elapsed_ms=2428 success=true
2026-09-22T14:55:37.079536Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=365560
2026-09-22T14:55:37.300428Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=39366
2026-09-22T14:55:38.861114Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=39366 elapsed_ms=1560 success=true
2026-09-22T14:55:39.409246Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=16932
2026-09-22T14:55:39.477597Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: compared ordinary and paired exchange schedules phase=3 candidates=486 ordinary_horizon=6855 paired_horizon=7003 use_paired=false
2026-09-22T14:55:39.533536Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=3 transfers=39366 maximum_identical_destinations=1
2026-09-22T14:55:39.533552Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=3 initial_horizon=6855 selected_horizon=6855 endpoint_lower_bound=6400 lower_bound_gap=455 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-22T14:55:41.851862Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=16932 elapsed_ms=2442 success=true
2026-09-22T14:55:43.025589Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: compared ordinary and paired exchange schedules phase=0 candidates=88 ordinary_horizon=29457 paired_horizon=29529 use_paired=false
2026-09-22T14:55:43.143924Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=0 transfers=16932 maximum_identical_destinations=1
2026-09-22T14:55:43.143946Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=0 initial_horizon=29457 selected_horizon=29457 endpoint_lower_bound=28008 lower_bound_gap=1449 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-22T14:55:43.359837Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=365560 elapsed_ms=6280 success=true
2026-09-22T14:55:48.178807Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: retrying exchange schedule with incremental instruction-alignment validation transfers=909792
2026-09-22T14:55:50.249064Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=2 transfers=365560 maximum_identical_destinations=1
2026-09-22T14:55:50.249094Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=2 initial_horizon=3211 selected_horizon=3211 endpoint_lower_bound=2352 lower_bound_gap=859 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-22T14:56:09.745539Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: finished incremental instruction-alignment validation transfers=909792 elapsed_ms=21566 success=true
2026-09-22T14:56:35.874194Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: prepared large physical exchange phase phase=5 transfers=909792 maximum_identical_destinations=1
2026-09-22T14:56:35.874252Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: optimized physical exchange schedule phase=5 initial_horizon=20277 selected_horizon=20277 endpoint_lower_bound=14976 lower_bound_gap=5301 selected_kind="full-duplex" neighborhood_improvements=0
2026-09-22T14:56:37.961971Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:schedule_provisional_exchanges: close time.busy=64.8s time.idle=4.92µs
2026-09-22T14:56:38.002902Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:plan_kernels: close time.busy=40.8ms time.idle=3.41µs
2026-09-22T14:56:38.004348Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:compile_kernels: close time.busy=1.42ms time.idle=2.30µs
2026-09-22T14:56:38.004588Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:link_runtime: close time.busy=161µs time.idle=802ns
2026-09-22T14:56:38.346249Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: exchange table and repeat patch storage exchange_table_bytes=19668 maximum_uncompressed_repeat_patch_bytes=0 maximum_elided_arithmetic_patch_bytes=0
2026-09-22T14:56:38.829451Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:size_tile_code: close time.busy=49.3ms time.idle=6.30µs
2026-09-22T14:56:38.829513Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: placing generated tile programs linked_end=318984 host_code_base=376832 host_code_bytes=572 generated_code_bytes=1140 exchange_table_bytes=19668 executable_ranges=[(377472, 524288)]
2026-09-22T14:56:38.829567Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: allocated package support memory linked_end=318984 profile_bytes=96 exchange_table_bytes=19668 host_code_bytes=572 host_data_bytes=392 generated_code_bytes=1140 code_address=377472 available_ranges=[(327680, 360448), (362600, 376832), (425984, 949168)]
2026-09-22T14:56:38.904337Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:place_storage: close time.busy=57.9ms time.idle=4.29µs
2026-09-22T14:56:39.145368Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=4
2026-09-22T14:56:39.152794Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=4 transfers=3096 maximum_identical_destinations=1
2026-09-22T14:56:39.152807Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=4 initial_horizon=15868 selected_horizon=15868 endpoint_lower_bound=15360 lower_bound_gap=508 selected_kind="reused" neighborhood_improvements=0
2026-09-22T14:56:39.617006Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=1
2026-09-22T14:56:39.666431Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=1 transfers=91056 maximum_identical_destinations=1
2026-09-22T14:56:39.666450Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=1 initial_horizon=7576 selected_horizon=7576 endpoint_lower_bound=7056 lower_bound_gap=520 selected_kind="reused" neighborhood_improvements=0
2026-09-22T14:56:39.737360Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=6
2026-09-22T14:56:39.771407Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=6 transfers=69984 maximum_identical_destinations=1
2026-09-22T14:56:39.771434Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=6 initial_horizon=554 selected_horizon=554 endpoint_lower_bound=384 lower_bound_gap=170 selected_kind="reused" neighborhood_improvements=0
2026-09-22T14:56:39.917009Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=3
2026-09-22T14:56:39.940924Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=3 transfers=39366 maximum_identical_destinations=1
2026-09-22T14:56:39.940940Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=3 initial_horizon=6855 selected_horizon=6855 endpoint_lower_bound=6400 lower_bound_gap=455 selected_kind="reused" neighborhood_improvements=0
2026-09-22T14:56:42.464370Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=0
2026-09-22T14:56:42.517921Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=0 transfers=16932 maximum_identical_destinations=1
2026-09-22T14:56:42.517948Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=0 initial_horizon=29457 selected_horizon=29457 endpoint_lower_bound=28008 lower_bound_gap=1449 selected_kind="reused" neighborhood_improvements=0
2026-09-22T14:56:45.680074Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=2
2026-09-22T14:56:45.763486Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=2 transfers=365560 maximum_identical_destinations=1
2026-09-22T14:56:45.763513Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=2 initial_horizon=3211 selected_horizon=3211 endpoint_lower_bound=2352 lower_bound_gap=859 selected_kind="reused" neighborhood_improvements=0
2026-09-22T14:56:58.830980Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: reused exchange optimization after validating relocated rows phase=5
2026-09-22T14:56:59.155471Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: prepared large physical exchange phase phase=5 transfers=909792 maximum_identical_destinations=1
2026-09-22T14:56:59.155497Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: optimized physical exchange schedule phase=5 initial_horizon=20277 selected_horizon=20277 endpoint_lower_bound=14976 lower_bound_gap=5301 selected_kind="reused" neighborhood_improvements=0
2026-09-22T14:57:01.829304Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:lower_exchanges: close time.busy=22.9s time.idle=4.99µs
2026-09-22T14:57:01.832802Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: costed final placed program final_cycles=450078 final_exchange=87998
2026-09-22T14:57:02.327107Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:prepare_tile_code: close time.busy=95.5ms time.idle=7.03µs
2026-09-22T14:57:02.353858Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:emit_tile_code: close time.busy=26.7ms time.idle=2.77µs
2026-09-22T14:57:02.509100Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:build_tile_images: close time.busy=149ms time.idle=2.30µs
2026-09-22T14:57:02.699779Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: wrote exact tile placement profile path=artifacts/planner-memory-20260922/fp8-b2-n1-384k/memory/placement-1171249.html
2026-09-22T14:57:02.784180Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: close time.busy=236s time.idle=5.34µs
2026-09-22T14:57:02.790648Z  INFO writing application package tiles=1472
2026-09-22T14:57:03.011696Z  INFO application package written
2026-09-22T14:57:03.012436Z  INFO reading application package
2026-09-22T14:57:03.260050Z  INFO application package read tiles=1472 compiler=0.1.0
2026-09-22T14:57:03.268168Z  INFO waiting for hardware ownership path=/tmp/ipu-stack-device.lock
2026-09-22T14:57:03.268184Z  INFO acquired hardware ownership path=/tmp/ipu-stack-device.lock
2026-09-22T14:57:03.268189Z  INFO opening IPU device path="/dev/ipu0"
2026-09-22T14:57:03.268238Z  INFO IPU device attached path=/dev/ipu0
2026-09-22T14:57:03.268243Z  INFO initializing IPU device
2026-09-22T14:57:03.276097Z  INFO IPU device initialized
2026-09-22T14:57:03.276104Z  INFO replaying device configuration register_writes=2003
2026-09-22T14:57:03.282150Z  INFO loading application tile_count=1472 final_mark=23
2026-09-22T14:57:03.478439Z  INFO application loaded and initial synchronization completed tile_count=1472
2026-09-22T14:57:04.358982Z  INFO prepared reference inputs elapsed_ms=877
2026-09-22T14:57:04.737065Z  INFO configured host reference BLAS threads=4
referencePrecision=device
2026-09-22T14:57:05.306610Z  INFO evaluated host reference elapsed_ms=947
2026-09-22T14:57:05.308458Z  INFO starting host exchange session startup_mark=23
2026-09-22T14:57:05.308480Z  INFO attaching host exchange pages pages=2
2026-09-22T14:57:05.309168Z  INFO host exchange session started
2026-09-22T14:57:05.309216Z  INFO invoking host exchange call call="initialize" command=0 phases=4
2026-09-22T14:57:05.312071Z  INFO invoking host exchange call call="run" command=0 phases=11
referenceInference=1 numericalTest=PASS
profile=artifacts/planner-memory-20260922/fp8-b2-n1-384k/profile.capnp tiles=1472 samples=30741
workload=siglip-mlp-f16-b2-t729-d1152-h4304-n1 benchmark=siglip-mlp-f16 batch=2 tokens=729 rows=1458 dimension=1152 hiddenDimension=4304 blocks=1 biases=false activeTiles=1458 executionTiles=1472 inputBytes=1679616 weightBytes=9934848 cycles=340494 minimumTileCycles=333816 deviceMicroseconds=226.996 effectiveGemmTflops=127.387 peakTflops=282.624 efficiencyPercent=45.07 maximumAbsoluteError=0.005493
package=artifacts/planner-memory-20260922/fp8-b2-n1-384k/model.ipuexe tiles=1472 hardwareTest=PASS