2026-09-12T08:04:26.376995Z WARN PE capacity probe: RoPE omitted; layernorm epsilon remains 1e-6 instead of PE's 1e-5; numerical results are not PE model accuracy 2026-09-12T08:04:26.377952Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="compile_runtime"}: package build phase finished phase="compile_runtime" elapsed_ms=0 success=true 2026-09-12T08:04:35.575415Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: wrote planner memory profile path=artifacts/pe-capacity-20260912/b5/memory/baseline-124031-0000.json scope="baseline" peak_step=115 effective_peak_bytes=559016 2026-09-12T08:04:46.190397Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: screened exchange table geometry expansion_ms=9135 expansion_and_footprint_ms=10614 estimated_row_bytes=126656 heuristic_row_bytes=346020 geometry_fragments_per_tile=15105 fragment_limit=16384 2026-09-12T08:04:46.350295Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: eliminated unread FP8 cast input padding clears removed=110181 2026-09-12T08:04:46.977116Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: placed tile storage tiles=1472 shards=468336 elapsed_us=551653 2026-09-12T08:04:47.159914Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=10 transfers=5670 maximum_identical_destinations=1 2026-09-12T08:04:47.159945Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=10 initial_horizon=25523 selected_horizon=25523 endpoint_lower_bound=12800 lower_bound_gap=12723 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.165771Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=10 tile=45 row_words=31 horizon=25523 2026-09-12T08:04:47.227872Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=65 transfers=1535 maximum_identical_destinations=1 2026-09-12T08:04:47.227900Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=65 initial_horizon=247 selected_horizon=247 endpoint_lower_bound=64 lower_bound_gap=183 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.229476Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=65 tile=854 row_words=47 horizon=247 2026-09-12T08:04:47.288722Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=6 candidates=16 ordinary_horizon=2169 paired_horizon=2280 use_paired=false 2026-09-12T08:04:47.308812Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=45 candidates=16 ordinary_horizon=2239 paired_horizon=2316 use_paired=false 2026-09-12T08:04:47.324812Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=9 transfers=1918 maximum_identical_destinations=1 2026-09-12T08:04:47.324842Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=9 initial_horizon=16634 selected_horizon=16634 endpoint_lower_bound=8832 lower_bound_gap=7802 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.333455Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=9 tile=25 row_words=53 horizon=16634 2026-09-12T08:04:47.334307Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=66 transfers=1279 maximum_identical_destinations=1 2026-09-12T08:04:47.334326Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=66 initial_horizon=2617 selected_horizon=2617 endpoint_lower_bound=2558 lower_bound_gap=59 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.335320Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=66 tile=0 row_words=1284 horizon=2617 2026-09-12T08:04:47.362486Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=77 transfers=1578 maximum_identical_destinations=1 2026-09-12T08:04:47.362507Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=77 initial_horizon=2641 selected_horizon=2641 endpoint_lower_bound=768 lower_bound_gap=1873 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.363867Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=53 transfers=1020 maximum_identical_destinations=1 2026-09-12T08:04:47.363888Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=53 initial_horizon=869 selected_horizon=869 endpoint_lower_bound=256 lower_bound_gap=613 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.364923Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=77 tile=1339 row_words=14 horizon=2641 2026-09-12T08:04:47.365035Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=53 tile=866 row_words=10 horizon=869 2026-09-12T08:04:47.379993Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=78 transfers=1344 maximum_identical_destinations=1 2026-09-12T08:04:47.380007Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=78 initial_horizon=1782 selected_horizon=1782 endpoint_lower_bound=1680 lower_bound_gap=102 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.381053Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=78 tile=63 row_words=26 horizon=1782 2026-09-12T08:04:47.389074Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=80 transfers=1279 maximum_identical_destinations=1 2026-09-12T08:04:47.389101Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=80 initial_horizon=2713 selected_horizon=2713 endpoint_lower_bound=2558 lower_bound_gap=155 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.391968Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=80 tile=0 row_words=1284 horizon=2713 2026-09-12T08:04:47.397734Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=4 candidates=16 ordinary_horizon=1215 paired_horizon=1291 use_paired=false 2026-09-12T08:04:47.398456Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=55 transfers=1535 maximum_identical_destinations=1 2026-09-12T08:04:47.398475Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=55 initial_horizon=247 selected_horizon=247 endpoint_lower_bound=64 lower_bound_gap=183 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.399802Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=55 tile=854 row_words=47 horizon=247 2026-09-12T08:04:47.406915Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=75 transfers=1473 maximum_identical_destinations=1 2026-09-12T08:04:47.406941Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=75 initial_horizon=10290 selected_horizon=10290 endpoint_lower_bound=10232 lower_bound_gap=58 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.408119Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=75 tile=0 row_words=1477 horizon=10290 2026-09-12T08:04:47.408402Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=68 candidates=160 ordinary_horizon=231 paired_horizon=1250 use_paired=false 2026-09-12T08:04:47.415043Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=79 transfers=1535 maximum_identical_destinations=1 2026-09-12T08:04:47.415058Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=79 initial_horizon=247 selected_horizon=247 endpoint_lower_bound=64 lower_bound_gap=183 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.416700Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=79 tile=854 row_words=47 horizon=247 2026-09-12T08:04:47.419049Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=56 transfers=1279 maximum_identical_destinations=1 2026-09-12T08:04:47.419071Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=56 initial_horizon=275 selected_horizon=275 endpoint_lower_bound=64 lower_bound_gap=211 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.420031Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=56 tile=0 row_words=45 horizon=275 2026-09-12T08:04:47.421076Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=83 transfers=1020 maximum_identical_destinations=1 2026-09-12T08:04:47.421102Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=83 initial_horizon=1152 selected_horizon=1152 endpoint_lower_bound=256 lower_bound_gap=896 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.422568Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=83 tile=946 row_words=10 horizon=1152 2026-09-12T08:04:47.434594Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=58 transfers=1400 maximum_identical_destinations=1 2026-09-12T08:04:47.434606Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=58 initial_horizon=376 selected_horizon=376 endpoint_lower_bound=280 lower_bound_gap=96 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.435696Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=58 tile=29 row_words=41 horizon=376 2026-09-12T08:04:47.438724Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=48 transfers=1974 maximum_identical_destinations=1 2026-09-12T08:04:47.438746Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=48 initial_horizon=18173 selected_horizon=18173 endpoint_lower_bound=9344 lower_bound_gap=8829 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.441042Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=48 tile=39 row_words=58 horizon=18173 2026-09-12T08:04:47.460264Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=63 transfers=1020 maximum_identical_destinations=1 2026-09-12T08:04:47.460279Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=63 initial_horizon=1152 selected_horizon=1152 endpoint_lower_bound=256 lower_bound_gap=896 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.461669Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=63 tile=946 row_words=10 horizon=1152 2026-09-12T08:04:47.466356Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=76 candidates=105 ordinary_horizon=10346 paired_horizon=12023 use_paired=false 2026-09-12T08:04:47.466756Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=71 transfers=8180 maximum_identical_destinations=1 2026-09-12T08:04:47.466775Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=71 initial_horizon=2237 selected_horizon=2237 endpoint_lower_bound=768 lower_bound_gap=1469 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.473495Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=71 tile=10 row_words=33 horizon=2237 2026-09-12T08:04:47.476227Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=62 candidates=75 ordinary_horizon=2681 paired_horizon=6122 use_paired=false 2026-09-12T08:04:47.476241Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=72 transfers=1330 maximum_identical_destinations=1 2026-09-12T08:04:47.476253Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=72 initial_horizon=1702 selected_horizon=1702 endpoint_lower_bound=1600 lower_bound_gap=102 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.477351Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=72 tile=132 row_words=14 horizon=1702 2026-09-12T08:04:47.495609Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=81 transfers=1279 maximum_identical_destinations=1 2026-09-12T08:04:47.495625Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=81 initial_horizon=2617 selected_horizon=2617 endpoint_lower_bound=2558 lower_bound_gap=59 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.496615Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=81 tile=0 row_words=1284 horizon=2617 2026-09-12T08:04:47.499452Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=74 transfers=2026 maximum_identical_destinations=1 2026-09-12T08:04:47.499468Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=74 initial_horizon=226 selected_horizon=226 endpoint_lower_bound=8 lower_bound_gap=218 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.501986Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=74 tile=1202 row_words=12 horizon=226 2026-09-12T08:04:47.516894Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=82 candidates=75 ordinary_horizon=2681 paired_horizon=6122 use_paired=false 2026-09-12T08:04:47.524327Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=49 transfers=7150 maximum_identical_destinations=1 2026-09-12T08:04:47.524350Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=49 initial_horizon=23541 selected_horizon=23541 endpoint_lower_bound=11200 lower_bound_gap=12341 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.529499Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=73 transfers=2153 maximum_identical_destinations=1 2026-09-12T08:04:47.529516Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=73 initial_horizon=226 selected_horizon=226 endpoint_lower_bound=80 lower_bound_gap=146 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.531136Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=49 tile=1229 row_words=54 horizon=23541 2026-09-12T08:04:47.531855Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=73 tile=90 row_words=37 horizon=226 2026-09-12T08:04:47.563659Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=37 transfers=4353 maximum_identical_destinations=1 2026-09-12T08:04:47.563688Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=37 initial_horizon=8221 selected_horizon=8221 endpoint_lower_bound=6144 lower_bound_gap=2077 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.567922Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=37 tile=339 row_words=28 horizon=8221 2026-09-12T08:04:47.570478Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=40 transfers=2279 maximum_identical_destinations=1 2026-09-12T08:04:47.570500Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=40 initial_horizon=18669 selected_horizon=18669 endpoint_lower_bound=9984 lower_bound_gap=8685 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.579289Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=40 tile=104 row_words=55 horizon=18669 2026-09-12T08:04:47.613773Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=38 transfers=4353 maximum_identical_destinations=1 2026-09-12T08:04:47.613796Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=38 initial_horizon=12316 selected_horizon=12316 endpoint_lower_bound=10240 lower_bound_gap=2076 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.617175Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=38 tile=484 row_words=34 horizon=12316 2026-09-12T08:04:47.771929Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=34 transfers=8519 maximum_identical_destinations=1 2026-09-12T08:04:47.771959Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=34 initial_horizon=19991 selected_horizon=19991 endpoint_lower_bound=10624 lower_bound_gap=9367 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.788423Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=34 tile=32 row_words=147 horizon=19991 2026-09-12T08:04:47.788464Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=26 transfers=1313 maximum_identical_destinations=1 2026-09-12T08:04:47.788485Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=26 initial_horizon=17061 selected_horizon=17061 endpoint_lower_bound=8576 lower_bound_gap=8485 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.795435Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=26 tile=20 row_words=61 horizon=17061 2026-09-12T08:04:47.915696Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=27 transfers=10542 maximum_identical_destinations=1 2026-09-12T08:04:47.915733Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=27 initial_horizon=15751 selected_horizon=15751 endpoint_lower_bound=7840 lower_bound_gap=7911 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:47.929344Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=27 tile=54 row_words=69 horizon=15751 2026-09-12T08:04:47.970281Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=31 candidates=16 ordinary_horizon=1226 paired_horizon=1121 use_paired=false 2026-09-12T08:04:48.527013Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=24 transfers=46127 maximum_identical_destinations=1 2026-09-12T08:04:48.527051Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=24 initial_horizon=4601 selected_horizon=4601 endpoint_lower_bound=2560 lower_bound_gap=2041 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:48.550691Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=24 tile=196 row_words=269 horizon=4601 2026-09-12T08:04:48.718163Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=0 candidates=14335 ordinary_horizon=6493 paired_horizon=10798 use_paired=false 2026-09-12T08:04:48.753917Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=33 transfers=8655 maximum_identical_destinations=1 2026-09-12T08:04:48.753957Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=33 initial_horizon=46467 selected_horizon=46467 endpoint_lower_bound=40480 lower_bound_gap=5987 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:48.763403Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=0 transfers=14425 maximum_identical_destinations=1 2026-09-12T08:04:48.763422Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=0 initial_horizon=6493 selected_horizon=6493 endpoint_lower_bound=5760 lower_bound_gap=733 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:48.780703Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=0 tile=13 row_words=241 horizon=6493 2026-09-12T08:04:48.797439Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=33 tile=234 row_words=548 horizon=46467 2026-09-12T08:04:48.902267Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=47 transfers=14425 maximum_identical_destinations=1 2026-09-12T08:04:48.902296Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=47 initial_horizon=15788 selected_horizon=15788 endpoint_lower_bound=15680 lower_bound_gap=108 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:48.924637Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=47 tile=53 row_words=337 horizon=15788 2026-09-12T08:04:48.963569Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=35 transfers=143360 maximum_identical_destinations=1 2026-09-12T08:04:48.963608Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=35 initial_horizon=14557 selected_horizon=14557 endpoint_lower_bound=12880 lower_bound_gap=1677 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:48.989546Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=25 transfers=20195 maximum_identical_destinations=1 2026-09-12T08:04:48.989579Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=25 initial_horizon=6515 selected_horizon=6515 endpoint_lower_bound=5600 lower_bound_gap=915 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:49.011698Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=25 tile=305 row_words=304 horizon=6515 2026-09-12T08:04:49.080696Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=35 tile=525 row_words=391 horizon=14557 2026-09-12T08:04:49.797497Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=8 candidates=11540 ordinary_horizon=25706 paired_horizon=23386 use_paired=false 2026-09-12T08:04:49.871447Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=8 transfers=11540 maximum_identical_destinations=1 2026-09-12T08:04:49.871470Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=8 initial_horizon=25706 selected_horizon=25706 endpoint_lower_bound=25600 lower_bound_gap=106 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:49.890504Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=8 tile=276 row_words=448 horizon=25706 2026-09-12T08:04:51.245004Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=3 transfers=186474 maximum_identical_destinations=1 2026-09-12T08:04:51.245036Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=3 initial_horizon=2347 selected_horizon=2347 endpoint_lower_bound=2048 lower_bound_gap=299 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:51.403379Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=3 tile=480 row_words=593 horizon=2347 2026-09-12T08:04:53.365467Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=39 candidates=43275 ordinary_horizon=37661 paired_horizon=33970 use_paired=false 2026-09-12T08:04:53.530191Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=39 transfers=43275 maximum_identical_destinations=1 2026-09-12T08:04:53.530242Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=39 initial_horizon=37661 selected_horizon=37661 endpoint_lower_bound=36000 lower_bound_gap=1661 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:53.594537Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=39 tile=292 row_words=728 horizon=37661 2026-09-12T08:04:54.095642Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=41 transfers=323700 maximum_identical_destinations=1 2026-09-12T08:04:54.095677Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=41 initial_horizon=25264 selected_horizon=25264 endpoint_lower_bound=18000 lower_bound_gap=7264 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:54.469653Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=41 tile=466 row_words=842 horizon=25264 2026-09-12T08:04:55.529007Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=42 candidates=8 ordinary_horizon=2496 paired_horizon=2396 use_paired=false 2026-09-12T08:04:55.764967Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=42 transfers=184470 maximum_identical_destinations=1 2026-09-12T08:04:55.765004Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=42 initial_horizon=2496 selected_horizon=2496 endpoint_lower_bound=2048 lower_bound_gap=448 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:55.919796Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=42 tile=531 row_words=493 horizon=2496 2026-09-12T08:04:56.489938Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=28 candidates=8 ordinary_horizon=2388 paired_horizon=2370 use_paired=false 2026-09-12T08:04:56.717986Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=28 transfers=184488 maximum_identical_destinations=1 2026-09-12T08:04:56.718015Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=28 initial_horizon=2388 selected_horizon=2388 endpoint_lower_bound=2048 lower_bound_gap=340 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:04:56.898595Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=28 tile=472 row_words=542 horizon=2388 2026-09-12T08:05:04.996885Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=57 transfers=207680 maximum_identical_destinations=1 2026-09-12T08:05:04.996942Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=57 initial_horizon=10334 selected_horizon=10334 endpoint_lower_bound=6656 lower_bound_gap=3678 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:05:05.328989Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=57 tile=179 row_words=2077 horizon=10334 2026-09-12T08:05:11.045055Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=13 transfers=138367 maximum_identical_destinations=1 2026-09-12T08:05:11.045125Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=13 initial_horizon=11929 selected_horizon=11929 endpoint_lower_bound=7680 lower_bound_gap=4249 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:05:11.228403Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=13 tile=576 row_words=975 horizon=11929 2026-09-12T08:05:11.325798Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=2 transfers=7020 maximum_identical_destinations=1 2026-09-12T08:05:11.325827Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=2 initial_horizon=8721 selected_horizon=8721 endpoint_lower_bound=7200 lower_bound_gap=1521 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:05:11.332153Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=2 tile=1335 row_words=41 horizon=8721 2026-09-12T08:05:22.720444Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=36 candidates=32 ordinary_horizon=14740 paired_horizon=13690 use_paired=false 2026-09-12T08:05:24.063291Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=36 transfers=737996 maximum_identical_destinations=1 2026-09-12T08:05:24.063335Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=36 initial_horizon=14740 selected_horizon=14740 endpoint_lower_bound=8192 lower_bound_gap=6548 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:05:25.157026Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=36 tile=313 row_words=1728 horizon=14740 2026-09-12T08:05:25.577441Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=52 candidates=320 ordinary_horizon=274 paired_horizon=1284 use_paired=false 2026-09-12T08:05:39.960934Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=50 candidates=16 ordinary_horizon=4910 paired_horizon=5604 use_paired=false 2026-09-12T08:05:40.584641Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=50 transfers=368928 maximum_identical_destinations=1 2026-09-12T08:05:40.584687Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=50 initial_horizon=4910 selected_horizon=4910 endpoint_lower_bound=4096 lower_bound_gap=814 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:05:41.031754Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=50 tile=179 row_words=1054 horizon=4910 2026-09-12T08:05:53.026060Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: compared ordinary and paired exchange schedules phase=11 candidates=24 ordinary_horizon=7210 paired_horizon=7016 use_paired=false 2026-09-12T08:05:54.135453Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=11 transfers=553420 maximum_identical_destinations=1 2026-09-12T08:05:54.135488Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=11 initial_horizon=7210 selected_horizon=7210 endpoint_lower_bound=6144 lower_bound_gap=1066 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:05:55.033181Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=11 tile=472 row_words=1523 horizon=7210 2026-09-12T08:05:55.747161Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: prepared large physical exchange phase phase=1 transfers=37896 maximum_identical_destinations=1 2026-09-12T08:05:55.747283Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: optimized physical exchange schedule phase=1 initial_horizon=9201 selected_horizon=9201 endpoint_lower_bound=4224 lower_bound_gap=4977 selected_kind="balanced-compact-streams" neighborhood_improvements=0 2026-09-12T08:05:55.768032Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: finished large physical exchange phase phase=1 tile=8 row_words=1073 horizon=9201 2026-09-12T08:05:55.876528Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="plan_kernels"}: package build phase finished phase="plan_kernels" elapsed_ms=93 success=true 2026-09-12T08:05:55.877039Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/gemm_f16_amp.S name="gemm_f8_interleaved_k64_c16_r10_r10" target=ipu21 2026-09-12T08:05:55.928987Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/5bad5e72bc3f9296daee6bcce99d35b21dffddc619e73131c78808425c3776f2/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/5bad5e72bc3f9296daee6bcce99d35b21dffddc619e73131c78808425c3776f2/kernel.json 2026-09-12T08:05:55.929447Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/gemm_f16_amp.S name="gemm_f8_interleaved_k96_c16_r10_r10" target=ipu21 2026-09-12T08:05:55.977932Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/4d6c8dbc9e671ca3253c1e7c333a0615bc58cc7aac1a12676f510e44319f5201/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/4d6c8dbc9e671ca3253c1e7c333a0615bc58cc7aac1a12676f510e44319f5201/kernel.json 2026-09-12T08:05:55.978229Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/gemm_f16_amp.S name="gemm_f8_interleaved_k96_c32_r10_r10" target=ipu21 2026-09-12T08:05:56.016525Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/3c59a0f988c8aa19d625931d0064027de547f02b7c0d11124aa395b63cfc4996/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/3c59a0f988c8aa19d625931d0064027de547f02b7c0d11124aa395b63cfc4996/kernel.json 2026-09-12T08:05:56.016991Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/gemm_f16_amp.S name="gemm_f8_interleaved_k128_c128_r80_r90" target=ipu21 2026-09-12T08:05:56.062493Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/a3222165b343be716696fe7191d6e338287330748f32028eeec0dc49e7eca257/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/a3222165b343be716696fe7191d6e338287330748f32028eeec0dc49e7eca257/kernel.json 2026-09-12T08:05:56.062958Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/gemm_f16_amp.S name="gemm_f8_interleaved_k160_c192_r60_r70" target=ipu21 2026-09-12T08:05:56.107666Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/2adf53e0a1e2920927d8cb69a855f23ff59116e31a4504979b134b2d806d0825/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/2adf53e0a1e2920927d8cb69a855f23ff59116e31a4504979b134b2d806d0825/kernel.json 2026-09-12T08:05:56.108132Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/gemm_f16_amp.S name="gemm_f8_interleaved_k160_c208_r60_r70" target=ipu21 2026-09-12T08:05:56.155293Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/a79c0cfdd6709540b1a6a7de8348e716769d37ca2734a927ed921b467828534d/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/a79c0cfdd6709540b1a6a7de8348e716769d37ca2734a927ed921b467828534d/kernel.json 2026-09-12T08:05:56.155767Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/gemm_f16_amp.S name="gemm_f8_interleaved_k192_c16_r10_r10" target=ipu21 2026-09-12T08:05:56.203065Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/83df25a8a4a1d946f1d33cdef862011b1ef3ef2e3d53e848b923ad7154ea3c00/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/83df25a8a4a1d946f1d33cdef862011b1ef3ef2e3d53e848b923ad7154ea3c00/kernel.json 2026-09-12T08:05:56.203610Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/gemm_f16_amp.S name="gemm_f8_interleaved_k224_c144_r130_r140" target=ipu21 2026-09-12T08:05:56.249760Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/7b6799d39759d3de57c48b9187fdeb59b85d442ea67005c5f9504e95890ed4c5/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/7b6799d39759d3de57c48b9187fdeb59b85d442ea67005c5f9504e95890ed4c5/kernel.json 2026-09-12T08:05:56.250266Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/gemm_f16_amp.S name="gemm_f8_interleaved_k224_c160_r130_r140" target=ipu21 2026-09-12T08:05:56.294709Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/276a03c5bfe806d28d6261357c9712f4e1453974d33c995b7eead9648a07a717/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/276a03c5bfe806d28d6261357c9712f4e1453974d33c995b7eead9648a07a717/kernel.json 2026-09-12T08:05:56.295143Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/gemm_f16_amp.S name="gemm_f8_interleaved_k256_c128_r190_r200" target=ipu21 2026-09-12T08:05:56.337247Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/54ed5779796ad56a7cb185ae2fdeab56d4f1b0c32f786251bbc6be9b33c94b54/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/54ed5779796ad56a7cb185ae2fdeab56d4f1b0c32f786251bbc6be9b33c94b54/kernel.json 2026-09-12T08:05:56.337712Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/gemm_f16_amp.S name="gemm_f8_interleaved_k288_c128_r240_r250" target=ipu21 2026-09-12T08:05:56.380513Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/15ec0ea01f5e852003b267cdf7d19733ac1854301efc102f149e8ffe87717ff7/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/15ec0ea01f5e852003b267cdf7d19733ac1854301efc102f149e8ffe87717ff7/kernel.json 2026-09-12T08:05:56.380714Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/gemm_f16_amp.S name="gemm_f8_interleaved_k352_c96_r220_r230" target=ipu21 2026-09-12T08:05:56.424468Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/71d55a654a3508f94ce4657ab72da489d522b5731b3ba7a647be0d5d44aeccfa/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/71d55a654a3508f94ce4657ab72da489d522b5731b3ba7a647be0d5d44aeccfa/kernel.json 2026-09-12T08:05:56.424949Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/gemm_f16_amp.S name="gemm_f8_interleaved_k352_c112_r220_r230" target=ipu21 2026-09-12T08:05:56.469379Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/42d5bf459b9a9989219be3d5d19fac3359f2683ff0977e97ea01e82a4914c799/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/42d5bf459b9a9989219be3d5d19fac3359f2683ff0977e97ea01e82a4914c799/kernel.json 2026-09-12T08:05:56.470531Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/gemm_f16_amp.S name="gemm_f16_k64_c64_r33_r33" target=ipu21 2026-09-12T08:05:56.512795Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/6f66a3becc8ada7f7fa5c7b85eee10d08d4d8bff9fc9f9efdaa8bc1bc2fa34d5/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/6f66a3becc8ada7f7fa5c7b85eee10d08d4d8bff9fc9f9efdaa8bc1bc2fa34d5/kernel.json 2026-09-12T08:05:56.513549Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/gemm_f16_amp.S name="gemm_f16_interleaved_k64_c64_r32_r33" target=ipu21 2026-09-12T08:05:56.568643Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/294e0d2fc9db74d27bdc199533ced7e6761ad7b2de3e55edcad5a3b45a6d448e/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/294e0d2fc9db74d27bdc199533ced7e6761ad7b2de3e55edcad5a3b45a6d448e/kernel.json 2026-09-12T08:05:56.569084Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/gemm_f16_amp.S name="gemm_f16_interleaved_k128_c32_r1_r1" target=ipu21 2026-09-12T08:05:56.614487Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/a66b6fbee7a30e88a5e4ac747eaf6602dc41cd419bc2f321ad4f5333cfda387d/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/a66b6fbee7a30e88a5e4ac747eaf6602dc41cd419bc2f321ad4f5333cfda387d/kernel.json 2026-09-12T08:05:56.615838Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/rearrange_amp_left_f16.S name="rearrange_f16_codelet_o0_r15_p16_c76_p128" target=ipu21 2026-09-12T08:05:56.659746Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/8c3e9b42679271d3e9e490ba58f78a7d0d4a12b942a22a8e759e63dd4e9e180d/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/8c3e9b42679271d3e9e490ba58f78a7d0d4a12b942a22a8e759e63dd4e9e180d/kernel.json 2026-09-12T08:05:56.659872Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/rearrange_amp_left_f16.S name="rearrange_f16_codelet_o0_r15_p16_c128_p128" target=ipu21 2026-09-12T08:05:56.703484Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/2743811dfc2580ab2d35eeb4fd010f654368b85f0290bd9c8d80bad2a8007252/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/2743811dfc2580ab2d35eeb4fd010f654368b85f0290bd9c8d80bad2a8007252/kernel.json 2026-09-12T08:05:56.703712Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/rearrange_amp_left_f16.S name="rearrange_f16_codelet_o0_r16_p16_c76_p128" target=ipu21 2026-09-12T08:05:56.747336Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/d188c2389a5da763875015cb90b9cb797396ab829767504bac21055b571fb62e/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/d188c2389a5da763875015cb90b9cb797396ab829767504bac21055b571fb62e/kernel.json 2026-09-12T08:05:56.747702Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/rearrange_amp_left_f16.S name="rearrange_f16_codelet_o0_r16_p16_c128_p128" target=ipu21 2026-09-12T08:05:56.792777Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/f533de81b788da5d81d9c39a45753bead55e8584856acb8c8de98b12fc323f5f/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/f533de81b788da5d81d9c39a45753bead55e8584856acb8c8de98b12fc323f5f/kernel.json 2026-09-12T08:05:56.793142Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/rearrange_amp_left_f16.S name="rearrange_f16_codelet_o0_r18_p18_c76_p128" target=ipu21 2026-09-12T08:05:56.848193Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/7efc860230dcbf654571f3ae04de33c6f128df066e68ce079df8b069f5a9ab47/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/7efc860230dcbf654571f3ae04de33c6f128df066e68ce079df8b069f5a9ab47/kernel.json 2026-09-12T08:05:56.848348Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/rearrange_amp_left_f16.S name="rearrange_f16_codelet_o0_r18_p18_c128_p128" target=ipu21 2026-09-12T08:05:56.893571Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/b73b6a3b1b6fa58522d39535bf34cac84ef4c4f3af107e3e32e221e6d99d9df7/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/b73b6a3b1b6fa58522d39535bf34cac84ef4c4f3af107e3e32e221e6d99d9df7/kernel.json 2026-09-12T08:05:56.893713Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/rearrange_amp_left_f16.S name="rearrange_f16_codelet_o0_r32_p32_c64_p64" target=ipu21 2026-09-12T08:05:56.939154Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/163214d98f8b56e929de24c88e9ad3e347a1ade7ca0a9596d59081a49a9f0354/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/163214d98f8b56e929de24c88e9ad3e347a1ade7ca0a9596d59081a49a9f0354/kernel.json 2026-09-12T08:05:56.939457Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/rearrange_amp_left_f16.S name="rearrange_f16_codelet_o0_r33_p33_c64_p64" target=ipu21 2026-09-12T08:05:56.984136Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/105389dda43cc49c9ac7af0c2f47de48b12130b552df246329736a00fbef1d7e/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/105389dda43cc49c9ac7af0c2f47de48b12130b552df246329736a00fbef1d7e/kernel.json 2026-09-12T08:05:56.984800Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/rearrange_f16.cpp name="rearrange_row_major_to_amp_f16_o1_r32_p32_c128_p128_codelet" target=ipu21 2026-09-12T08:05:57.314086Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/1c441bd297f6efc30cfcdb5c62ba8b21aebf80c01bb9001451744843d7381206/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/1c441bd297f6efc30cfcdb5c62ba8b21aebf80c01bb9001451744843d7381206/kernel.json 2026-09-12T08:05:57.314486Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/worker_call.S name="rearrange_row_major_to_amp_f16_o1_r32_p32_c128_p128_wrapper" target=ipu21 2026-09-12T08:05:57.357637Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/9d392ce01ddf0d74b4b743c9bc5bdfba723b6dfd7fdf312b9b5c9dda4c5ff0d9/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/9d392ce01ddf0d74b4b743c9bc5bdfba723b6dfd7fdf312b9b5c9dda4c5ff0d9/kernel.json 2026-09-12T08:05:57.357778Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/rearrange_f16.cpp name="rearrange_row_major_to_amp_f16_o1_r64_p64_c64_p64_codelet" target=ipu21 2026-09-12T08:05:57.596086Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/5b18756ef18e1ab17999c0dae395b0a7c88e3924d8be62f2030c6764dc9365c8/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/5b18756ef18e1ab17999c0dae395b0a7c88e3924d8be62f2030c6764dc9365c8/kernel.json 2026-09-12T08:05:57.596427Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/worker_call.S name="rearrange_row_major_to_amp_f16_o1_r64_p64_c64_p64_wrapper" target=ipu21 2026-09-12T08:05:57.641994Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/9f6159544bc1872551251210be655783213b0dd2143ae9e665e02580a2cbdd54/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/9f6159544bc1872551251210be655783213b0dd2143ae9e665e02580a2cbdd54/kernel.json 2026-09-12T08:05:57.642326Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/rearrange_block_major_f16.S name="rearrange_f16_codelet_o2_r0_p64_c64_p64_b64x16" target=ipu21 2026-09-12T08:05:57.692293Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/1dc2583819c9c0f6433ee65ab23b849f7dc34ac6549a588b1256f13c0f9f2954/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/1dc2583819c9c0f6433ee65ab23b849f7dc34ac6549a588b1256f13c0f9f2954/kernel.json 2026-09-12T08:05:57.692517Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: compiling kernel source=artifacts/attention-residual-20260911/source/rearrange_block_major_f16.S name="rearrange_f16_codelet_o2_r64_p64_c64_p64_b64x16" target=ipu21 2026-09-12T08:05:57.742344Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: kernel artifact created object=/srv/home/.cache/ipu-stack/kernels/74022d9c81c3bb187deb143c2a82306e741544db01b0a105c7f3a0f20b1d2fbf/kernel.o metadata=/srv/home/.cache/ipu-stack/kernels/74022d9c81c3bb187deb143c2a82306e741544db01b0a105c7f3a0f20b1d2fbf/kernel.json 2026-09-12T08:05:57.743872Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="compile_kernels"}: package build phase finished phase="compile_kernels" elapsed_ms=1867 success=true 2026-09-12T08:05:57.752047Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="link_runtime"}: package build phase finished phase="link_runtime" elapsed_ms=0 success=true 2026-09-12T08:05:58.078788Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: exchange table and repeat patch storage exchange_table_bytes=74148 maximum_uncompressed_repeat_patch_bytes=768 maximum_elided_arithmetic_patch_bytes=768 2026-09-12T08:06:00.315460Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="size_tile_code"}: package build phase finished phase="size_tile_code" elapsed_ms=107 success=true 2026-09-12T08:06:00.315534Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: placing generated tile programs linked_end=392184 host_code_base=392184 host_code_bytes=8128 generated_code_bytes=13172 exchange_table_bytes=74148 executable_ranges=[(377828, 377832), (378220, 378224), (379620, 379624), (380412, 380416), (381204, 381208), (381500, 381504), (383076, 383080), (383388, 383392), (383700, 383704), (384284, 384288), (386300, 386304), (388012, 388016), (390028, 390032), (391436, 391440), (400376, 524288)] 2026-09-12T08:06:00.315572Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: allocated package support memory linked_end=392184 profile_bytes=0 exchange_table_bytes=74148 host_code_bytes=8128 host_data_bytes=5256 generated_code_bytes=13172 code_address=400376 available_ranges=[(327680, 360448), (367464, 376832), (507904, 949168)] 2026-09-12T08:06:00.830856Z DEBUG tile allocation does not fit tile=736 class=Ipu21Standard bytes=77448 alignment=32 first=105 last=107 free=[(327680, 360448), (376680, 376832), (510240, 524288), (912640, 949168)] active=[(4294967295, 524288, 73728), (4294967295, 598016, 73728), (4294967295, 671744, 61440), (4294967295, 733184, 24576), (4294967295, 367464, 3072), (4294967295, 370536, 3072), (4294967295, 373608, 1024), (4294967295, 374632, 1024), (4294967295, 375656, 1024), (4294967295, 507904, 1024), (4294967295, 508928, 800), (4294967295, 509728, 512), (105, 757760, 154880)] tensor_type=TensorType { shape: TensorShape([5, 577, 1024]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1443, replicas: 37, axes: [AxisTiling { axis: FromEnd(1), partitions: 3, padding_groups: 1, block_size: 352, padding_multiple: 352, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(2), partitions: 13, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.830837Z DEBUG tile allocation does not fit tile=92 class=Ipu21Standard bytes=80968 alignment=32 first=129 last=138 free=[(337920, 360448), (376680, 376832), (518144, 519664), (522520, 524288), (757760, 768000), (929920, 949168)] active=[(4294967295, 524288, 73728), (4294967295, 598016, 73728), (4294967295, 671744, 61440), (4294967295, 733184, 24576), (4294967295, 367464, 3072), (4294967295, 370536, 3072), (4294967295, 373608, 1536), (4294967295, 375144, 1024), (4294967295, 519664, 1024), (4294967295, 520688, 1024), (4294967295, 521712, 808), (4294967295, 376168, 512), (171, 327680, 10240), (164, 507904, 10240), (129, 768000, 161920)] tensor_type=TensorType { shape: TensorShape([5, 577, 1024]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1443, replicas: 37, axes: [AxisTiling { axis: FromEnd(1), partitions: 3, padding_groups: 1, block_size: 352, padding_multiple: 352, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(2), partitions: 13, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.830831Z DEBUG tile allocation does not fit tile=184 class=Ipu21Standard bytes=80968 alignment=32 first=129 last=131 free=[(337920, 360448), (376680, 376832), (518144, 519664), (522520, 524288), (757760, 768000), (929920, 949168)] active=[(4294967295, 524288, 73728), (4294967295, 598016, 73728), (4294967295, 671744, 61440), (4294967295, 733184, 24576), (4294967295, 367464, 3072), (4294967295, 370536, 3072), (4294967295, 373608, 1536), (4294967295, 375144, 1024), (4294967295, 519664, 1024), (4294967295, 520688, 1024), (4294967295, 521712, 808), (4294967295, 376168, 512), (162, 327680, 10240), (155, 507904, 10240), (129, 768000, 161920)] tensor_type=TensorType { shape: TensorShape([5, 577, 1024]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1443, replicas: 37, axes: [AxisTiling { axis: FromEnd(1), partitions: 3, padding_groups: 1, block_size: 352, padding_multiple: 352, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(2), partitions: 13, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.831017Z DEBUG tile allocation does not fit tile=0 class=Ipu21Standard bytes=80968 alignment=32 first=153 last=161 free=[(337920, 360448), (376680, 376832), (518144, 519664), (522520, 524288), (757760, 786432), (948352, 949168)] active=[(4294967295, 524288, 73728), (4294967295, 598016, 73728), (4294967295, 671744, 61440), (4294967295, 733184, 24576), (4294967295, 367464, 3072), (4294967295, 370536, 3072), (4294967295, 373608, 1536), (4294967295, 375144, 1024), (4294967295, 519664, 1024), (4294967295, 520688, 1024), (4294967295, 521712, 808), (4294967295, 376168, 512), (201, 327680, 10240), (194, 507904, 10240), (153, 786432, 161920)] tensor_type=TensorType { shape: TensorShape([5, 577, 1024]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1443, replicas: 37, axes: [AxisTiling { axis: FromEnd(1), partitions: 3, padding_groups: 1, block_size: 352, padding_multiple: 352, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(2), partitions: 13, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.831055Z DEBUG tile allocation does not fit tile=736 class=Ipu21Standard bytes=61440 alignment=8 first=0 last=4294967295 free=[(356808, 360448), (367464, 376832), (507904, 524288), (904072, 949168)] active=[] tensor_type=TensorType { shape: TensorShape([1024, 3072]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 1344, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 7, padding_groups: 1, block_size: 32, padding_multiple: 32, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(1), partitions: 192, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.831082Z DEBUG tile allocation does not fit tile=92 class=Ipu21Standard bytes=61440 alignment=8 first=0 last=4294967295 free=[(359048, 360448), (367464, 376832), (507904, 524288), (914632, 949168)] active=[] tensor_type=TensorType { shape: TensorShape([1024, 3072]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 1344, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 7, padding_groups: 1, block_size: 32, padding_multiple: 32, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(1), partitions: 192, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.831098Z DEBUG tile allocation does not fit tile=184 class=Ipu21Standard bytes=61440 alignment=8 first=0 last=4294967295 free=[(359048, 360448), (367464, 376832), (507904, 524288), (914632, 949168)] active=[] tensor_type=TensorType { shape: TensorShape([1024, 3072]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 1344, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 7, padding_groups: 1, block_size: 32, padding_multiple: 32, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(1), partitions: 192, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.831037Z DEBUG tile allocation does not fit tile=368 class=Ipu21Interleaved bytes=51200 alignment=32 first=26 last=27 free=[(337920, 360448), (376680, 376832), (510760, 524288), (778240, 847872), (899080, 949168)] active=[(4294967295, 524288, 73728), (4294967295, 598016, 73728), (4294967295, 671744, 49152), (4294967295, 720896, 24576), (4294967295, 367464, 3072), (4294967295, 370536, 3072), (4294967295, 373608, 1536), (4294967295, 375144, 1024), (4294967295, 507904, 1024), (4294967295, 508928, 1024), (4294967295, 509952, 808), (4294967295, 376168, 512), (160, 327680, 10240), (26, 847872, 51208), (26, 745472, 32768)] tensor_type=TensorType { shape: TensorShape([4, 5, 577, 3072]), format: TensorFormat { precision: F16, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1440, replicas: 1, axes: [AxisTiling { axis: FromStart(3), partitions: 24, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromStart(2), partitions: 15, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(96) }, AxisTiling { axis: FromStart(0), partitions: 4, padding_groups: 1, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(24) }] }, memory_class: Ipu21Interleaved } } } 2026-09-12T08:06:00.831272Z DEBUG tile allocation does not fit tile=126 class=Ipu21Standard bytes=80968 alignment=32 first=129 last=131 free=[(337920, 360448), (376760, 376832), (518144, 519664), (522440, 524288), (757760, 768000), (929920, 949168)] active=[(4294967295, 524288, 73728), (4294967295, 598016, 73728), (4294967295, 671744, 61440), (4294967295, 733184, 24576), (4294967295, 367464, 3072), (4294967295, 370536, 3072), (4294967295, 373608, 1536), (4294967295, 375144, 1024), (4294967295, 519664, 1024), (4294967295, 520688, 1024), (4294967295, 376168, 592), (4294967295, 521712, 512), (4294967295, 522224, 216), (162, 327680, 10240), (155, 507904, 10240), (129, 768000, 161920)] tensor_type=TensorType { shape: TensorShape([5, 577, 1024]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1443, replicas: 37, axes: [AxisTiling { axis: FromEnd(1), partitions: 3, padding_groups: 1, block_size: 352, padding_multiple: 352, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(2), partitions: 13, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.831330Z DEBUG tile allocation does not fit tile=0 class=Ipu21Standard bytes=61440 alignment=8 first=0 last=4294967295 free=[(359048, 360448), (367464, 376832), (507904, 524288), (914632, 949168)] active=[] tensor_type=TensorType { shape: TensorShape([1024, 3072]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 1344, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 7, padding_groups: 1, block_size: 32, padding_multiple: 32, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(1), partitions: 192, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.831333Z DEBUG tile allocation does not fit tile=46 class=Ipu21Interleaved bytes=51200 alignment=32 first=26 last=27 free=[(337920, 360448), (376680, 376832), (507904, 519664), (522520, 524288), (778240, 847872), (899080, 949168)] active=[(4294967295, 524288, 73728), (4294967295, 598016, 73728), (4294967295, 671744, 49152), (4294967295, 720896, 24576), (4294967295, 367464, 3072), (4294967295, 370536, 3072), (4294967295, 373608, 1536), (4294967295, 375144, 1024), (4294967295, 519664, 1024), (4294967295, 520688, 1024), (4294967295, 521712, 808), (4294967295, 376168, 512), (163, 327680, 10240), (26, 847872, 51208), (26, 745472, 32768)] tensor_type=TensorType { shape: TensorShape([4, 5, 577, 3072]), format: TensorFormat { precision: F16, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1440, replicas: 1, axes: [AxisTiling { axis: FromStart(3), partitions: 24, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromStart(2), partitions: 15, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(96) }, AxisTiling { axis: FromStart(0), partitions: 4, padding_groups: 1, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(24) }] }, memory_class: Ipu21Interleaved } } } 2026-09-12T08:06:00.831439Z DEBUG tile allocation does not fit tile=368 class=Ipu21Standard bytes=49152 alignment=8 first=0 last=4294967295 free=[(356808, 360448), (367464, 376832), (507904, 524288), (904072, 949168)] active=[] tensor_type=TensorType { shape: TensorShape([1024, 3072]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 1344, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 7, padding_groups: 1, block_size: 32, padding_multiple: 32, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(1), partitions: 192, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.831349Z DEBUG tile allocation does not fit tile=138 class=Ipu21Interleaved bytes=51200 alignment=32 first=26 last=27 free=[(337920, 360448), (376680, 376832), (507904, 519664), (522520, 524288), (778240, 847872), (899080, 949168)] active=[(4294967295, 524288, 73728), (4294967295, 598016, 73728), (4294967295, 671744, 49152), (4294967295, 720896, 24576), (4294967295, 367464, 3072), (4294967295, 370536, 3072), (4294967295, 373608, 1536), (4294967295, 375144, 1024), (4294967295, 519664, 1024), (4294967295, 520688, 1024), (4294967295, 521712, 808), (4294967295, 376168, 512), (160, 327680, 10240), (26, 847872, 51208), (26, 745472, 32768)] tensor_type=TensorType { shape: TensorShape([4, 5, 577, 3072]), format: TensorFormat { precision: F16, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1440, replicas: 1, axes: [AxisTiling { axis: FromStart(3), partitions: 24, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromStart(2), partitions: 15, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(96) }, AxisTiling { axis: FromStart(0), partitions: 4, padding_groups: 1, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(24) }] }, memory_class: Ipu21Interleaved } } } 2026-09-12T08:06:00.831440Z DEBUG tile allocation does not fit tile=552 class=Ipu21Interleaved bytes=48640 alignment=32 first=26 last=27 free=[(337920, 360448), (376712, 376832), (510720, 524288), (778240, 842752), (891400, 949168)] active=[(4294967295, 524288, 73728), (4294967295, 598016, 73728), (4294967295, 671744, 49152), (4294967295, 720896, 24576), (4294967295, 367464, 3072), (4294967295, 370536, 3072), (4294967295, 373608, 1536), (4294967295, 375144, 1024), (4294967295, 507904, 1024), (4294967295, 508928, 1024), (4294967295, 509952, 768), (4294967295, 376168, 512), (4294967295, 376680, 32), (163, 327680, 10240), (26, 842752, 48648), (26, 745472, 32768)] tensor_type=TensorType { shape: TensorShape([4, 5, 577, 3072]), format: TensorFormat { precision: F16, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1440, replicas: 1, axes: [AxisTiling { axis: FromStart(3), partitions: 24, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromStart(2), partitions: 15, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(96) }, AxisTiling { axis: FromStart(0), partitions: 4, padding_groups: 1, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(24) }] }, memory_class: Ipu21Interleaved } } } 2026-09-12T08:06:00.831607Z DEBUG tile allocation does not fit tile=115 class=Ipu21Standard bytes=80968 alignment=32 first=129 last=131 free=[(337920, 360448), (376680, 376832), (518144, 519664), (522520, 524288), (757760, 768000), (929920, 949168)] active=[(4294967295, 524288, 73728), (4294967295, 598016, 73728), (4294967295, 671744, 61440), (4294967295, 733184, 24576), (4294967295, 367464, 3072), (4294967295, 370536, 3072), (4294967295, 373608, 1536), (4294967295, 375144, 1024), (4294967295, 519664, 1024), (4294967295, 520688, 1024), (4294967295, 521712, 808), (4294967295, 376168, 512), (170, 327680, 10240), (163, 507904, 10240), (129, 768000, 161920)] tensor_type=TensorType { shape: TensorShape([5, 577, 1024]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1443, replicas: 37, axes: [AxisTiling { axis: FromEnd(1), partitions: 3, padding_groups: 1, block_size: 352, padding_multiple: 352, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(2), partitions: 13, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.831601Z DEBUG tile allocation does not fit tile=161 class=Ipu21Standard bytes=80968 alignment=32 first=129 last=131 free=[(337920, 360448), (376680, 376832), (518144, 519664), (522520, 524288), (757760, 768000), (929920, 949168)] active=[(4294967295, 524288, 73728), (4294967295, 598016, 73728), (4294967295, 671744, 61440), (4294967295, 733184, 24576), (4294967295, 367464, 3072), (4294967295, 370536, 3072), (4294967295, 373608, 1536), (4294967295, 375144, 1024), (4294967295, 519664, 1024), (4294967295, 520688, 1024), (4294967295, 521712, 808), (4294967295, 376168, 512), (162, 327680, 10240), (155, 507904, 10240), (129, 768000, 161920)] tensor_type=TensorType { shape: TensorShape([5, 577, 1024]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1443, replicas: 37, axes: [AxisTiling { axis: FromEnd(1), partitions: 3, padding_groups: 1, block_size: 352, padding_multiple: 352, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(2), partitions: 13, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.831690Z DEBUG tile allocation does not fit tile=126 class=Ipu21Standard bytes=61440 alignment=8 first=0 last=4294967295 free=[(359048, 360448), (367464, 376832), (507904, 524288), (914632, 949168)] active=[] tensor_type=TensorType { shape: TensorShape([1024, 3072]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 1344, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 7, padding_groups: 1, block_size: 32, padding_multiple: 32, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(1), partitions: 192, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.831745Z DEBUG tile allocation does not fit tile=46 class=Ipu21Standard bytes=49152 alignment=8 first=0 last=4294967295 free=[(359048, 360448), (367464, 376832), (507904, 524288), (914632, 949168)] active=[] tensor_type=TensorType { shape: TensorShape([1024, 3072]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 1344, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 7, padding_groups: 1, block_size: 32, padding_multiple: 32, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(1), partitions: 192, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.831784Z DEBUG tile allocation does not fit tile=552 class=Ipu21Standard bytes=49152 alignment=8 first=0 last=4294967295 free=[(356808, 360448), (367464, 376832), (507904, 524288), (904072, 949168)] active=[] tensor_type=TensorType { shape: TensorShape([1024, 3072]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 1344, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 7, padding_groups: 1, block_size: 32, padding_multiple: 32, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(1), partitions: 192, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.831786Z DEBUG tile allocation does not fit tile=138 class=Ipu21Standard bytes=49152 alignment=8 first=0 last=4294967295 free=[(359048, 360448), (367464, 376832), (507904, 524288), (914632, 949168)] active=[] tensor_type=TensorType { shape: TensorShape([1024, 3072]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 1344, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 7, padding_groups: 1, block_size: 32, padding_multiple: 32, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(1), partitions: 192, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.832011Z DEBUG tile allocation does not fit tile=161 class=Ipu21Standard bytes=61440 alignment=8 first=0 last=4294967295 free=[(359048, 360448), (367464, 376832), (507904, 524288), (914632, 949168)] active=[] tensor_type=TensorType { shape: TensorShape([1024, 3072]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 1344, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 7, padding_groups: 1, block_size: 32, padding_multiple: 32, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(1), partitions: 192, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.832021Z DEBUG tile allocation does not fit tile=115 class=Ipu21Standard bytes=61440 alignment=8 first=0 last=4294967295 free=[(359048, 360448), (367464, 376832), (507904, 524288), (914632, 949168)] active=[] tensor_type=TensorType { shape: TensorShape([1024, 3072]), format: TensorFormat { precision: F8F143 { scale_exponent: -4 }, layout: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 1344, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 7, padding_groups: 1, block_size: 32, padding_multiple: 32, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(1), partitions: 192, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } } } 2026-09-12T08:06:00.900233Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}:ipu_codegen.package.phase{phase="place_storage"}: package build phase finished phase="place_storage" elapsed_ms=584 success=false 2026-09-12T08:06:02.272549Z INFO ipu_codegen.package.build{tile_count=1472 operations=26}:ipu_codegen.package.phase{phase="plan_package"}: package build phase finished phase="plan_package" elapsed_ms=95894 success=false Error: placement failed: tile 736 has insufficient Ipu21Standard SRAM for 61440 bytes Caused by: tile 736 has insufficient Ipu21Standard SRAM for 61440 bytes