Compiling ipu-codegen v0.1.0 (/srv/home/gc-sdk/ipu-stack/crates/ipu-codegen)
   Compiling ipu-tests v0.1.0 (/srv/home/gc-sdk/ipu-stack/crates/ipu-tests)
    Finished `release` profile [optimized] target(s) in 20.30s
     Running `target/release/ipu-trivial-test c600-init.ipucfg --workload siglip-mlp-benchmark --mlp-batch 1 --mlp-tokens 729 --mlp-dim 1152 --mlp-hidden-dim 4304 --mlp-blocks 1 --tiles 1472 --profiling none --timeout-seconds 30`
2026-08-17T13:11:57.641297Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=2279 proxy_frontier_grids=239 generated_variants=2832 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:11:57.768685Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=570 proxy_frontier_grids=111 generated_variants=2624 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:11:59.607495Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=1661 proxy_frontier_grids=180 generated_variants=2128 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:11:59.684428Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=464 proxy_frontier_grids=88 generated_variants=2080 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:00.273832Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=641 proxy_frontier_grids=106 generated_variants=1228 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:00.303256Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=168 proxy_frontier_grids=54 generated_variants=1248 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:00.401413Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=126 proxy_frontier_grids=27 generated_variants=324 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=59
2026-08-17T13:12:00.408550Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=33 proxy_frontier_grids=17 generated_variants=408 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:00.421065Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.421091Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.429195Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.429244Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.436659Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.436668Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.443164Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.443172Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.448861Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.448868Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.453282Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.453289Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.456433Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.456439Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.458339Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.458345Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.458924Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:00.458929Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:03.432373Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=2266 proxy_frontier_grids=207 generated_variants=2452 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:03.560878Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=570 proxy_frontier_grids=114 generated_variants=2680 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:06.260119Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=1876 proxy_frontier_grids=227 generated_variants=2688 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:06.349556Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=517 proxy_frontier_grids=94 generated_variants=2216 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:06.651567Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained planning beam operation=0 retained=64 expanded=1890 pareto_dominated=1123 equivalent=124 diversity_representatives=6 best_cycles=59718 generated_plan_sets=1 plan_cache_hits=0
2026-08-17T13:12:10.463010Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained planning beam operation=1 retained=64 expanded=254 pareto_dominated=35 equivalent=24 diversity_representatives=11 best_cycles=81489 generated_plan_sets=29 plan_cache_hits=35
2026-08-17T13:12:13.750508Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=1693 proxy_frontier_grids=234 generated_variants=2716 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:14.004652Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=1171 proxy_frontier_grids=226 generated_variants=5168 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:16.039820Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=1352 proxy_frontier_grids=197 generated_variants=2320 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:16.219093Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=964 proxy_frontier_grids=202 generated_variants=4736 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:17.033092Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=575 proxy_frontier_grids=145 generated_variants=1704 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:17.115857Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=384 proxy_frontier_grids=144 generated_variants=3400 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:17.381627Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=141 proxy_frontier_grids=62 generated_variants=732 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:17.404943Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=93 proxy_frontier_grids=54 generated_variants=1296 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:17.417467Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.417500Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.425080Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.425095Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.432097Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.432108Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.438183Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.438191Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.443584Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.443607Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.447838Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.447845Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.450992Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.450998Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.452865Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.452870Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.453442Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:17.453447Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=0 proxy_frontier_grids=0 generated_variants=0 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=0
2026-08-17T13:12:20.564428Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=1688 proxy_frontier_grids=220 generated_variants=2572 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:20.815849Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=1168 proxy_frontier_grids=225 generated_variants=5152 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:23.482348Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Normal generated_grids=1497 proxy_frontier_grids=229 generated_variants=2656 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:23.682037Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained parallel GEMM candidates orientation=Swapped generated_grids=1053 proxy_frontier_grids=207 generated_variants=4872 generated_grouped_variants=0 retained_grouped_variants=0 retained_variants=64
2026-08-17T13:12:31.081408Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained planning beam operation=2 retained=64 expanded=4325 pareto_dominated=2343 equivalent=492 diversity_representatives=9 best_cycles=234069 generated_plan_sets=1 plan_cache_hits=63
2026-08-17T13:12:31.083846Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained operator-plan finalist finalist=0 values=8 operations=5 estimated_cycles=234069 estimated_exchange_cycles=63408 exchange_row_bytes=62856 peak_standard_bytes=271496 peak_interleaved_bytes=115520 peak_total_bytes=387016 maximum_standard_allocation_bytes=180224 active_tile_counts={1350, 1440, 1472} padding_group_counts={} conversion_sources=[Some(OperationId(1)), Some(OperationId(2))]
2026-08-17T13:12:31.083874Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: retained operator-plan details finalist=0 plans=[(Some(OperationId(0)), BlockedGemm(BlockedGemmPlan { kernel: GemmKernelFamily { multiply: F16, accumulate: F16, weights: Interleaved }, geometry: GemmGeometry { block: GemmBlockShape { inner: 240, output_columns: 32 }, orientation: Normal, result: GemmResultGrid { rows: 10, columns: 135 }, order: ColumnsFast, distribution: ParallelReduction(ParallelReductionPlan { compute: GemmGrid { rows: 2, columns: 135, inner: 5 }, staging: Complete }) } }), [Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1350, replicas: 135, axes: [AxisTiling { axis: FromEnd(1), partitions: 5, padding_groups: 1, block_size: 240, padding_multiple: 240, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(2), partitions: 2, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Standard }, Layout { order: BlockMajor(Matrix { row_block: 240, column_block: 16 }), tiling: TensorTiling { tile_count: 675, replicas: 1, axes: [AxisTiling { axis: FromEnd(1), partitions: 135, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 5, padding_groups: 1, block_size: 240, padding_multiple: 240, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Interleaved }], Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1350, replicas: 1, axes: [AxisTiling { axis: FromEnd(1), partitions: 135, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 10, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(135) }] }, memory_class: Interleaved }, 89252, 13616), (Some(OperationId(1)), Pointwise { kernel: Gelu, input_mapping: TileLocal }, [Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: Linear, partitions: 1472, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Reject, tile_stride: None }] }, memory_class: Interleaved }], Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: Linear, partitions: 1472, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Reject, tile_stride: None }] }, memory_class: Interleaved }, 21451, 0), (Some(OperationId(2)), BlockedGemm(BlockedGemmPlan { kernel: GemmKernelFamily { multiply: F16, accumulate: F16, weights: Interleaved }, geometry: GemmGeometry { block: GemmBlockShape { inner: 288, output_columns: 48 }, orientation: Normal, result: GemmResultGrid { rows: 60, columns: 24 }, order: ColumnsFast, distribution: ParallelReduction(ParallelReductionPlan { compute: GemmGrid { rows: 4, columns: 24, inner: 15 }, staging: Complete }) } }), [Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1440, replicas: 24, axes: [AxisTiling { axis: FromEnd(1), partitions: 15, padding_groups: 1, block_size: 288, padding_multiple: 288, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(2), partitions: 4, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Standard }, Layout { order: BlockMajor(Matrix { row_block: 288, column_block: 16 }), tiling: TensorTiling { tile_count: 360, replicas: 1, axes: [AxisTiling { axis: FromEnd(1), partitions: 24, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 15, padding_groups: 1, block_size: 288, padding_multiple: 288, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Interleaved }], Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1440, replicas: 1, axes: [AxisTiling { axis: FromEnd(1), partitions: 24, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 60, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(24) }] }, memory_class: Interleaved }, 92646, 19728)] conversions=[(Some(OperationId(1)), Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1350, replicas: 1, axes: [AxisTiling { axis: FromEnd(1), partitions: 135, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 10, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(135) }] }, memory_class: Interleaved }, Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: Linear, partitions: 1472, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Reject, tile_stride: None }] }, memory_class: Interleaved }, 3264, 2968), (Some(OperationId(2)), Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: Linear, partitions: 1472, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Reject, tile_stride: None }] }, memory_class: Interleaved }, Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1440, replicas: 24, axes: [AxisTiling { axis: FromEnd(1), partitions: 15, padding_groups: 1, block_size: 288, padding_multiple: 288, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(2), partitions: 4, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Standard }, 27456, 27096)]
2026-08-17T13:12:31.087433Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_mid"}: package build phase finished phase="lower_mid" elapsed_ms=36883 success=true
2026-08-17T13:12:31.087465Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}: selected analytical operator plan estimated_cycles=234069 estimated_exchange_cycles=63408
2026-08-17T13:12:31.087469Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}: selected mid-level operation source=Some(OperationId(0)) kind=Operator(OperatorPlan { operator: Gemm { options: GemmOptions { transpose_left: false, transpose_right: false }, multiply: F16, accumulate: F16 }, dispatch: BlockedGemm(BlockedGemmPlan { kernel: GemmKernelFamily { multiply: F16, accumulate: F16, weights: Interleaved }, geometry: GemmGeometry { block: GemmBlockShape { inner: 240, output_columns: 32 }, orientation: Normal, result: GemmResultGrid { rows: 10, columns: 135 }, order: ColumnsFast, distribution: ParallelReduction(ParallelReductionPlan { compute: GemmGrid { rows: 2, columns: 135, inner: 5 }, staging: Complete }) } }), requirements: OperatorRequirements { inputs: [OperandRequirement { format: TensorFormat { precision: F16, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1350, replicas: 135, axes: [AxisTiling { axis: FromEnd(1), partitions: 5, padding_groups: 1, block_size: 240, padding_multiple: 240, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(2), partitions: 2, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Standard } }, allocation: AllocationRequirements { alignment: 32, access_tail_bytes: 16, memory_element: Any }, local_staging: Direct, materialization: Complete }, OperandRequirement { format: TensorFormat { precision: F16, layout: Layout { order: BlockMajor(Matrix { row_block: 240, column_block: 16 }), tiling: TensorTiling { tile_count: 675, replicas: 1, axes: [AxisTiling { axis: FromEnd(1), partitions: 135, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 5, padding_groups: 1, block_size: 240, padding_multiple: 240, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Interleaved } }, allocation: AllocationRequirements { alignment: 32, access_tail_bytes: 0, memory_element: Any }, local_staging: Direct, materialization: Complete }], output: OperandRequirement { format: TensorFormat { precision: F16, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1350, replicas: 1, axes: [AxisTiling { axis: FromEnd(1), partitions: 135, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 10, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(135) }] }, memory_class: Interleaved } }, allocation: AllocationRequirements { alignment: 32, access_tail_bytes: 0, memory_element: Any }, local_staging: Direct, materialization: Complete }, output_aliasing: Fresh, memory_space: MemorySpaceRequirements { distinct_element_groups: [[Output, Input(0)]] } }, deferred_output: None, deferred_inputs: [None, None] }) memory=MemoryEstimate { live: MemoryUsage { standard: 175680, interleaved: 20096 }, temporary: MemoryUsage { standard: 28416, interleaved: 39744 }, peak: MemoryUsage { standard: 204096, interleaved: 59840 }, maximum_standard_temporary_allocation: 0 } plan=Some(OperatorPlan { operator: Gemm { options: GemmOptions { transpose_left: false, transpose_right: false }, multiply: F16, accumulate: F16 }, dispatch: BlockedGemm(BlockedGemmPlan { kernel: GemmKernelFamily { multiply: F16, accumulate: F16, weights: Interleaved }, geometry: GemmGeometry { block: GemmBlockShape { inner: 240, output_columns: 32 }, orientation: Normal, result: GemmResultGrid { rows: 10, columns: 135 }, order: ColumnsFast, distribution: ParallelReduction(ParallelReductionPlan { compute: GemmGrid { rows: 2, columns: 135, inner: 5 }, staging: Complete }) } }), requirements: OperatorRequirements { inputs: [OperandRequirement { format: TensorFormat { precision: F16, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1350, replicas: 135, axes: [AxisTiling { axis: FromEnd(1), partitions: 5, padding_groups: 1, block_size: 240, padding_multiple: 240, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(2), partitions: 2, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Standard } }, allocation: AllocationRequirements { alignment: 32, access_tail_bytes: 16, memory_element: Any }, local_staging: Direct, materialization: Complete }, OperandRequirement { format: TensorFormat { precision: F16, layout: Layout { order: BlockMajor(Matrix { row_block: 240, column_block: 16 }), tiling: TensorTiling { tile_count: 675, replicas: 1, axes: [AxisTiling { axis: FromEnd(1), partitions: 135, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 5, padding_groups: 1, block_size: 240, padding_multiple: 240, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Interleaved } }, allocation: AllocationRequirements { alignment: 32, access_tail_bytes: 0, memory_element: Any }, local_staging: Direct, materialization: Complete }], output: OperandRequirement { format: TensorFormat { precision: F16, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1350, replicas: 1, axes: [AxisTiling { axis: FromEnd(1), partitions: 135, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 10, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(135) }] }, memory_class: Interleaved } }, allocation: AllocationRequirements { alignment: 32, access_tail_bytes: 0, memory_element: Any }, local_staging: Direct, materialization: Complete }, output_aliasing: Fresh, memory_space: MemorySpaceRequirements { distinct_element_groups: [[Output, Input(0)]] } }, deferred_output: None, deferred_inputs: [None, None] })
2026-08-17T13:12:31.087512Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}: selected mid-level operation source=Some(OperationId(1)) kind=Rearrange { from: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1350, replicas: 1, axes: [AxisTiling { axis: FromEnd(1), partitions: 135, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 10, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(135) }] }, memory_class: Interleaved }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: Linear, partitions: 1472, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Reject, tile_stride: None }] }, memory_class: Interleaved } } memory=MemoryEstimate { live: MemoryUsage { standard: 0, interleaved: 9024 }, temporary: MemoryUsage { standard: 0, interleaved: 0 }, peak: MemoryUsage { standard: 0, interleaved: 9024 }, maximum_standard_temporary_allocation: 0 } plan=None
2026-08-17T13:12:31.087521Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}: selected mid-level operation source=Some(OperationId(1)) kind=Operator(OperatorPlan { operator: Gelu, dispatch: Pointwise { kernel: Gelu, input_mapping: TileLocal }, requirements: OperatorRequirements { inputs: [OperandRequirement { format: TensorFormat { precision: F16, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: Linear, partitions: 1472, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Reject, tile_stride: None }] }, memory_class: Interleaved } }, allocation: AllocationRequirements { alignment: 8, access_tail_bytes: 0, memory_element: Any }, local_staging: Direct, materialization: Complete }], output: OperandRequirement { format: TensorFormat { precision: F16, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: Linear, partitions: 1472, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Reject, tile_stride: None }] }, memory_class: Interleaved } }, allocation: AllocationRequirements { alignment: 8, access_tail_bytes: 0, memory_element: Any }, local_staging: Direct, materialization: Complete }, output_aliasing: MayAliasInputs([0]), memory_space: MemorySpaceRequirements { distinct_element_groups: [] } }, deferred_output: None, deferred_inputs: [None] }) memory=MemoryEstimate { live: MemoryUsage { standard: 0, interleaved: 8576 }, temporary: MemoryUsage { standard: 0, interleaved: 0 }, peak: MemoryUsage { standard: 0, interleaved: 8576 }, maximum_standard_temporary_allocation: 0 } plan=Some(OperatorPlan { operator: Gelu, dispatch: Pointwise { kernel: Gelu, input_mapping: TileLocal }, requirements: OperatorRequirements { inputs: [OperandRequirement { format: TensorFormat { precision: F16, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: Linear, partitions: 1472, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Reject, tile_stride: None }] }, memory_class: Interleaved } }, allocation: AllocationRequirements { alignment: 8, access_tail_bytes: 0, memory_element: Any }, local_staging: Direct, materialization: Complete }], output: OperandRequirement { format: TensorFormat { precision: F16, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: Linear, partitions: 1472, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Reject, tile_stride: None }] }, memory_class: Interleaved } }, allocation: AllocationRequirements { alignment: 8, access_tail_bytes: 0, memory_element: Any }, local_staging: Direct, materialization: Complete }, output_aliasing: MayAliasInputs([0]), memory_space: MemorySpaceRequirements { distinct_element_groups: [] } }, deferred_output: None, deferred_inputs: [None] })
2026-08-17T13:12:31.087535Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}: selected mid-level operation source=Some(OperationId(2)) kind=Rearrange { from: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: Linear, partitions: 1472, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Reject, tile_stride: None }] }, memory_class: Interleaved }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1440, replicas: 24, axes: [AxisTiling { axis: FromEnd(1), partitions: 15, padding_groups: 1, block_size: 288, padding_multiple: 288, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(2), partitions: 4, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Standard } } memory=MemoryEstimate { live: MemoryUsage { standard: 105984, interleaved: 4288 }, temporary: MemoryUsage { standard: 0, interleaved: 0 }, peak: MemoryUsage { standard: 105984, interleaved: 4288 }, maximum_standard_temporary_allocation: 0 } plan=None
2026-08-17T13:12:31.087542Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}: selected mid-level operation source=Some(OperationId(2)) kind=Operator(OperatorPlan { operator: Gemm { options: GemmOptions { transpose_left: false, transpose_right: false }, multiply: F16, accumulate: F16 }, dispatch: BlockedGemm(BlockedGemmPlan { kernel: GemmKernelFamily { multiply: F16, accumulate: F16, weights: Interleaved }, geometry: GemmGeometry { block: GemmBlockShape { inner: 288, output_columns: 48 }, orientation: Normal, result: GemmResultGrid { rows: 60, columns: 24 }, order: ColumnsFast, distribution: ParallelReduction(ParallelReductionPlan { compute: GemmGrid { rows: 4, columns: 24, inner: 15 }, staging: Complete }) } }), requirements: OperatorRequirements { inputs: [OperandRequirement { format: TensorFormat { precision: F16, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1440, replicas: 24, axes: [AxisTiling { axis: FromEnd(1), partitions: 15, padding_groups: 1, block_size: 288, padding_multiple: 288, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(2), partitions: 4, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Standard } }, allocation: AllocationRequirements { alignment: 32, access_tail_bytes: 16, memory_element: Any }, local_staging: Direct, materialization: Complete }, OperandRequirement { format: TensorFormat { precision: F16, layout: Layout { order: BlockMajor(Matrix { row_block: 288, column_block: 16 }), tiling: TensorTiling { tile_count: 360, replicas: 1, axes: [AxisTiling { axis: FromEnd(1), partitions: 24, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 15, padding_groups: 1, block_size: 288, padding_multiple: 288, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Interleaved } }, allocation: AllocationRequirements { alignment: 32, access_tail_bytes: 0, memory_element: Any }, local_staging: Direct, materialization: Complete }], output: OperandRequirement { format: TensorFormat { precision: F16, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1440, replicas: 1, axes: [AxisTiling { axis: FromEnd(1), partitions: 24, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 60, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(24) }] }, memory_class: Interleaved } }, allocation: AllocationRequirements { alignment: 32, access_tail_bytes: 0, memory_element: Any }, local_staging: Direct, materialization: Complete }, output_aliasing: Fresh, memory_space: MemorySpaceRequirements { distinct_element_groups: [[Output, Input(0)]] } }, deferred_output: None, deferred_inputs: [None, None] }) memory=MemoryEstimate { live: MemoryUsage { standard: 105984, interleaved: 28992 }, temporary: MemoryUsage { standard: 21504, interleaved: 48288 }, peak: MemoryUsage { standard: 127488, interleaved: 77280 }, maximum_standard_temporary_allocation: 0 } plan=Some(OperatorPlan { operator: Gemm { options: GemmOptions { transpose_left: false, transpose_right: false }, multiply: F16, accumulate: F16 }, dispatch: BlockedGemm(BlockedGemmPlan { kernel: GemmKernelFamily { multiply: F16, accumulate: F16, weights: Interleaved }, geometry: GemmGeometry { block: GemmBlockShape { inner: 288, output_columns: 48 }, orientation: Normal, result: GemmResultGrid { rows: 60, columns: 24 }, order: ColumnsFast, distribution: ParallelReduction(ParallelReductionPlan { compute: GemmGrid { rows: 4, columns: 24, inner: 15 }, staging: Complete }) } }), requirements: OperatorRequirements { inputs: [OperandRequirement { format: TensorFormat { precision: F16, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1440, replicas: 24, axes: [AxisTiling { axis: FromEnd(1), partitions: 15, padding_groups: 1, block_size: 288, padding_multiple: 288, shard_padding_multiple: 1, padding: Zero, tile_stride: None }, AxisTiling { axis: FromEnd(2), partitions: 4, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Standard } }, allocation: AllocationRequirements { alignment: 32, access_tail_bytes: 16, memory_element: Any }, local_staging: Direct, materialization: Complete }, OperandRequirement { format: TensorFormat { precision: F16, layout: Layout { order: BlockMajor(Matrix { row_block: 288, column_block: 16 }), tiling: TensorTiling { tile_count: 360, replicas: 1, axes: [AxisTiling { axis: FromEnd(1), partitions: 24, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 15, padding_groups: 1, block_size: 288, padding_multiple: 288, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Interleaved } }, allocation: AllocationRequirements { alignment: 32, access_tail_bytes: 0, memory_element: Any }, local_staging: Direct, materialization: Complete }], output: OperandRequirement { format: TensorFormat { precision: F16, layout: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1440, replicas: 1, axes: [AxisTiling { axis: FromEnd(1), partitions: 24, padding_groups: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 60, padding_groups: 1, block_size: 2, padding_multiple: 2, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(24) }] }, memory_class: Interleaved } }, allocation: AllocationRequirements { alignment: 32, access_tail_bytes: 0, memory_element: Any }, local_staging: Direct, materialization: Complete }, output_aliasing: Fresh, memory_space: MemorySpaceRequirements { distinct_element_groups: [[Output, Input(0)]] } }, deferred_output: None, deferred_inputs: [None, None] })
2026-08-17T13:12:31.092717Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: assigned parameter storage group to tiles placement_group=MidValueId(1) offset=0 shards=675
2026-08-17T13:12:31.095779Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: assigned parameter storage group to tiles placement_group=MidValueId(2) offset=675 shards=360
2026-08-17T13:12:31.145869Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: scheduled exchange phase phase=0 operation=Some(0) value=Some(1) reason=OperatorInput { input: 1 }
2026-08-17T13:12:31.146676Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: prepared parallel GEMM partials partial_groups=270 output_shards=1350 result_row_partitions=5 result_column_partitions=1
2026-08-17T13:12:31.329066Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: scheduled exchange phase phase=1 operation=Some(0) value=Some(3) reason=OperatorInputs
2026-08-17T13:12:31.330597Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: materialized packed parallel reduction reduction_roots=1485
2026-08-17T13:12:31.330726Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: lowered mid operation to tile work operation=0 source=Some(0) elapsed_ms=232 shards=17485 exchange_phases=2
2026-08-17T13:12:31.582170Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: scheduled exchange phase phase=2 operation=Some(1) value=Some(4) reason=LayoutRearrangement
2026-08-17T13:12:31.582458Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: lowered mid operation to tile work operation=1 source=Some(1) elapsed_ms=251 shards=17485 exchange_phases=3
2026-08-17T13:12:31.589072Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: lowered mid operation to tile work operation=2 source=Some(1) elapsed_ms=6 shards=17485 exchange_phases=3
2026-08-17T13:12:33.739893Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: scheduled exchange phase phase=3 operation=Some(2) value=Some(6) reason=LayoutRearrangement
2026-08-17T13:12:33.740363Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: lowered mid operation to tile work operation=3 source=Some(2) elapsed_ms=2151 shards=17485 exchange_phases=4
2026-08-17T13:12:33.759429Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: consolidated independent exchange transfers phase=3 operation=Some(2)
2026-08-17T13:12:33.760656Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: prepared parallel GEMM partials partial_groups=96 output_shards=1440 result_row_partitions=15 result_column_partitions=1
2026-08-17T13:12:33.805597Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: scheduled exchange phase phase=4 operation=Some(2) value=Some(7) reason=OperatorInputs
2026-08-17T13:12:33.807144Z DEBUG ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: materialized packed parallel reduction reduction_roots=1512
2026-08-17T13:12:33.807495Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: lowered mid operation to tile work operation=4 source=Some(2) elapsed_ms=67 shards=24901 exchange_phases=5
2026-08-17T13:12:33.807517Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}:ipu_codegen.low.lower_to_tiles{tile_count=1472 operations=5 profiling=Disabled}: built logical tile schedule shards=24901 exchange_phases=5
2026-08-17T13:12:33.807529Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="select_finalist"}: package build phase finished phase="select_finalist" elapsed_ms=2720 success=true
2026-08-17T13:12:33.807536Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}: lowered graph for package construction logical_shards=24901 exchange_phases=5
2026-08-17T13:12:33.807682Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="compile_runtime"}: package build phase finished phase="compile_runtime" elapsed_ms=0 success=true
2026-08-17T13:12:33.809952Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="plan_kernels"}: package build phase finished phase="plan_kernels" elapsed_ms=2 success=true
2026-08-17T13:12:33.810285Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="compile_kernels"}: package build phase finished phase="compile_kernels" elapsed_ms=0 success=true
2026-08-17T13:12:33.810941Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="link_runtime"}: package build phase finished phase="link_runtime" elapsed_ms=0 success=true
2026-08-17T13:12:35.517761Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="plan_exchange_storage"}: package build phase finished phase="plan_exchange_storage" elapsed_ms=1706 success=true
2026-08-17T13:12:35.763659Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_exchanges_provisional"}: prepared large physical exchange phase phase=1 transfers=12105 maximum_identical_destinations=1
2026-08-17T13:12:35.763701Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_exchanges_provisional"}: optimized physical exchange schedule phase=1 initial_horizon=6953 selected_horizon=6657 endpoint_lower_bound=4800 lower_bound_gap=1857 selected_kind="matching-waves" neighborhood_improvements=0
2026-08-17T13:12:35.775449Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_exchanges_provisional"}: finished large physical exchange phase phase=1 tile=670 row_words=83 horizon=6657
2026-08-17T13:12:58.591851Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_exchanges_provisional"}: prepared large physical exchange phase phase=2 transfers=195361 maximum_identical_destinations=1
2026-08-17T13:12:58.591919Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_exchanges_provisional"}: optimized physical exchange schedule phase=2 initial_horizon=15536 selected_horizon=15536 endpoint_lower_bound=1184 lower_bound_gap=14352 selected_kind="full-duplex" neighborhood_improvements=0
2026-08-17T13:12:58.704916Z  INFO ipu_codegen.package.build{tile_count=1472 operations=3}:ipu_codegen.package.phase{phase="lower_exchanges_provisional"}: finished large physical exchange phase phase=2 tile=287 row_words=700 horizon=15536