Test inventory
Questi contenuti non sono ancora disponibili nella tua lingua.
This is the audit queue for Rapid Software Testing in Iron. It inventories declarations by source file rather than treating a green workspace exit as evidence that the intended corpus ran. Audit findings and focused repair PRs are tracked in issue #220.
The current tree contains 1463 Rust tests, 1881 declarative GPU test cases, and 1311 benchmark declarations across 528 files. Parameterized declarative GPU tests may expand into several dtype cases at runtime, so declaration counts are not execution counts.
Regenerate this file from the repository root:
scripts/render-test-inventory.sh > docs/test-inventory.mdThe renderer is intentionally static. During an audit, corroborate each row
with the selected Cargo/iron test lane, backend and device identity, executed
and skipped counts, and a known-bad mutation. A declaration is not considered
audited merely because it appears here.
Status vocabulary
Section titled โStatus vocabularyโqueued: declaration and execution reachability have not been audited.auditing: a focused RST session is in progress.strong: claim, production reachability, independent oracle, and a realistic known-bad mutation are demonstrated.repair: the current check is weak, false-green, redundant, or exposes a product defect; link the repairing PR in place of this state.blocked: required hardware, consumer, fixture, or observability is absent; the boundary column names that missing evidence per file.
Boundary codes are performance-baseline (no pinned threshold or same-machine
baseline), backend-device (a non-local accelerator/backend), and
external-fixture (captured production data is absent).
Declaration inventory
Section titled โDeclaration inventoryโ| File | Layer | #[test] |
#[test_kernel] |
Bench | Ignored | Audit | Boundary |
|---|---|---|---|---|---|---|---|
crates/wh-iron-cli/src/cmd/bench.rs |
unit / inline | 11 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-cli/src/cmd/diff.rs |
unit / inline | 17 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-cli/src/cmd/init.rs |
unit / inline | 4 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-cli/src/cmd/snap.rs |
unit / inline | 7 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-cli/src/config.rs |
unit / inline | 12 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-cli/src/git.rs |
unit / inline | 2 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/backend.rs |
unit / inline | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/cuda/mod.rs |
unit / inline | 26 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/emit.rs |
unit / inline | 31 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/hip/mod.rs |
unit / inline | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/ml_local.rs |
unit / inline | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/msl/config.rs |
unit / inline | 2 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/msl/mod.rs |
unit / inline | 40 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/algebraic_simplify.rs |
unit / inline | 11 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/const_fold.rs |
unit / inline | 12 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/copy_prop.rs |
unit / inline | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/cse.rs |
unit / inline | 7 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/dead_store_elim.rs |
unit / inline | 4 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/dead_value_elim.rs |
unit / inline | 9 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/fma_fusion.rs |
unit / inline | 3 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/fusion.rs |
unit / inline | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/if_conversion.rs |
unit / inline | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/kernel_inline.rs |
unit / inline | 7 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/licm.rs |
unit / inline | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/occupancy.rs |
unit / inline | 9 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/register_estimate.rs |
unit / inline | 4 | 0 | 0 | 1 | blocked | performance-baseline |
crates/wh-iron-codegen/src/passes/remap.rs |
unit / inline | 14 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/schedule.rs |
unit / inline | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/type_check.rs |
unit / inline | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/unroll.rs |
unit / inline | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/value_sink.rs |
unit / inline | 4 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/passes/vectorize.rs |
unit / inline | 7 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/src/spirv/mod.rs |
unit / inline | 7 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-codegen/tests/msl_snapshots.rs |
crate integration | 21 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-core/src/dsl/constexpr.rs |
unit / inline | 10 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-core/src/dsl/dtype.rs |
unit / inline | 8 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-core/src/dsl/shape.rs |
unit / inline | 18 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-core/src/ir/kernel.rs |
unit / inline | 10 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-core/src/protocol.rs |
unit / inline | 10 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-macros/src/kernel/body.rs |
unit / inline | 13 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-macros/src/kernel/mod.rs |
unit / inline | 8 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-macros/src/kernel/variants.rs |
unit / inline | 84 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-macros/src/test.rs |
unit / inline | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-msl/tests/roundtrip.rs |
crate integration | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-runtime/src/batch_hazard.rs |
unit / inline | 4 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-runtime/src/buffer.rs |
unit / inline | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-runtime/src/cache/msl_cache.rs |
unit / inline | 2 | 0 | 0 | 2 | blocked | performance-baseline |
crates/wh-iron-runtime/src/cache/pso_cache.rs |
unit / inline | 14 | 0 | 0 | 1 | blocked | performance-baseline |
crates/wh-iron-runtime/src/context.rs |
unit / inline | 26 | 0 | 0 | 2 | blocked | performance-baseline |
crates/wh-iron-runtime/src/device/cuda/mxfp4_moe.rs |
unit / inline | 3 | 0 | 0 | 0 | blocked | backend-device |
crates/wh-iron-runtime/src/device/gpu_family.rs |
unit / inline | 9 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-runtime/src/device/vulkan/mod.rs |
unit / inline | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-runtime/src/dispatch/buffer_plan.rs |
unit / inline | 15 | 0 | 0 | 3 | blocked | performance-baseline |
crates/wh-iron-runtime/src/dispatch/single_dispatch.rs |
unit / inline | 1 | 0 | 0 | 1 | blocked | performance-baseline |
crates/wh-iron-runtime/src/dispatch/validate.rs |
unit / inline | 7 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-runtime/src/spirv_cache.rs |
unit / inline | 3 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-runtime/src/workgroup.rs |
unit / inline | 4 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-runtime/tests/cuda_cutlass_fp4.rs |
crate integration | 1 | 0 | 0 | 1 | blocked | backend-device |
crates/wh-iron-runtime/tests/cuda_cutlass_mxfp4.rs |
crate integration | 5 | 0 | 0 | 5 | blocked | backend-device |
crates/wh-iron-runtime/tests/cuda_external_enqueue.rs |
crate integration | 3 | 0 | 0 | 3 | blocked | backend-device |
crates/wh-iron-runtime/tests/cuda_smoke.rs |
crate integration | 6 | 0 | 0 | 6 | blocked | backend-device |
crates/wh-iron-runtime/tests/hip_smoke.rs |
crate integration | 4 | 0 | 0 | 3 | blocked | backend-device |
crates/wh-iron-runtime/tests/vulkan_smoke.rs |
crate integration | 5 | 0 | 0 | 3 | blocked | backend-device |
crates/wh-iron-std/src/kernels/audio/lstm.rs |
kernel declaration | 0 | 6 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/audio/mel_spectrogram.rs |
kernel declaration | 0 | 3 | 4 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/audio/snake1d.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/audio/upsample_nearest1d.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/audio/vocoder.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/consolidated/conv1d.rs |
kernel declaration | 0 | 2 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/conv1d_causal.rs |
kernel declaration | 0 | 6 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/conv1d_causal_many.rs |
kernel declaration | 1 | 6 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/conv1d_causal_step_silu_cast_many.rs |
kernel declaration | 1 | 7 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/conv2d.rs |
kernel declaration | 0 | 5 | 4 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/conv2d_block_scaled.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/conv2d_mma.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/conv2d_mma_block_scaled.rs |
kernel declaration | 0 | 30 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/conv3d.rs |
kernel declaration | 0 | 4 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/conv3d_block_scaled.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/conv3d_mma.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/conv3d_mma_block_scaled.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/depthwise_conv1d.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/depthwise_conv2d.rs |
kernel declaration | 0 | 4 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/depthwise_conv2d_block_scaled.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/depthwise_conv2d_nhwc.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/convolution/grouped_dynamic_conv.rs |
kernel declaration | 0 | 2 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/convolution/winograd_conv.rs |
kernel declaration | 0 | 4 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/dsv4_decode_state.rs |
kernel declaration | 2 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/gemm/batched_4_block_scaled_qgemv.rs |
kernel declaration | 0 | 30 | 30 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/batched_4_block_scaled_qmm.rs |
kernel declaration | 0 | 30 | 30 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/batched_4_qgemv.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/batched_4_qmm.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/batched_qkv_block_scaled_qgemv.rs |
kernel declaration | 0 | 30 | 30 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/batched_qkv_block_scaled_qmm.rs |
kernel declaration | 0 | 30 | 30 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/batched_qkv_qgemv.rs |
kernel declaration | 0 | 2 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/batched_qkv_qmm.rs |
kernel declaration | 1 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/block_scaled_matmul.rs |
kernel declaration | 0 | 30 | 30 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/block_scaled_mma.rs |
kernel declaration | 0 | 30 | 30 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/block_scaled_qmm.rs |
kernel declaration | 0 | 30 | 30 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/block_scaled_qmm_mpp.rs |
kernel declaration | 0 | 30 | 29 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/block_scaled_qmm_nax.rs |
kernel declaration | 0 | 30 | 29 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/dense.rs |
kernel declaration | 0 | 4 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/dequant_gemv.rs |
kernel declaration | 0 | 6 | 4 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/fp_quantized.rs |
kernel declaration | 0 | 3 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/fp_quantized_mma.rs |
kernel declaration | 0 | 4 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/fp_quantized_nax.rs |
kernel declaration | 3 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/gemm_q4_bm8.rs |
kernel declaration | 1 | 3 | 4 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/gemm_q4_bm8_narrow.rs |
kernel declaration | 1 | 6 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/gemm/gemm_q4_mpp.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/gemm_q8.rs |
kernel declaration | 0 | 0 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/gemm_q8_mpp.rs |
kernel declaration | 0 | 0 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/gemv.rs |
kernel declaration | 0 | 7 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/gemv_axpy_inplace.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/gemv_f16_exact.rs |
kernel declaration | 6 | 20 | 4 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/gemv_f16_exact_multi.rs |
kernel declaration | 3 | 6 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/gemv_masked.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/gemv_q8_0_exact.rs |
kernel declaration | 10 | 16 | 6 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/gemv_q8_0_multi.rs |
kernel declaration | 4 | 8 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/gemv_q8_0_shexp.rs |
kernel declaration | 2 | 6 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/gemv_quantized.rs |
kernel declaration | 3 | 14 | 29 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/gguf_k_gemv.rs |
kernel declaration | 0 | 5 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/gemm/patch_embed.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/patch_embed_block_scaled.rs |
kernel declaration | 0 | 30 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/patch_embed_mma.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/patch_embed_mma_block_scaled.rs |
kernel declaration | 0 | 30 | 29 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/q4_projections.rs |
kernel declaration | 2 | 32 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/gemm/quantized.rs |
kernel declaration | 8 | 33 | 41 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/quantized_coop.rs |
kernel declaration | 5 | 6 | 12 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/quantized_mma_dynamic_m.rs |
kernel declaration | 4 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/gemm/quantized_mpp.rs |
kernel declaration | 3 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/quantized_mpp_int8.rs |
kernel declaration | 3 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/quantized_nax.rs |
kernel declaration | 3 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/quantized_nax_int8.rs |
kernel declaration | 3 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/reduction_contract.rs |
kernel declaration | 3 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/gemm/steel/steel_gemm_fused.rs |
kernel declaration | 0 | 4 | 7 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/steel/steel_gemm_fused_nax.rs |
kernel declaration | 3 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/steel/steel_gemm_gather.rs |
kernel declaration | 0 | 2 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/steel/steel_gemm_gather_nax.rs |
kernel declaration | 3 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/steel/steel_gemm_masked.rs |
kernel declaration | 0 | 2 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/steel/steel_gemm_segmented.rs |
kernel declaration | 0 | 2 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/steel/steel_gemm_splitk.rs |
kernel declaration | 0 | 4 | 4 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/gemm/steel/steel_gemm_splitk_nax.rs |
kernel declaration | 5 | 30 | 8 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/hyper_connections/mhc.rs |
kernel declaration | 0 | 12 | 4 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/hyper_connections/mhc_head_pre.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/hyper_connections/mhc_mean.rs |
kernel declaration | 0 | 2 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/hyper_connections/mhc_sinkhorn_split.rs |
kernel declaration | 0 | 4 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/kv_cache/cache.rs |
kernel declaration | 0 | 14 | 9 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/kv_cache/compact_kv.rs |
kernel declaration | 0 | 3 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/kv_cache/fft.rs |
kernel declaration | 0 | 10 | 5 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/kv_cache/update_many.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/block_scaled_moe.rs |
kernel declaration | 0 | 30 | 30 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/dequant_gemv_expert_indexed.rs |
kernel declaration | 0 | 2 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/dequant_gemv_expert_indexed_block_scaled.rs |
kernel declaration | 0 | 30 | 30 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/hash_route.rs |
kernel declaration | 0 | 4 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_bgemm_iq2xxs_bm64.rs |
kernel declaration | 0 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_bgemm_iq2xxs_mpp.rs |
kernel declaration | 0 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_bgemm_iq2xxs_view.rs |
kernel declaration | 0 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_bgemm_iq2xxs_view_u16_bm64.rs |
kernel declaration | 0 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_bgemm_q2k_bm64.rs |
kernel declaration | 0 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_bgemm_q2k_mpp.rs |
kernel declaration | 0 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_bgemm_q2k_view.rs |
kernel declaration | 0 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_bgemm_q2k_view_u16_bm64.rs |
kernel declaration | 0 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_bgemm_q4_bm64.rs |
kernel declaration | 0 | 1 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/moe/moe_compact_active.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_densify_remap.rs |
kernel declaration | 0 | 2 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_down_swiglu_accum.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_down_weighted_sum_f16.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_gather_down_q2k.rs |
kernel declaration | 0 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_gather_down_q2k_x_q8k.rs |
kernel declaration | 3 | 7 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_gather_gemv_iq1s.rs |
kernel declaration | 0 | 2 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/moe/moe_gather_gemv_iq2xxs.rs |
kernel declaration | 1 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_gather_gemv_iq2xxs_x_q8k.rs |
kernel declaration | 3 | 5 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_gather_gemv_iq4_nl.rs |
kernel declaration | 0 | 2 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/moe/moe_gather_qmm.rs |
kernel declaration | 1 | 9 | 11 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_gather_qmm_coop.rs |
kernel declaration | 3 | 4 | 24 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_gather_qmm_coop_paired.rs |
kernel declaration | 3 | 9 | 48 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_gather_qmm_expert_mpp.rs |
kernel declaration | 2 | 9 | 6 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_gather_qmm_expert_mpp_bn64wn1.rs |
kernel declaration | 2 | 8 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_gemv_rows_iq2xxs.rs |
kernel declaration | 0 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_gemv_rows_q2k.rs |
kernel declaration | 0 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_gemv_rows_view_iq2xxs.rs |
kernel declaration | 0 | 0 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_gemv_ws_iq2xxs.rs |
kernel declaration | 0 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_gemv_ws_q2k.rs |
kernel declaration | 0 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_mpp.rs |
kernel declaration | 3 | 4 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_mpp_active.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_mpp_block_scaled.rs |
kernel declaration | 0 | 31 | 30 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_mpp_bm16_bk32_block_scaled.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_mpp_bm64.rs |
kernel declaration | 2 | 3 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_mpp_bm64_bk64_nvfp4.rs |
kernel declaration | 0 | 9 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_mpp_bm64_block_scaled.rs |
kernel declaration | 0 | 30 | 30 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_mpp_bm8.rs |
kernel declaration | 2 | 3 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_mpp_bm8_block_scaled.rs |
kernel declaration | 0 | 30 | 30 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_mpp_expert_grid.rs |
kernel declaration | 2 | 4 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_mpp_tileplan.rs |
kernel declaration | 4 | 0 | 19 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_mxfp4_expert_decode.rs |
kernel declaration | 4 | 22 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_permute.rs |
kernel declaration | 0 | 3 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_route_sort_plan_fused.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_routed_down_reduce_nvfp4.rs |
kernel declaration | 0 | 1 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_routed_qmv_nvfp4.rs |
kernel declaration | 0 | 2 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_router_group_limited_topk.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_router_sigmoid_bias.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_router_sqrtsoftplus.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_router_topk.rs |
kernel declaration | 2 | 10 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_router_topk_biased.rs |
kernel declaration | 0 | 3 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_router_topk_biased_rows_fused.rs |
kernel declaration | 0 | 4 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_sigmoid_bias.rs |
kernel declaration | 0 | 2 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/moe/moe_sigmoid_bias_rows.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_sort_plan_counting.rs |
kernel declaration | 0 | 3 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_tile_plan_builder.rs |
kernel declaration | 1 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_tile_plan_builder_bm32.rs |
kernel declaration | 1 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_tile_plan_builder_bm32_own.rs |
kernel declaration | 1 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_tile_plan_builder_bm32_own_parallel.rs |
kernel declaration | 1 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_tile_plan_builder_bm32_paired.rs |
kernel declaration | 1 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_tile_plan_builder_bm32_parallel.rs |
kernel declaration | 1 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_tile_plan_builder_paired.rs |
kernel declaration | 13 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/moe/moe_tile_plan_builder_parallel.rs |
kernel declaration | 1 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_tile_plan_expert_counts.rs |
kernel declaration | 1 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/moe/moe_unpermute_tail.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/norm/adain1d.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/norm/gated_rms_norm_block_scaled_qgemv.rs |
kernel declaration | 0 | 30 | 30 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/norm/gated_rms_norm_pregate_qgemv.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/norm/gated_rms_norm_qgemv.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/norm/gated_rmsnorm.rs |
kernel declaration | 0 | 12 | 4 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/norm/layer_norm.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/norm/rms_norm.rs |
kernel declaration | 2 | 20 | 6 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/norm/rms_norm_block_scaled_qgemv.rs |
kernel declaration | 0 | 30 | 30 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/norm/rms_norm_multi.rs |
kernel declaration | 2 | 4 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/norm/rms_norm_qgemv.rs |
kernel declaration | 0 | 3 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/norm/rms_norm_residual.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/norm/rms_norm_rope.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/norm/rms_norm_wide_acc.rs |
kernel declaration | 2 | 4 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/arange.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/arg_reduce.rs |
kernel declaration | 0 | 4 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/argmax_rows.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/axpy_scalar_inplace.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/binary.rs |
kernel declaration | 0 | 10 | 10 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/binary_two.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/clamp_scalar.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/copy.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/dp4a_u32.rs |
kernel declaration | 0 | 2 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/ops/exact_exp.rs |
kernel declaration | 4 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/ops/gate_up_swiglu_fused.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/gated_activation.rs |
kernel declaration | 1 | 9 | 6 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/gather.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/gather_axis.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/gelu_erf.rs |
kernel declaration | 0 | 3 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/gelu_erf_exact.rs |
kernel declaration | 4 | 6 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/ops/hadamard.rs |
kernel declaration | 0 | 6 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/hadamard_m.rs |
kernel declaration | 6 | 3 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/indexed_argmax.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/indexing.rs |
kernel declaration | 0 | 4 | 4 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/leaky_relu.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/logsumexp.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/masked_tail.rs |
kernel declaration | 0 | 3 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/ops/random.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/reduce.rs |
kernel declaration | 0 | 15 | 16 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/round_bf16.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/scan.rs |
kernel declaration | 0 | 6 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/scatter_axis.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/scatter_row.rs |
kernel declaration | 0 | 4 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/signed_literal.rs |
kernel declaration | 2 | 2 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/ops/split_scatter_columns.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/strided.rs |
kernel declaration | 0 | 4 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/swiglu_limit.rs |
kernel declaration | 0 | 4 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/swiglu_limit_weighted.rs |
kernel declaration | 3 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/ternary.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ops/unary.rs |
kernel declaration | 0 | 7 | 14 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/activation_quantize.rs |
kernel declaration | 2 | 7 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/activation_roundtrip.rs |
kernel declaration | 2 | 8 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/aura_dequant_rotated.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/aura_encode.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/block_scaled_dequant.rs |
kernel declaration | 0 | 30 | 30 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/codec.rs |
kernel declaration | 14 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/quant/dequant_gather.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/dequant_gather_block_scaled.rs |
kernel declaration | 0 | 31 | 30 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/dequant_q4.rs |
kernel declaration | 0 | 2 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/quant/format.rs |
kernel declaration | 8 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/quant/fp4_indexer_roundtrip.rs |
kernel declaration | 0 | 4 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/fp8_block_dequant.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/fp8_kv_roundtrip.rs |
kernel declaration | 3 | 5 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/quant/gguf.rs |
kernel declaration | 4 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/quant/gguf_dequant_iq1_s.rs |
kernel declaration | 0 | 1 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/quant/gguf_dequant_iq2_xxs.rs |
kernel declaration | 1 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/gguf_dequant_iq2_xxs_raw.rs |
kernel declaration | 1 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/gguf_dequant_iq4_nl.rs |
kernel declaration | 0 | 1 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/quant/gguf_dequant_q2_k.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/gguf_dequant_q8_0.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/gguf_iq2_xxs_extract_qs.rs |
kernel declaration | 1 | 0 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/mxfp4_dequant.rs |
kernel declaration | 6 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/mxfp4_expert_gather.rs |
kernel declaration | 0 | 7 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/mxfp4_ggml_split.rs |
kernel declaration | 4 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/mxfp4_quant_act.rs |
kernel declaration | 6 | 4 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/trellis_bgemm.rs |
kernel declaration | 1 | 6 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/trellis_gemm.rs |
kernel declaration | 0 | 9 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/trellis_gemv.rs |
kernel declaration | 0 | 46 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/quant/trellis_ngram_decode.rs |
kernel declaration | 1 | 1 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/rope/base.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/rope/partial_rope.rs |
kernel declaration | 0 | 6 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/rope/partial_rope_compressor_dyn.rs |
kernel declaration | 2 | 4 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/rope/partial_rope_dyn.rs |
kernel declaration | 1 | 4 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/rope/rope_2d.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/rope/rope_banded.rs |
kernel declaration | 0 | 3 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/rope/rope_yarn.rs |
kernel declaration | 0 | 6 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sampling/categorical_sample.rs |
kernel declaration | 0 | 2 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sampling/logits_min_p.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sampling/logits_processors.rs |
kernel declaration | 0 | 2 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sampling/logits_top_p.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sampling/logits_topk.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sampling/logits_topk_tiled.rs |
kernel declaration | 0 | 2 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sampling/row_topk.rs |
kernel declaration | 0 | 1 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/sampling/softmax.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sampling/sort.rs |
kernel declaration | 0 | 3 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/attn_head_gate.rs |
kernel declaration | 0 | 4 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/aura_flash_pass1.rs |
kernel declaration | 0 | 4 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/aura_flash_pass2.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/aura_flash_sdpa.rs |
kernel declaration | 0 | 8 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/aura_score.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/aura_value.rs |
kernel declaration | 0 | 5 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/compressor_overlap_slide.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/compressor_pool.rs |
kernel declaration | 0 | 2 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/compressor_pool_v2.rs |
kernel declaration | 3 | 5 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/csa_sdpa_decode.rs |
kernel declaration | 1 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/csa_sdpa_decode_sink.rs |
kernel declaration | 2 | 1 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/csa_sdpa_decode_sink_dyn.rs |
kernel declaration | 1 | 4 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/csa_select_union.rs |
kernel declaration | 0 | 7 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/dsa_sdpa_decode_d256.rs |
kernel declaration | 0 | 1 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/sdpa/dsa_weighted_kpool.rs |
kernel declaration | 0 | 2 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/sdpa/flash_block_scaled_sdpa.rs |
kernel declaration | 0 | 35 | 31 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/flash_quantized_sdpa.rs |
kernel declaration | 0 | 14 | 4 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/indexer_score.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/indexer_score_dyn.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/indexer_topk.rs |
kernel declaration | 3 | 12 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/indexer_topk_dyn.rs |
kernel declaration | 1 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/indexer_topk_large.rs |
kernel declaration | 5 | 23 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/mla_sdpa_decode.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/mla_sdpa_prefill.rs |
kernel declaration | 0 | 4 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/scaled_dot_product_attention.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_2pass_nq_d256.rs |
kernel declaration | 2 | 1 | 5 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_bidirectional.rs |
kernel declaration | 0 | 10 | 5 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_bidirectional_blockmask.rs |
kernel declaration | 0 | 6 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_bidirectional_d128_relpos.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_bidirectional_windowed.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_decode.rs |
kernel declaration | 2 | 25 | 6 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_decode_2pass.rs |
kernel declaration | 6 | 13 | 9 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_decode_batched.rs |
kernel declaration | 8 | 3 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_decode_d512_sink.rs |
kernel declaration | 2 | 5 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_decode_d512_sink_twopass.rs |
kernel declaration | 4 | 4 | 6 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_multi.rs |
kernel declaration | 0 | 2 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_multi_d256.rs |
kernel declaration | 1 | 2 | 4 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_multi_d256_gqa.rs |
kernel declaration | 1 | 10 | 7 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_multi_d512.rs |
kernel declaration | 1 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_prefill_mma_d256.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_prefill_mma_d256_bk16.rs |
kernel declaration | 0 | 7 | 4 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_prefill_qtiled_d256.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_prefill_sink.rs |
kernel declaration | 1 | 0 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_rel_pos_conformer.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/sdpa_vector.rs |
kernel declaration | 0 | 14 | 7 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/steel_attn/mod.rs |
kernel declaration | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/sdpa/steel_attn/steel_attention.rs |
kernel declaration | 0 | 3 | 4 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/sdpa/steel_attn/steel_attention_mma.rs |
kernel declaration | 0 | 7 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/sdpa/steel_attn/steel_attention_nax.rs |
kernel declaration | 4 | 4 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ssm/gated_delta.rs |
kernel declaration | 1 | 3 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ssm/gated_delta_gate_beta.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ssm/gated_delta_packed_prework.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ssm/gated_delta_prep.rs |
kernel declaration | 2 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ssm/gated_delta_prep_chunk.rs |
kernel declaration | 3 | 17 | 9 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ssm/gated_delta_qknorm_prepass.rs |
kernel declaration | 1 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ssm/gated_delta_qwen35_decode_fused.rs |
kernel declaration | 0 | 5 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/ssm/gated_delta_replay.rs |
kernel declaration | 0 | 2 | 4 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ssm/gated_delta_vec.rs |
kernel declaration | 0 | 5 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ssm/gated_delta_wy.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ssm/gated_delta_wy_plan.rs |
kernel declaration | 3 | 4 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ssm/gated_delta_wy_scan.rs |
kernel declaration | 3 | 6 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ssm/glm5_kda.rs |
kernel declaration | 0 | 5 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/ssm/glm5_qk_l2_norm.rs |
kernel declaration | 0 | 1 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/kernels/ssm/kda_gate.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ssm/scan.rs |
kernel declaration | 3 | 5 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ssm/scan_inplace.rs |
kernel declaration | 0 | 10 | 3 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/ssm/ssm_replay.rs |
kernel declaration | 1 | 9 | 5 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/vision/avg_pool2d_nhwc.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/vision/broadcast_affine.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/vision/dsv4_vision_qkv_rope.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/vision/dsv4_vision_qkv_rope_dyn.rs |
kernel declaration | 0 | 6 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/vision/dsv4_vision_swiglu_split.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/vision/dsv4_vision_unshuffle3.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/vision/frame_diff_luma.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/vision/im2col_patch.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/vision/im2col_patch_interleaved.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/vision/patch_unfold.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/vision/patch_unfold_qwen.rs |
kernel declaration | 0 | 3 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/vision/pos_emb_2d_add.rs |
kernel declaration | 0 | 2 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/vision/resize_normalize.rs |
kernel declaration | 0 | 3 | 2 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/kernels/vision/transpose_th.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/probes/inline_local_probe.rs |
kernel declaration | 0 | 1 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/probes/live_compile_opt_out_probe.rs |
kernel declaration | 2 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/probes/mpp_matmul_probe.rs |
kernel declaration | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/src/probes/simdgroup_load_probe.rs |
kernel declaration | 0 | 1 | 1 | 0 | blocked | performance-baseline |
crates/wh-iron-std/src/utils.rs |
kernel declaration | 21 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/aura_encode_decode_roundtrip_gpu.rs |
GPU / integration | 2 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/aura_msl_snapshots.rs |
GPU / integration | 7 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/batched_4_qgemv_oracle.rs |
GPU / integration | 3 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/batched_4_qmm_fast_oracle.rs |
GPU / integration | 3 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/bm64_vs_gemvrows_iq2xxs.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/conv1d_causal_many_parity_gpu.rs |
GPU / integration | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/cuda_bench_corpus.rs |
GPU / integration | 1 | 0 | 0 | 1 | blocked | backend-device |
crates/wh-iron-std/tests/cuda_buffer_pool_stress_gpu.rs |
GPU / integration | 2 | 0 | 0 | 2 | blocked | backend-device |
crates/wh-iron-std/tests/cuda_kernel_corpus.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/cuda_mmasync_gpu.rs |
GPU / integration | 3 | 0 | 0 | 3 | blocked | backend-device |
crates/wh-iron-std/tests/cuda_wmma_gpu.rs |
GPU / integration | 1 | 0 | 0 | 1 | strong | โ |
crates/wh-iron-std/tests/dequant_gemv_indirect_wrappers.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/dispatch_pinning_guard.rs |
GPU / integration | 2 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/exact_exp_probe.rs |
GPU / integration | 2 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/exp_precision_probe.rs |
GPU / integration | 2 | 0 | 0 | 0 | repair | โ |
crates/wh-iron-std/tests/fft_bluestein_roundtrip_gpu.rs |
GPU / integration | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/gated_delta_prep_chunk_correctness.rs |
GPU / integration | 8 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/gated_delta_prep_step_correctness.rs |
GPU / integration | 14 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/gated_delta_vec_step_vs_chunk_gpu.rs |
GPU / integration | 3 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/gated_delta_wy_cpu_oracle.rs |
GPU / integration | 7 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/gated_delta_wy_microbench.rs |
GPU / integration | 6 | 0 | 0 | 6 | blocked | performance-baseline |
crates/wh-iron-std/tests/gated_delta_wy_pipeline_integration.rs |
GPU / integration | 8 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/gdn_family_cuda_corpus_scoped.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/gdn_qwen36_27b_shape_cuda.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/gemm_q8_correctness.rs |
GPU / integration | 2 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/gemm_q8_mpp_correctness.rs |
GPU / integration | 2 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/gemv_q8_correctness.rs |
GPU / integration | 3 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/gguf_iq2_xxs_correctness_gpu.rs |
GPU / integration | 4 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/glm5_abi_contract.rs |
GPU / integration | 3 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/grouped_gemm_q8_correctness.rs |
GPU / integration | 2 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/grouped_gemv_q8_rows_correctness.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/grouped_gemv_q8_rows_tiled_correctness.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/hip_kernel_corpus.rs |
GPU / integration | 1 | 0 | 0 | 1 | blocked | backend-device |
crates/wh-iron-std/tests/indexer_topk_multiblock_gpu.rs |
GPU / integration | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/kernel_descriptor.rs |
GPU / integration | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/kernel_mode_harvest.rs |
GPU / integration | 1 | 0 | 0 | 1 | repair | โ |
crates/wh-iron-std/tests/kernel_registry_consistency.rs |
GPU / integration | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/kernel_test_inventory.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/kernel_tests_harness.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/kv_cache_quant_roundtrip_gpu.rs |
GPU / integration | 13 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/metal_buffer_pool_stress_gpu.rs |
GPU / integration | 4 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/mla_sdpa_prefill_vs_decode_gpu.rs |
GPU / integration | 3 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/mma_layout_probe.rs |
GPU / integration | 2 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_bgemm_iq2xxs_bm64_correctness.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_bgemm_iq2xxs_mpp_correctness.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_bgemm_iq2xxs_view_correctness.rs |
GPU / integration | 2 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_bgemm_q2k_bm64_correctness.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_bgemm_q2k_mpp_correctness.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_bgemm_q2k_view_correctness.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_bm64_ragged_correctness.rs |
GPU / integration | 3 | 0 | 0 | 1 | blocked | external-fixture |
crates/wh-iron-std/tests/moe_build_tile_plan_bm32_correctness.rs |
GPU / integration | 13 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_build_tile_plan_bm32_own_correctness.rs |
GPU / integration | 13 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_build_tile_plan_bm32_paired_correctness.rs |
GPU / integration | 15 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_build_tile_plan_correctness.rs |
GPU / integration | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_coop_idle_tile_ablation.rs |
GPU / integration | 1 | 0 | 0 | 1 | blocked | performance-baseline |
crates/wh-iron-std/tests/moe_gather_down_q2k_correctness.rs |
GPU / integration | 3 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_gather_gemv_iq2xxs_correctness.rs |
GPU / integration | 2 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_gather_qmm_int4_m16_m32_correctness.rs |
GPU / integration | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_gather_qmm_microbench.rs |
GPU / integration | 3 | 0 | 0 | 3 | blocked | performance-baseline |
crates/wh-iron-std/tests/moe_gather_qmm_mma_bitwidth_correctness.rs |
GPU / integration | 4 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_gather_qmm_mpp_bm64_correctness.rs |
GPU / integration | 4 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_gather_qmm_mpp_bm64_int8_correctness.rs |
GPU / integration | 3 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_gather_qmm_mpp_bm8_correctness.rs |
GPU / integration | 7 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_gather_qmm_mpp_bm8_int8_correctness.rs |
GPU / integration | 7 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_gather_qmm_mpp_correctness.rs |
GPU / integration | 2 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_gather_qmm_mpp_int8_correctness.rs |
GPU / integration | 3 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_gather_qmm_tileplan_correctness.rs |
GPU / integration | 3 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_gemv_rows_iq2xxs_correctness.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_gemv_rows_q2k_correctness.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_gemv_rows_view_iq2xxs_correctness.rs |
GPU / integration | 2 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_gemv_ws_iq2xxs_correctness.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_gemv_ws_q2k_correctness.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_q2k_view_u16_correctness.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_route_sort_plan_fused_correctness.rs |
GPU / integration | 7 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_router_topk_biased_correctness.rs |
GPU / integration | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_sort_plan_counting_correctness.rs |
GPU / integration | 8 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_sort_plan_counting_isolated_bench.rs |
GPU / integration | 2 | 0 | 0 | 2 | blocked | performance-baseline |
crates/wh-iron-std/tests/moe_tile_plan_parallel_correctness.rs |
GPU / integration | 8 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/moe_tile_plan_parallel_isolated_bench.rs |
GPU / integration | 1 | 0 | 0 | 1 | blocked | performance-baseline |
crates/wh-iron-std/tests/moe_view_u16_correctness.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/mpp_matmul_probe.rs |
GPU / integration | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/msl_reduction_loop_gpu.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/mxfp4_decode_graph_capture_gpu.rs |
GPU / integration | 1 | 0 | 0 | 1 | blocked | backend-device |
crates/wh-iron-std/tests/nvfp4_fast_decode_gpu_correctness.rs |
GPU / integration | 4 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/optional_buffer_placeholder_gpu.rs |
GPU / integration | 10 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/partial_rope_correctness.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/prod_shape_gemm_isolation_bench.rs |
GPU / integration | 4 | 0 | 0 | 4 | blocked | performance-baseline |
crates/wh-iron-std/tests/qmm_bm8_correctness.rs |
GPU / integration | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/qmm_mma_bm8_correctness.rs |
GPU / integration | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/qmm_mma_dynamic_m_correctness.rs |
GPU / integration | 7 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/qmm_mpp_correctness.rs |
GPU / integration | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/qmm_mpp_int8_correctness.rs |
GPU / integration | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/rms_norm_per_head_gpu.rs |
GPU / integration | 9 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/rms_warp_rows_correctness.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/rms_warp_rows_numerical.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/rope_yarn_many_parity_gpu.rs |
GPU / integration | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/sdpa_2pass_nq_d256_gpu.rs |
GPU / integration | 4 | 0 | 0 | 1 | blocked | performance-baseline |
crates/wh-iron-std/tests/sdpa_2pass_nq_d256_tq_gpu.rs |
GPU / integration | 5 | 0 | 0 | 1 | blocked | performance-baseline |
crates/wh-iron-std/tests/sdpa_decode_2pass_gpu.rs |
GPU / integration | 19 | 0 | 0 | 7 | blocked | performance-baseline |
crates/wh-iron-std/tests/sdpa_decode_d512_sink_nkv1.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/sdpa_decode_swa_gpu.rs |
GPU / integration | 3 | 0 | 0 | 3 | blocked | performance-baseline |
crates/wh-iron-std/tests/sdpa_prefill_d512_sink_correctness.rs |
GPU / integration | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/sdpa_prefill_mma_bf16_gpu.rs |
GPU / integration | 3 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/sdpa_prefill_mma_long_t.rs |
GPU / integration | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/sdpa_prefill_qtiled_gpu.rs |
GPU / integration | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/sdpa_prefill_sink_window_gpu.rs |
GPU / integration | 13 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/simdgroup_load_probe_gpu.rs |
GPU / integration | 4 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/smallm_occupancy_microbench.rs |
GPU / integration | 4 | 0 | 0 | 4 | blocked | performance-baseline |
crates/wh-iron-std/tests/softmax_categorical_sample_matrix.rs |
GPU / integration | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/softmax_categorical_sample_perf.rs |
GPU / integration | 1 | 0 | 0 | 1 | blocked | performance-baseline |
crates/wh-iron-std/tests/ssm_scan_a2d_inplace_parity_gpu.rs |
GPU / integration | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/ssm_scan_inplace_parity_gpu.rs |
GPU / integration | 9 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/ssm_scan_parity_gpu.rs |
GPU / integration | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/steel_msl_snapshots.rs |
GPU / integration | 7 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/swiglu_joined_rounded.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/swiglu_joined_rounded_numerical.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/vulkan_add_bias_rows.rs |
GPU / integration | 1 | 0 | 0 | 1 | blocked | backend-device |
crates/wh-iron-std/tests/vulkan_coopmat_gemm.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron-std/tests/vulkan_kernel_corpus.rs |
GPU / integration | 1 | 0 | 0 | 1 | blocked | backend-device |
crates/wh-iron-std/tests/vulkan_sdpa_multi.rs |
GPU / integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron/src/harness/bench.rs |
unit / inline | 9 | 0 | 0 | 0 | strong | โ |
crates/wh-iron/src/harness/test.rs |
unit / inline | 3 | 0 | 0 | 0 | strong | โ |
crates/wh-iron/src/runner/args.rs |
unit / inline | 7 | 0 | 0 | 0 | strong | โ |
crates/wh-iron/src/runner/backend.rs |
unit / inline | 4 | 0 | 0 | 0 | strong | โ |
crates/wh-iron/src/runner/device_specs.rs |
unit / inline | 5 | 0 | 0 | 0 | strong | โ |
crates/wh-iron/src/runner/gpu.rs |
unit / inline | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron/src/runner/harness.rs |
unit / inline | 12 | 0 | 0 | 0 | strong | โ |
crates/wh-iron/src/runner/profile.rs |
unit / inline | 6 | 0 | 0 | 0 | strong | โ |
crates/wh-iron/tests/compile_fail.rs |
crate integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron/tests/msl_elementwise_tail_guard.rs |
crate integration | 1 | 0 | 0 | 0 | strong | โ |
crates/wh-iron/tests/variants_row_metadata.rs |
crate integration | 2 | 0 | 0 | 0 | strong | โ |
Completed audit classification
Section titled โCompleted audit classificationโEvery inventory row is classified: 198 strong,
328 blocked, 2 repair, and
0 auditing (no rows are queued). repair rows name checks
that run (or are ignored) but do not gate on an independent oracle; each is
expected to be deleted or replaced by a stronger owner in a focused PR. The
blocked ledger is complete: 313 performance-baseline, 14
backend-device, and 1 external-fixture boundaries. The
classification is deliberately file-wide: a file remains blocked when any
declaration in it is an ignored campaign, a benchmark
without a committed regression threshold, or an accelerator integration that
cannot run in the default lane. This avoids calling a mixed file strong merely
because its cheaper declarations passed.
The always-on Rust workspace and registered Metal kernel corpus were replayed
from the audited revision with cargo test --workspace. Strong rows executed
in that replay and own an asserted value, structural, protocol, snapshot, or
CPU-reference oracle. The registry witness and realistic divergence cases
(including bit flips, wrong mappings, invalid geometry, and stale-state
mutants) establish that the harness does not silently omit registered kernel
claims. Accelerator campaigns now use explicit #[ignore] gates and fail
closed when deliberately selected; absence of CUDA, HIP, or Vulkan is recorded
as blocked evidence rather than a passing test.
Blocked benchmark rows require a correctness owner plus a pinned performance claim (threshold or baseline receipt) before promotion. A benchmark that only prints timing remains blocked even when it compiles and runs.
Execution lanes that own the inventory
Section titled โExecution lanes that own the inventoryโ| Lane | Intended reach | RST evidence boundary |
|---|---|---|
make test |
Workspace Rust tests and the Cargo bridge for declarative kernel checks | A pass may include backend/capability skips; inspect executed and skipped totals. |
iron test |
Registered GPU correctness corpus | Record selected names or inventory digest, device/backend, filters, and skip reasons; see issue #190. |
make test-validate |
Metal validation-instrumented diagnostic slice | Instrumentation changes resource limits and is not a universal correctness gate. |
| Ignored correctness companions | Explicit campaigns invoked with --ignored |
Ordinary CI does not execute these declarations. |
iron bench / ignored benches |
Performance corpus | Timing is evidence only after correctness and harness validation. |
Audit order
Section titled โAudit orderโ- Prove the renderer recognizes a known declaration and fails visibly if a new declaration form is introduced.
- Audit registration, filtering, skip accounting, and CI lane reachability.
- Audit independent CPU oracles and dtype/shape/tail partitions by kernel family.
- Audit DSL, IR, pass, codegen, snapshot, and compile-fail contracts.
- Audit runtime binding, aliasing, ordering, graph, and lifetime contracts.
- Audit benchmark harnesses only after the corresponding correctness owners.
For each focused PR, apply the Atlas questions: exact claim, production setup reachability, oracle relevance, realistic mutant survival, and redundancy with a stronger owner. If the repaired test exposes a product defect, fix that defect in the same PR.