All issues

搜索 results

Blockscaled GEMV accumulates past K when floor(K/tile) is not a multiple of the stage count: example 91 fails for batch > 1

Status: Open.

#3568 In NVIDIA/cutlass;

· TANGBUDU opened on Aug 29, 2026

[BUG] CuTe DSL: _merge_gpu_link_libraries emits link-libraries in Python set order, so the compile pipeline string varies per process

Status: Open.

#3564 In NVIDIA/cutlass;

· yurekami opened on Aug 28, 2026

[DOC] Request: Error handling documentation for CuTe DSL exported C API

? - Needs Triage

documentationDocumentationDocumentation

Status: Open.

#3562 In NVIDIA/cutlass;

· UnpureRationalist opened on Aug 28, 2026

Conv device wrappers: update() leaves stale geometry and fused-output pointers, two wrappers miss the 2 GiB gate, negative output extents not rejected

Status: Open.

#3545 In NVIDIA/cutlass;

· VaggelisGian opened on Aug 25, 2026

Small profiler defects: device-ID validation off by one, kTensorNCHW mapped to TensorNHWC, sparse-meta host/device type mismatch

Status: Open.

#3544 In NVIDIA/cutlass;

· VaggelisGian opened on Aug 25, 2026

Three host reference utility defects: wrong MRE formula, operator precedence in padded-diagonal fill, unseeded per-element RNG in gaussian fill

Status: Open.

#3543 In NVIDIA/cutlass;

· VaggelisGian opened on Aug 25, 2026

Validation gaps: grouped/split-K can_implement accept crashing configs; batch grid wraps at 65536; Sm80 EVT mode gates are assert-only

Status: Open.

#3542 In NVIDIA/cutlass;

· VaggelisGian opened on Aug 25, 2026

Several 2.x device wrappers have update()/operator() paths that never compile or silently apply stale state

Status: Open.

#3541 In NVIDIA/cutlass;

· VaggelisGian opened on Aug 25, 2026

GemmUniversalWithBroadcast ColumnMajor-output transpose leaves the bias vector indexed in the wrong coordinate system

Status: Open.

#3540 In NVIDIA/cutlass;

· VaggelisGian opened on Aug 25, 2026

EllGemm ColumnMajor-output workspace sizing uses the unswapped problem shape and undersizes split-K semaphores

Status: Open.

#3539 In NVIDIA/cutlass;

· VaggelisGian opened on Aug 25, 2026

GemmWithKReduction parallel split-K: workspace initialization is skipped, partials land in the user output buffer

Status: Open.

#3538 In NVIDIA/cutlass;

· VaggelisGian opened on Aug 25, 2026

Blockwise-scaled split-K GEMM applies slice-0 scale factors to every K slice

Status: Open.

#3537 In NVIDIA/cutlass;

· VaggelisGian opened on Aug 25, 2026