biondizzle 197cac875c fix: CUDA 13 TMA descriptor — 3D rank + byte strides + mbarrier byte count ...

Three critical fixes for CUDA 13.x on Blackwell:
1. globalStrides are in BYTES not elements (CUDA 13 change)
2. Use 3D descriptors (degenerate 3rd dim=1) — CUDA 13 TMA requires rank >= 2
3. mbarrier init uses expected byte count (4096 for 128x16 BF16 tile)
4. cp.async.bulk.tensor.3d instead of .2d for 3D descriptors
5. BFLOAT16 data type instead of UINT16

2026-05-29 06:34:58 +00:00

..

attention

fix: CUDA 13 TMA descriptor — 3D rank + byte strides + mbarrier byte count

2026-05-29 06:34:58 +00:00

cache

KV Cache: schema, allocator, pools, manager, append_swa kernel

2026-05-22 00:08:38 +00:00

compressor

Restructure: cutedsl/ -> dsv4/ with proper layering

2026-05-21 17:30:44 +00:00

cuda

Clean up E2M1 dequant: use LUT approach (consultant recommendation)

2026-05-28 16:17:47 +00:00

decode

Restructure: cutedsl/ -> dsv4/ with proper layering

2026-05-21 17:30:44 +00:00

gemm

NVFP4-1.1: Mark fp4_quant.py as toolchain-blocked, clean up test files

2026-05-28 04:59:01 +00:00

indexer

Clean up E2M1 dequant: use LUT approach (consultant recommendation)

2026-05-28 16:17:47 +00:00

router

Router: Blackwell-native fused decode kernel — real CuTeDSL implementation

2026-05-21 22:04:20 +00:00

__init__.py

Restructure: cutedsl/ -> dsv4/ with proper layering

2026-05-21 17:30:44 +00:00