Skip to content

Pull requests: NVIDIA/cutlass

New pull request New
158 Open 885 Closed
158 Open 885 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Enable tcgen05 blockscaled ops on Thor SM110
#3283 opened May 28, 2026 by xiangg-nv Loading…
[SM120] Add ptr-array TMA collective for tensor/token-scaled FP8 grouped GEMM
#3280 opened May 27, 2026 by tgmerritt Loading…
1
1
fix: missing PDL wait on main_sf_load in sm103 blockscaled GEMM
#3279 opened May 27, 2026 by tianyuxbear Loading…
Fix the ScatterD issue in predicated_tile_iterator
#3278 opened May 27, 2026 by pengpeng-yu Loading…
1
[CuTeDSL] Fix the incorrect import of from_dlpack from docs
#3276 opened May 26, 2026 by kainzhong Loading…
[CuTeDSL] Add SM120 MXF4/NVFP4 native-TMA path
#3273 opened May 25, 2026 by alecco Loading…
1
[CuTeDSL] Make @cute.struct instances flattenable across scf.if / scf.while
#3270 opened May 25, 2026 by cheshire Loading…
1
4
[CuTeDSL] Lower scalar Float16/BFloat16 load through Uint16+bitcast
#3267 opened May 24, 2026 by cheshire Loading…
1
fix: handle ComposedLayout slicing with dynamic strides (fixes #3255)
#3261 opened May 22, 2026 by zhils Loading…
[DOC] Drop empty duplicated Numeric Conversion code block from fundamental_types.md
#3260 opened May 22, 2026 by adityasingh2400 Loading…
3 tasks done
1
fix(CuTeDSL): restore trailing Int<1> dimension in SM90 MMA atom TV L…
#3258 opened May 21, 2026 by zhils Loading…
1
1
[Cutlass SM90] Per-group aux TMA descriptor update for grouped GEMM + Gated-SwiGLU example)
#3256 opened May 21, 2026 by Butterfingrz Loading…
test: add CPU-only unit coverage for sharding helpers
#3250 opened May 19, 2026 by Pritiks23 Loading…
2
fix(base_dsl): drop ArchMeta alias so Arch.sm_*.value is correct
#3248 opened May 19, 2026 by lingolin128 Loading…
7
Filter SM120 mixed 8-bit tiles for FP6 ElementD
#3247 opened May 19, 2026 by zhils Loading…
1
1
Fix FastDivmod divisor SSA transport for kernel regions (#3243)
#3246 opened May 19, 2026 by zhils Loading…
1
2
[cutlass-library] Alias cutlass_lib to the static target when shared is off (fixes #3179)
#3245 opened May 18, 2026 by LeSingh1 Loading…
1
[CuTe] Add missing include for smem_ptr_flag_bits in print_tensor.hpp
#3244 opened May 18, 2026 by LeSingh1 Loading…
1
[fast_math] Add bfloat16_t PTX specializations for fast_exp and fast_tanh
#3242 opened May 16, 2026 by VittoriaLanzo Loading…
5 tasks done
fix an intermittent accuracy isse
#3233 opened May 15, 2026 by dishengbin Loading…
1
Fix MSVC CUDA build: is_unsigned_v not available in cutlass::platform
#3229 opened May 12, 2026 by TxsharDev Loading…
W4a8 speedup v2
#3226 opened May 11, 2026 by mak-corp Loading…
[example Cute C++]Add CuTe C++ tutorial for Blackwell MXFP8 block-scaled GEMM.
#3225 opened May 11, 2026 by haowen-han Loading…
Avoid unordered_map for runtime datatype mapping
#3223 opened May 11, 2026 by LwhJesse Loading…
1
FMHA examples: use cute::min in device functions
#3222 opened May 11, 2026 by LwhJesse Loading…
1
ProTip! no:milestone will show everything without a milestone.