cutlass

cutlass copied to clipboard

Tiny fix tutorial text

6

from my naive understanding the second arrow (output "?") is correct

fzyzcjy

[BUG] CuteDSL example hits IMA for large tensors due to strides in Int32

6

Tensors in Cute DSL uses strides in int32 by default. This causes IMA for large tensors. Is there a way to force strides to be int64? **Steps/Code to reproduce bug**...

tridao

[FEA] [CuTeDSL] `print_latex` in `CuTeDSL`

12

**Is your feature request related to a problem? Please describe.** It would be nice to have utility function in `CuTeDSL` like `print_latex` in `C++` API **Describe the solution you'd like**...

simveit

[QST] [CuTeDSL] Nsight Compute Profiler Link to Source Code

8

**What is your question?** When profiling CUDA/CUTLASS, the profiler can provide line-by-line profiling for user code, in addition to PTX and SASS. Triton can also do this, likely because its...

HanGuo97

Mixed Precision Grouped Gemm with zero points and GPT-Q semantics closes #2261

8

Hello! This MR provides two things: 1) Zero points for default mode 2) GPT-Q [semantics](https://pytorch.org/blog/accelerating-triton/) Closes #2261

ankutalev

Missing PYPI releases

10

At least 4.1 and `3.9.0.0` is missing from PYPI. As cuda-python 12.6.2 is required for CUDA 12.6 and that has API deprecations (`import cuda.bindings.cuda` instead of `import cuda.cuda`) nvidia-cutlass 4.1...

Flamefire

Rtoax

[QST] The strange bank conflict in the CuTeDSL python gemm demo.

1

## code https://github.com/NVIDIA/cutlass/blob/main/examples/python/CuTeDSL/ampere/tensorop_gemm.py ## env ``` pip list | grep -i cutlass nvidia-cutlass-dsl 4.3.0 nvcc -V nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2025 NVIDIA Corporation Built on Wed_Aug_20_01:57:39_PM_PDT_2025...

LRlr239