Skip to content

[None][chore] Upgrade CuteDSL to 4.3.0 #9444

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
syuoni merged 3 commits into from
Nov 26, 2025
Merged

[None][chore] Upgrade CuteDSL to 4.3.0 #9444

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
0267ce0
upgrade
syuoni Nov 24, 2025
c3d4bfa
fix
syuoni Nov 25, 2025
f68dff3
fix
syuoni Nov 25, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion ATTRIBUTIONS-Python.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25250,7 +25250,7 @@ License: `NVIDIA Proprietary Software`
- `Homepage`: https://developer.nvidia.com/cusparselt


## nvidia-cutlass-dsl (4.2.1)
## nvidia-cutlass-dsl (4.3.0)

### Licenses
License: `None`
Expand Down
2 changes: 1 addition & 1 deletion requirements.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -69,7 +69,7 @@ triton==3.5.0; platform_machine == "x86_64"
tiktoken
blobfile
openai-harmony==0.0.4
nvidia-cutlass-dsl==4.3.0.dev0; python_version >= "3.10"
nvidia-cutlass-dsl==4.3.0; python_version >= "3.10"
plotly
numexpr<2.14.0 # WAR for attempted use of nonexistent numpy.typing
partial_json_parser
7 changes: 4 additions & 3 deletions .../_torch/cute_dsl_kernels/blackwell/blockscaled_contiguous_grouped_gemm_finalize_fusion.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1552,6 +1552,8 @@ def kernel(
epi_tidx, tCtAcc_base, tCgC, epi_tile, use_2cta_instrs
)

tTR_rC = cute.make_rmem_tensor(tTR_rAcc.shape, self.out_dtype)

copy_atom_r2s = sm100_utils.get_smem_store_op(
self.gemm_output_layout, self.out_dtype, self.acc_dtype, tiled_copy_t2r
)
Expand Down Expand Up @@ -1641,8 +1643,6 @@ def kernel(
layout = cute.make_layout(shape=(cute.size(tTR_rAcc),), stride=(1,))
loop_size = cute.size(tTR_rAcc)

rOut_epi = cute.make_rmem_tensor(layout, self.out_dtype)

for subtile_idx in cutlass.range(subtile_cnt):
#
# Load accumulator from tensor memory buffer to register
Expand All @@ -1657,7 +1657,8 @@ def kernel(
# Apply router scale to the entire row (broadcast scalar to vector)
acc_vec_finalized = token_scale * acc_vec_scaled

rOut_epi.store(acc_vec_finalized.to(self.out_dtype))
tTR_rC.store(acc_vec_finalized.to(self.out_dtype))
rOut_epi = cute.make_tensor(tTR_rC.iterator, layout)

if permuted_row < tile_mn_limit:
coord_n = mma_tile_coord_mnl[1] * self.cta_tile_shape_mnk[
Expand Down
1 change: 1 addition & 0 deletions tensorrt_llm/commands/eval.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -135,6 +135,7 @@ def main(ctx, model: str, tokenizer: Optional[str], log_level: str,

profiler.start("trtllm init")
if backend == 'pytorch':
llm_args.pop("build_config", None)
llm = PyTorchLLM(**llm_args)
elif backend == 'tensorrt':
llm = LLM(**llm_args)
Expand Down