Search results

test_mhas_v2.py spurious cuMemCreate NOT_PERMITTED errors after FE 1.24.0 → 1.25.0 PYTORCH_CUDA_ALLOC_CONF change

Status: Open.

#287 In NVIDIA/cudnn-frontend;

· kangbintNV opened on Jun 5, 2026

Missing THD support for cuDNN 9.23 + cuDNN FE 1.24 for D=256 bwd on sm10.x

cat-featureRequests for new functionality, APIs, examples, or behavior improvements.Requests for new functionality, APIs, examples, or behavior improvements.

mod-backendcuDNN backend API, graph execution, descriptors, engines, or backend integration.cuDNN backend API, graph execution, descriptors, engines, or backend integration.

orig-nv-engReported or requested by NVIDIA engineering.Reported or requested by NVIDIA engineering.

Status: Open.

Enhancement

#276 In NVIDIA/cudnn-frontend;

· KshitijLakhani opened on Jun 3, 2026

[Question] About SM90 support for DSA IndexerForward

cat-enhancements

cat-questionQuestions, usage help, clarifications, or support requests.Questions, usage help, clarifications, or support requests.

mod-cutedslCuTeDSL kernels, generated kernels, examples, or related integration work.CuTeDSL kernels, generated kernels, examples, or related integration work.

orig-externalReported or requested by an external user, customer, or community contributor.Reported or requested by an external user, customer, or community contributor.

Status: Open.

#264 In NVIDIA/cudnn-frontend;

· tkandi opened on Jun 1, 2026

·  

Frontend 1.25.0

Add SDPA test coverage for all-negative attention scores (negative row max)

cat-enhancements

mod-frontendcuDNN frontend APIs, operation graph construction, plans, and user-facing wrappers.cuDNN frontend APIs, operation graph construction, plans, and user-facing wrappers.

Status: Open.

#262 In NVIDIA/cudnn-frontend;

· Anerudhan opened on May 29, 2026

[PyTorch][CUDNN][Windows] Version 1.22.1 fails to compile with NV_CUDNN_FRONTEND_USE_DYNAMIC_LOADING

cat-enhancements

mod-frontendcuDNN frontend APIs, operation graph construction, plans, and user-facing wrappers.cuDNN frontend APIs, operation graph construction, plans, and user-facing wrappers.

open-todoOpen work that is ready to be picked up, scheduled, or prioritized.Open work that is ready to be picked up, scheduled, or prioritized.

Status: Open.

#261 In NVIDIA/cudnn-frontend;

· Anerudhan opened on May 29, 2026

·  

Frontend 1.26.0

cuDNN frontend grouped GEMM fusion API is restricted to PyTorch

cat-featureRequests for new functionality, APIs, examples, or behavior improvements.Requests for new functionality, APIs, examples, or behavior improvements.

mod-frontendcuDNN frontend APIs, operation graph construction, plans, and user-facing wrappers.cuDNN frontend APIs, operation graph construction, plans, and user-facing wrappers.

orig-nv-engReported or requested by NVIDIA engineering.Reported or requested by NVIDIA engineering.

Status: Open.

#260 In NVIDIA/cudnn-frontend;

· Anerudhan opened on May 29, 2026

·  

Frontend 1.26.0

grouped_gemm_dglu_wrapper_sm100 became slower between 1.23.0 and 1.24.0

cat-perf-bugPerformance regressions or cases where behavior is correct but too slow.Performance regressions or cases where behavior is correct but too slow.

mod-backendcuDNN backend API, graph execution, descriptors, engines, or backend integration.cuDNN backend API, graph execution, descriptors, engines, or backend integration.

mod-cutedslCuTeDSL kernels, generated kernels, examples, or related integration work.CuTeDSL kernels, generated kernels, examples, or related integration work.

open-todoOpen work that is ready to be picked up, scheduled, or prioritized.Open work that is ready to be picked up, scheduled, or prioritized.

orig-nv-engReported or requested by NVIDIA engineering.Reported or requested by NVIDIA engineering.

Status: Open.

Bug

#256 In NVIDIA/cudnn-frontend;

· jepio opened on May 27, 2026

·  

Frontend 1.25.0

Add enforce_precompiled flag to graph deserialization

cat-featureRequests for new functionality, APIs, examples, or behavior improvements.Requests for new functionality, APIs, examples, or behavior improvements.

mod-frontendcuDNN frontend APIs, operation graph construction, plans, and user-facing wrappers.cuDNN frontend APIs, operation graph construction, plans, and user-facing wrappers.

open-todoOpen work that is ready to be picked up, scheduled, or prioritized.Open work that is ready to be picked up, scheduled, or prioritized.

orig-nv-engReported or requested by NVIDIA engineering.Reported or requested by NVIDIA engineering.

Status: Open.

Task

#254 In NVIDIA/cudnn-frontend;

· Anerudhan opened on May 26, 2026

·  

Frontend 1.25.0

[FI] SM103 cuDNN BF16 GEMM does not support FP16 as an output datatype

cat-featureRequests for new functionality, APIs, examples, or behavior improvements.Requests for new functionality, APIs, examples, or behavior improvements.

mod-frontendcuDNN frontend APIs, operation graph construction, plans, and user-facing wrappers.cuDNN frontend APIs, operation graph construction, plans, and user-facing wrappers.

Status: Open.

Bug

#203 In NVIDIA/cudnn-frontend;

· raayandhar opened on Feb 10, 2026

·  

Frontend 1.25.0

[Question] Need clarification on the bn_finalize and dbn_weight interfaces

Status: Open.

#191 In NVIDIA/cudnn-frontend;

· GeezYuven opened on Dec 8, 2025

Scale is optional input in norms

Status: Open.

#188 In NVIDIA/cudnn-frontend;

· zcbenz opened on Dec 3, 2025

The populate_cuda_graph creates an unnecessary CUDA graph

Status: Open.

#186 In NVIDIA/cudnn-frontend;

· zcbenz opened on Dec 3, 2025