Search results

[Bug] k_grouped_fp8_gemm_nt_contiguous crashes with n = 768 on H100

Status: Open.

#237 In deepseek-ai/DeepGEMM;

· xinqiu opened on Dec 2, 2025

Feature Request: Support sm_120 ( 5090 and blackwell 6000 pro )

Status: Open.

#236 In deepseek-ai/DeepGEMM;

· createthis opened on Dec 1, 2025

[Question] Contiguous Grouped Gemm for CUDA Graph

Status: Open.

#235 In deepseek-ai/DeepGEMM;

· yuhyao opened on Nov 28, 2025

[Feature Request] Pip install support

Status: Open.

#232 In deepseek-ai/DeepGEMM;

· yewentao256 opened on Nov 21, 2025

__version__ doesn't match Git tags for post-releases, breaking build tools

Status: Open.

#228 In deepseek-ai/DeepGEMM;

· pavank63 opened on Nov 18, 2025

Question: Are ARM prebuilt wheels

Status: Open.

#223 In deepseek-ai/DeepGEMM;

· bbartels opened on Oct 28, 2025

[QUESTION] Can fp8_paged_mqa_logits support next_n > 2?

Status: Open.

#222 In deepseek-ai/DeepGEMM;

· starwang1024 opened on Oct 24, 2025

Can fp8_paged_mqa_logits support non-contiguous layout?

Status: Open.

#221 In deepseek-ai/DeepGEMM;

· MARD1NO opened on Oct 17, 2025

Question: planned support for Geforce GPUs (at least Blackwell sm_12x)..

Status: Open.

#216 In deepseek-ai/DeepGEMM;

· oscarbg opened on Oct 15, 2025

[Question] Supported CUDA/Torch/Python combinations

Status: Open.

#212 In deepseek-ai/DeepGEMM;

· ko3n1g opened on Oct 11, 2025

[Feature] Support Bias

Status: Open.

#210 In deepseek-ai/DeepGEMM;

· complexfilter opened on Oct 8, 2025

[QUESTION] dsv32 mqa_logits kernel not considering causal masking?

Status: Open.

#209 In deepseek-ai/DeepGEMM;

· carlushuang opened on Oct 7, 2025