Search results

The comparison between cublaslt and deepgemm in test_fp8.py is unaligned

Status: Open.

#199 In deepseek-ai/DeepGEMM;

· Insideyyy opened on Sep 26, 2025

Tested tests/test_fp8.py on SM100 with very poor performance

Status: Open.

#197 In deepseek-ai/DeepGEMM;

· wasamtc opened on Sep 24, 2025

[Feature] Speed up DeepGEMM v2 warm up

Status: Open.

#196 In deepseek-ai/DeepGEMM;

· yuan-luo opened on Sep 22, 2025

Perf regression since sha f85ec6

Status: Open.

#195 In deepseek-ai/DeepGEMM;

· hj-mistral opened on Sep 19, 2025

Wheel is not supported in the platform

Status: Open.

#194 In deepseek-ai/DeepGEMM;

· acphile opened on Sep 16, 2025

The current strategy in get_best_config is suboptimal in some cases on H20

Status: Open.

#191 In deepseek-ai/DeepGEMM;

· Sulfur6 opened on Sep 11, 2025

关于Wgrad 计算中 fp8_gemm_nt 输入量化方式的疑问与实验观察

Status: Open.

#189 In deepseek-ai/DeepGEMM;

· HoshinoAkua opened on Sep 10, 2025

How can I use DeepGemm with sm120a（5060）?

Status: Open.

#185 In deepseek-ai/DeepGEMM;

· Lyxien opened on Sep 4, 2025

Create Git tags and releases for versioning

Status: Open.

#177 In deepseek-ai/DeepGEMM;

· EmilienM opened on Aug 28, 2025

B200(sm=100a) FP8 accumulator bits

Status: Open.

#176 In deepseek-ai/DeepGEMM;

· lisuying214 opened on Aug 28, 2025

Support grouped gemm bf16 tn/nn contiguous?

Status: Open.

#171 In deepseek-ai/DeepGEMM;

· zfan2356 opened on Aug 26, 2025

Partner Listing Request：PaddlePaddle 的 PyTorch C++/Python 接口兼容项目

Status: Open.

#170 In deepseek-ai/DeepGEMM;

· yongqiangma opened on Aug 25, 2025