deepgemm

Here are 4 public repositories matching this topic...

idrees2516 / deepgemm-rust

DeepGEMM in pure Rust: FP8/BF16 fine-grained-scaled GEMM + MoE grouped GEMM for NVIDIA SM90a (TMA+WGMMA warp-specialized kernels, NVRTC runtime JIT, zero torch/nvcc dependency)

rust gpu cuda moe hopper gemm b200 nvrtc blackwell fp8 deepseek fp4 wgmma deepgemm tcgen05

Updated Oct 4, 2026
Rust

GTC2080 / ascend-operator-skills

Star

Portable Ascend C and TileLang operator engineering skill with tested workflows and lessons from DeepGEMM and TileKernels.

npu ascend agent-skills kernel-optimization deepgemm tilelang

Updated Oct 16, 2026
Python

jessiedong01 / deepgemm-swapab-bench

Star

What DeepGEMM's SwapAB layout is worth on B300, with DeepSeek-V3.2 GEMM shapes in FP8, FP4 and BF16. CUPTI-measured.

benchmark cuda gemm blackwell fp8 deepgemm

Updated Aug 14, 2026
Python

ecohash-co / deepseek-v4-flash-dgx-spark-vllm027

Star

Running DeepSeek-V4-Flash with DSpark speculative decoding on 2x DGX Spark (GB10/sm_121) using upstream vLLM 0.27.1 — the three blockers, the patches, and the measurements

blackwell vllm llm-inference speculative-decoding deepseek flashinfer gb10 sm120 dgx-spark deepgemm

Updated Aug 12, 2026
Python

Add this topic to your repo

To associate your repository with the deepgemm topic, visit your repo's landing page and select "manage topics."

Learn more