Popular repositories Loading

1. sm120-moe-bench sm120-moe-bench Public

   SM120 MoE Inference Benchmark: Qwen3.5-397B on RTX PRO 6000 Blackwell — K=64 CUTLASS kernel fix + real-world legal prompt benchmarks

   Cuda 3

2. flashinfer flashinfer Public

   Forked from flashinfer-ai/flashinfer

   FlashInfer: Kernel Library for LLM Serving

   Python 1

3. vllm vllm Public

   Forked from vllm-project/vllm

   A high-throughput and memory-efficient inference and serving engine for LLMs

   Python 1

4. cutlass cutlass Public

   Forked from NVIDIA/cutlass

   CUDA Templates and Python DSLs for High-Performance Linear Algebra

   C++ 1

5. free-code free-code Public

   Forked from paoloanzn/free-code

   The free build of Claude Code. All telemetry removed, security-prompt guardrails stripped, all experimental features enabled.

   TypeScript 1

6. verdict-warp-decode verdict-warp-decode Public

   Neuron-centric fused MoE kernel for SM120 NVFP4 — 17.5μs/layer, 1.02x faster than VerdictMoE, 5.6x faster than CUTLASS

   Cuda 1