Environment:Sgl project Sglang CUDA
Sgl_project_Sglang_CUDA is the NVIDIA CUDA development toolkit environment for SGLang, providing the compiler, libraries, and headers needed to build and run custom GPU kernels.
Requirements
- NVIDIA GPU (SM75+ / Turing or newer)
- CUDA Toolkit 12.3+ (12.8+ for Blackwell SM100/SM120)
- NVCC compiler and CUDA runtime libraries
- cuBLAS, cuDNN, cuSPARSE
- NVIDIA driver 550+
- `cuda-python` == 12.9
- `nvidia-cutlass-dsl` >= 4.3.4
Required By
- Implementation:Sgl_project_Sglang_GPU_Trace_To_Graph
- Implementation:Sgl_project_Sglang_PR_Test_Workflow
- Implementation:Sgl_project_Sglang_SM100_MLA_Device
- Implementation:Sgl_project_Sglang_SM100_MLA_TMA_WarpSpecialized
See Also
Page Connections
Double-click a node to navigate. Hold to expand connections.
Principle
Implementation
Heuristic
Environment