Updated NVIDIA CUDA For WSL Brings Better Performance, PTX JIT
Earlier this summer building off the latest Windows Subsystem for Linux 2 advancements by Microsoft, NVIDIA released early support for CUDA / GPU compute on WSL2. This week NVIDIA offered up a new version of their CUDA WSL support.
The updated NVIDIA CUDA implementation with Windows Subsystem for Linux brings better performance particularly for smaller workloads, DirectML API for DirectX 12 GPU acceleration, and support for PTX JIT. The PTX JIT support allows developers to run the PTX representation on WSL and loads from the driver store directly.
The better performance comes by optimizing the driver for launch and synchronization tasks. Additional performance improvements for NVIDIA GPU acceleration within WSL is still being worked on for future updates.
NVIDIA also is still planning WSL support for OptiX and other driver libraries, improved memory access performance, support for NVML and NVIDIA SMI for GPU management, and better support for multi-GPU systems.
More details on this week's NVIDIA CUDA WSL update via developer.nvidia.com.
The updated NVIDIA CUDA implementation with Windows Subsystem for Linux brings better performance particularly for smaller workloads, DirectML API for DirectX 12 GPU acceleration, and support for PTX JIT. The PTX JIT support allows developers to run the PTX representation on WSL and loads from the driver store directly.
The better performance comes by optimizing the driver for launch and synchronization tasks. Additional performance improvements for NVIDIA GPU acceleration within WSL is still being worked on for future updates.
NVIDIA also is still planning WSL support for OptiX and other driver libraries, improved memory access performance, support for NVML and NVIDIA SMI for GPU management, and better support for multi-GPU systems.
More details on this week's NVIDIA CUDA WSL update via developer.nvidia.com.
10 Comments