Vulkan compute
Definition pending
We've cataloged "Vulkan compute" but haven't written a full definition yet. Definitions are hand-curated rather than auto-generated, so it takes time to cover every term.
Want this one prioritized? Email us and we'll bump it.
Practical example
Vulkan is a cross-platform GPU API that works on NVIDIA, AMD, Intel, and even some ARM GPUs. It's the "run anywhere" GPU backend for llama.cpp. Performance is typically 5–15% slower than CUDA (NVIDIA) or Metal (Apple) but works on hardware those backends don't support — AMD GPUs, Intel Arc, Raspberry Pi 5 GPU.
Workflow example
Use Vulkan backend when: (1) you have an AMD GPU (no CUDA), (2) you have an Intel Arc GPU, (3) you want a single build that runs everywhere. In llama.cpp: build with LLAMA_VULKAN=1 make. In Ollama: Vulkan backend is available on Linux. Benchmark CUDA vs Vulkan on your NVIDIA GPU — if Vulkan is within 10% of CUDA performance, consider using it for broader hardware compatibility in your deployment.