3 ms·
What is needed are true NPUs as dedicated co-processors, especially for prosumer desktop systems (devs, other professionals, gamers). GPUs work in the enterpris
by uyzstvqs 2y ago
What is needed are true NPUs as dedicated co-processors, especially for prosumer desktop systems (devs, other professionals, gamers). GPUs work in the enterprise, but they're a hassle to use for AI on the personal computing side of the market. Especially VRAM limitations, but also the lack of a standard open API other than Vulkan (again, using video stuff for AI).
- dartos 2y agoFwiw, Vulkan isn’t specifically a graphics api and has had compute specific features for a while now. (Potentially since its inception)
- the__alchemist 2y agoCompared to CUDA, Vulkan is... not fun to code compute in! The serialization bridge and duplicating data structures and functions between CPU and GPU is tedious.
- dartos 2y agoI hear both CUDA and Vulkan are not fun to code in. But yeah Vulkan is famously verbose. It takes about 1000 LoC to draw a triangle
- KeplerBoy 2y agoCUDA is very much fun to code in! Nvidia provides devs with great tools (Nsight Systems and Nsight Compute), so you know where you have to optimize.