Alternative(s) to run CUDA on non-Nvidia hardware

Post by woodrowbarlow » Wed, Jul 15, 2026, 12:00 PM UTC

inigyou wrote: post #48

woodrowbarlow wrote: post #44

i'm also interested in tenstorrent. they're building GPUs with cheap GDDR6 using a fast SRAM cache, and writing their own compiler stack (used instead of CUDA) that pipelines data to the SRAM ahead-of-time so you (in theory) never need to suffer the slow speed of GDDR6 for AI workloads. also they've got built-in SFP cages where the video ports would normally be.

Is tenstorrent building GPUs now, not just tensor processors?

you're right, i misspoke by saying it's a GPU since it doesn't have video output. the just-in-time data pipelining wouldn't work for graphics workloads anyway.

Post by zdw » Wed, Jul 15, 2026, 2:45 PM UTC

LogicFailsMe wrote: post #47

woctordho wrote: post #13

There's nothing wrong to run CUDA on non-Nvidia hardware. CUDA has an interface that is reasonably well-designed, well-documented/reverse-engineered, and battle-tested for decades. What we need is not to invent another interface just under the name of 'open standard', but to implement the same interface. ROCm is exactly doing this, and so are other hardware SDKs such as MooreThread and Alibaba T-Head.

Someone needs to stand up a benchmark suite for ROCM, this, and everyone else attempting it to really get the ball rolling here. SemiAnalysis could have a blast with this.

A YouTuber automated ROCm version testing for Strix Halo/R9700 + LLM workloads:

https://kyuz0.github.io/amd-strix-halo-toolboxes/

https://kyuz0.github.io/amd-r9700-ai-toolboxes/

Re: Alternative(s) to run CUDA on non-Nvidia hardware

Re: Alternative(s) to run CUDA on non-Nvidia hardware