Re:Open source ChatGPT with LLaMA implementation trains 15x ...
by dragonwriter on Monday February 27, 2023 @01:39PM
(#34960215)
> Keep in mind that many of the GPU cores go unused, since they are dedicated to geometry or ray tracing or whatnot
I mean, “many” is not usually the case; an RTX 3090 has 10,496 CUDA cores, 328 Tensor cores, 82 RT (raytracing) cores, and 96 render output pipelines. ML apps will uses the first and, depending on software, second set. The vast majority of the cores being CUDA cores is the norm.
I mean, “many” is not usually the case; an RTX 3090 has 10,496 CUDA cores, 328 Tensor cores, 82 RT (raytracing) cores, and 96 render output pipelines. ML apps will uses the first and, depending on software, second set. The vast majority of the cores being CUDA cores is the norm.