3 ms·
This is a big part of AMD still not having a proper foothold in the space: AMD Instinct is quite different from what regular folks can easily put in their works
by KeplerBoy 8mo ago
This is a big part of AMD still not having a proper foothold in the space: AMD Instinct is quite different from what regular folks can easily put in their workstation. In Nvidia-land I can put anything from mid-range gaming cards, over a 5090 to an RTX 6000 Pro in my machine and be confident that my CUDA code will scale somewhat acceptably to a datacenter GPU.
- bigyabai 8mo agoThis is where I feel like Khronos could contribute, making a Compute Capability-equivalent hardware standard for vendors to implement. CUDA's versioning of hardware capabilities plays a huge role in clarifying the support matrix. ...but that requires buy-in from the rest of the industry, and it's doubtful FAANG is willing to thread that needle together. Nvidia's hedged bet against industry-wide cooperation is making Jensen the 21st century Mansa Musa.