4 ms·
I believe “world models” are the future of the field so I really need better performance in areas like IVP, FFT, special functions (e.g., harmonics), and dynami
by Q6T46nT668w6i3m 3y ago
I believe “world models” are the future of the field so I really need better performance in areas like IVP, FFT, special functions (e.g., harmonics), and dynamic programming. The H100 advances (e.g., DPX instructions) are terrific but they feel like a starting point. Hell, improved geometric operations (e.g., triangulation and intersection) would be killer too and surely that expertise exists at NVIDIA! The H100, especially for the price, feels terrible when you’re training a neural network bottlenecked on an operation that flies on a consumer CPU when you know there’s GPU optimizations that have been left on the floor.
- tgtweak 3y agoI suspect these can be patched in as well - most of these functions have implementations in CUDA implying they should be able to run on the hardware even without dedicated instructions.
- rbanffy 3y agoThis was one of the selling points for RISC: to focus on the most frequently used instructions and implement the rest in software would yield smaller and faster designs. I have the feeling that GPUs are on the sweet spot where smaller footprint directly translates into more executions and, therefore, higher throughput for the same chip area.