2 ms·
The reason is AMD and Nvidia don't is that they don't want to cannibalize their high end AI market. Intel doesn't have a high end AI market to protect.
by bryanlarsen 2y ago
The reason is AMD and Nvidia don't is that they don't want to cannibalize their high end AI market. Intel doesn't have a high end AI market to protect.
- fweimer 2y agoThere are products like this one: https://www.intel.com/content/www/us/en/products/sku/232592/intel-xeon-cpu-max-9480-processor-112-5m-cache-1-90-ghz/specifications.html https://www.intel.com/content/www/us/en/products/sku/232592/... As far as I understand it, it gives you 64 GiB of HBM per socket.
- knowitnone 2y agothat's a CPU. We are talking about GPUs here for highly parallel matrix multiplication problems. 2 different beasts.
- fweimer 2y agoThis CPU is a bit of a hybrid because it as integrated matrix multiplication accelerators: https://en.wikipedia.org/wiki/Advanced_Matrix_Extensions https://en.wikipedia.org/wiki/Advanced_Matrix_Extensions