[peppermint_gum]

AMD clearly believes that this newfangled "GPU compute" fad will pass soon, so there's no point to invest in it.

This is one of the worst acts of self-sabotage I have ever seen in the tech business.

[jart]

Zen4 AVX512 must be really good then.

[imtringued]

To be fair a lot of the GPU edge comes from fast memory. A GPU with 20tflops running a 30 billion parameter model has a compute budget of 700flops per parameter. Meanwhile the sheer size of the model prevents you from loading it more than 20 times from memory per second.