vecchio

Intel Gaudi 2 is supposedly good but is mysterious to me, and Ponte Vecchio has barely started shipping.

From my understanding, Gaudi specializes in a specific use case (deep learning/AI) while Ponte Vecchio is more generic HPC.

If Ponte Vecchio is successful and doesn't end up a dead-end then little doubt its performance/latency will improve with time.

The cool thing about it is architecturally speaking it is basically a mini version of the Ponte Vecchio (PVC) datacenter GPU.

Note that this is about Intel Ponte Vecchio professional GPUs, not the consumer Alchemist GPUs.

I still can't fathom how Intel thought Arc and/or Ponte Vecchio would pay for themselves on day one.

I imagine if it had been retained, tourism UK would be calling it the Ponte Vecchio of the north.

It's good risk reduction, especially since Ponte Vecchio failed.

Does Microsft use Gaudi 2 or Ponte Vecchio much?

- Ponte Vecchio surprised us with its performance on our custom face swap model, beating everyone including the mighty H100.

> With that in mind, Ponte Vecchio is better seen as a learning experience.

> I still can't fathom how Intel thought Arc and/or Ponte Vecchio would pay for themselves on day one.

Check out Epyc, Ponte Vecchio, all high end GPUs and ML accelerators using HBM memory.

It takes 12 years to make the best, aceto balsamico tradizionale (traditional balsamic vinegar), and at least 25 to make the finest, extra Vecchio.

Like Ponte Vecchio, very interesting tech, and something the hyperscalers and supercomputer can be excited about, but it's hard to see a path towards long term success.