>As you can see from the table, at 16-bit floating point precision, Gaudi 3 is neck and neck with where Nvidia’s Blackwell B100 will be later this year when it ships. At FP8, however, Blackwell will have the advantage, and Blackwell also supports FP4, which Gaudi 3 does not.
>If you add in support, power, environmental, and management costs, which are the same, then the gap between Nvidia and Intel starts to get a little smaller, but clearly Intel can argue some pretty impressive price/performance advantages at certain precisions.
Intel can argue good performance with huge asterisk.
As the article points out, Intel volumes are small
Not enough volumes to put hard price pressure against Nvidia. Gaudi sales can't fill even one of those new 100k GPU monster datacenters.
When the whole baseboard+memory+GPU system costs $300 - $400k and runs 4-6 years, configuring and finetuning the system for new models, algorithms and libraries that come along the way is a big part of the deal.
>If you add in support, power, environmental, and management costs, which are the same, then the gap between Nvidia and Intel starts to get a little smaller, but clearly Intel can argue some pretty impressive price/performance advantages at certain precisions.
Intel can argue good performance with huge asterisk.
As the article points out, Intel volumes are small
2024 sales:
Not enough volumes to put hard price pressure against Nvidia. Gaudi sales can't fill even one of those new 100k GPU monster datacenters.When the whole baseboard+memory+GPU system costs $300 - $400k and runs 4-6 years, configuring and finetuning the system for new models, algorithms and libraries that come along the way is a big part of the deal.