Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

The likelihood intel AI was going to catch up with efforts like AWS Trainium, let alone Nvidia was already vanishingly small. This gives intel a chance at maintaining leading edge fab technologies.

I feel bad for gamers - I’ve been considering buying a B580 - but honestly the consumer welfare of that market is a complete sidenote.



I don’t agree. OneAPI gets a lot of things right that ROCM doesn’t, simply because ROCM is a 1:1 rip of what nvidia provides (warts and historical baggage included) whereas OneAPI was thoughtfully designed and did away with all of that. Intel has a strong history in networking, much stronger than Xilinx/AMD, and really was the best hope we had for an open standard to replace nvidia’s hellscape.


> This gives intel a chance at maintaining leading edge fab technologies.

I don't think so:

> The chip giant hasn’t disclosed whether it will use Intel Foundry to produce any of these products yet.

It seems pretty likely this is an x86 licensing strategy for nvidia. I doubt they're going to be manufacturing anything on intel fabs. I even wonder if this is a play to get an in with Trump by "supporting" his nationalizing intel strategy.


nvidia doesn’t need x86, they’re moving forward on aarch64 and won’t look back. For example, one of the headlines from CUDA 13 is that sbsa can be targeted from all toolkits, not as a separate download, which is important for making it easy to target grace. They have c2c silicon on grace for native host side nvlink. They’re not looking back.


They're clearly looking back though, investing in Intel and announcing quite substantial partnerships. Maybe they're not looking back for technical reasons, but they are looking back.


I don't think Nvidia cares about the CPU ISA.


I think literally just the cash is a big deal at this point. Additionally, this deal probably increases the chances that Nvidia at least uses some Intel Foundry technology (like packing) and maybe very down the road, fabrication.


> The likelihood intel AI was going to catch up with efforts like AWS Trainium, let alone Nvidia

...and yet Nvidia is not gambling with the odds. Intel could have challenged Nvidia on performance-per-dollar or per watt, even if they failed to match performance in absolute terms (see AMD's Zen 1 vs Intel)


You misunderstand enterprise GPUs. Their cost of ownership is dominated by electricity - they're already optimized for price per watt.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: