News for hackers. Stuff that matters.
Re:How Meta trains large language models at scale by dinobones on Wednesday June 12, 2024 @09:34PM (#40665036)
Do you really think Google’s hardware expertise is better than Nvidia’s?

If needed these other companies have the $$$ to buy the best chips money can buy from Nvidia. Better chips than Google could ever produce.

If anything, this is why IMO Google will fail.

Re:How Meta trains large language models at scale by candiddevmike on Wednesday June 12, 2024 @09:54PM (#40665163)
I thought NVIDIA's moat was mostly software/CUDA?
Re:How Meta trains large language models at scale by rapsey on Wednesday June 12, 2024 @11:28PM (#40665643)
They have by far the fastest chip, the best software with CUDA and feverishly working on next gen chips. They also bought out multiple years worth of global high bandwidth memory manufacturing capacity.

No one will beat them at their game. However if there are any major breakthroughs that might render those processing capacities unneeded, or the major players hitting a wall regarding AI spending, then they will take a massive hit. It will come eventually because the chip business is always in boom/bust cycles.

Re:How Meta trains large language models at scale by mike_d on Wednesday June 12, 2024 @09:55PM (#40665171)
Yes. Google was building custom HPC hardware 5-8 years before Nvidia decided to expand outside the consumer and "workstation" markets.
Re:How Meta trains large language models at scale by threeseed on Wednesday June 12, 2024 @10:42PM (#40665400)
Nvidia acquired Mellanox who know far more about custom HPC hardware than Google.
Re:How Meta trains large language models at scale by mike_d on Thursday June 13, 2024 @04:21AM (#40667214)
Mellanox and friends not being able to build fast enough switching gear at a reasonable price point was what got Google into the hardware game in the first place ;)