Not a lot of public data about things like that at scale, but there are some things like the Meta Llama report that had ~ 9% annualized gpu failure rate while training Llama 3. Not 100% clear from the Meta report if those are dead hardware or just 'unhappy' because they just put them into the 'interruption caused by' bucket.
Would be interested if anyone had better data on actual live failure rates of h100 or better class gpu that suck so much power.
Specifically I think this will be true around defence procurement: Canada was just added as the first non-EU state in the EU SAFE defence agreements, and just did a major booth at the International Defence Industry Exhibition EU defence trade show. They decided to buy new subs from Germany instead of Korea, and have slow-played the new fighter acquisitions and will potentially replace at least some F35 orders with Saab Gripens. And they have a new deal to start manufacturing Ukrainian designed military drones in Canada (partnership with Ukraine announced last week iirc?).
If getting this status helps simplify process and costs of EU<->Canada military procurement, then it will be a clear win for both sides given the current desire to reduce dependence on US defence providers.
It could be, but common, we all know that Anthropic's watermark is the using "load bearing", "genuine", and "seam" 1000x more in the same paragraph than any human in history.
Better yet, he gets loans, collateralized against those unrealized gains to the current value of the gains, and then writes off the loan interest against his present and future taxes! Amazing!
Not a lot of public data about things like that at scale, but there are some things like the Meta Llama report that had ~ 9% annualized gpu failure rate while training Llama 3. Not 100% clear from the Meta report if those are dead hardware or just 'unhappy' because they just put them into the 'interruption caused by' bucket.
Would be interested if anyone had better data on actual live failure rates of h100 or better class gpu that suck so much power.
https://ai.meta.com/research/publications/the-llama-3-herd-o... https://www.trendforce.com/news/2024/10/31/news-datacenter-g...