Hacker News Reader

Given the worst-case assumptions of people arguing the cloud is required[1], it's a one-time investment of $60,000, $200,000 in the absolute worst case.

[1] https://news.ycombinator.com/item?id=24601264

View on HN

That's for model inference. For training OpenAI said they used around 3000 PetaFLOPS / days on the largest GPT-3 model. That translates to about 300 Nvidia A100 GPUs if you want training to finish in a month (any slower and your researchers are not going to be able to make much progress). A system like that would cost at least $5M, probably more like $10M.

Is that unit PetaFLOPS/day right? I think it should be PetaFLOPS-day which has dimensions of FLOP (total number of operations), rather than FLOP/time^2.

The cost wouldn't be the cost of the hardware because it still exists afterwards. You'd have to discount it for the amount of time it was in use.

Surely $10 million should be well within the spending abilities of a fair number of tech people? Many universities now have fairly large computing clusters as well.

People give more than this to get their lackluster children into the Ivy League.

Reply on news.ycombinator.com