Hacker News Reader

>It seems likely that Microsoft and Google will introduce ARM-based instances.

Actually google was the first. Its called TPU - Tensor Processing Units and has been available for their ML platform for a long time. Interestingly, TPU replaces nvidia in this case.

There are already largescale Kaggle benchmarks on CPU vs GPU vs TPU - https://towardsdatascience.com/when-to-use-cpus-vs-gpus-vs-t...

View on HN

I'm pretty sure TPUs are not based on ARM.

I'm not claiming they are on ARM - and that in itself is a moot point.

I'm mentioning that Google Cloud primarily hosts and bills your ML code on TPUs. Today TPU v4 run at exascale.

so the equivalence "scared" behavior exists for both Intel and nvidia on Google already.

GCP cloud TPUs are comprised of multiple ASICS underneath. To say 1 TPU is faster than 1 GPU is moot as the Cloud TPU is a logical construct.

Instead $/ML training job or time to complete training given X budget is likely a better measure.

Reply on news.ycombinator.com