3 ms·
Yeah, I've seen pod slices allocate 7TB. A TPUv3 pod is actually a bunch of individual TPUv3-8's linked together. There's 8 cores per device, so a TPUv3-512 ha
by sillysaurusx 7y ago
Yeah, I've seen pod slices allocate 7TB.
A TPUv3 pod is actually a bunch of individual TPUv3-8's linked together. There's 8 cores per device, so a TPUv3-512 has 512 cores divided by 8 cores per device = 64 individual TPUs. (You can get each individual TPU's IP address using `gcloud compute tpus list`: https://imgur.com/Qym4l17 https://imgur.com/Qym4l17)
The big question is, since there are 64 individual TPUs, does that mean we have access to 300GB * 64 = 19.2 TB of memory?
I haven't tested that, but I would bet the answer is yes, for two reasons. 1. I've seen allocations of up to 7TB according to memory usage logs, so 19TB doesn't seem far fetched in comparison. 2. If you create 64 individual TPUv3-8's, then you definitely will have access to 300GB of memory on each TPU, so it's the same engineering problem either way.
Right now, people only seem to use the TPU's CPU for infeed processing / input pipeline transformations. But the CPU is quite fast – it's almost as fast as an actual TPU core.
I wrote up some more about this in a tweet chain if you're interested: https://twitter.com/theshawwn/status/1223395022814339073 https://twitter.com/theshawwn/status/1223395022814339073
Also, if you want to play around with a few TPUv3-8's and you have a GCE project, feel free to DM me on twitter. We just figured out how to forward TPUs to VMs in different projects: https://twitter.com/theshawwn/status/1221241517626445826 https://twitter.com/theshawwn/status/1221241517626445826
Is there an official specification clarifying this somewhere?
Not that I've seen. I stumbled across it by accident. https://twitter.com/theshawwn/status/1163799288771698688 https://twitter.com/theshawwn/status/1163799288771698688
- nil-sec 7y agoSo you are saying the system memory is 300GB and you can train your model on the cpu instead? Well yeah you can always do that but training will be slow because your model is not trained on the GPU. What’s the point?
- sillysaurusx 7y agoIt's not that slow. And you can use many TPUs together to make up the speed difference.
- nil-sec 7y agoIf that were the case I am wondering why anyone would buy GPUs? I invite you to retrain a state of the art model of your choice on a CPU and see how far you get.
- sillysaurusx 7y agoWe fine-tuned GPT-2 1.5B for subreddit simulator using this technique. https://www.reddit.com/r/SubSimulatorGPT2Meta/comments/entfgx/update_upgrading_to_15b_gpt2_and_adding_22_new/ https://www.reddit.com/r/SubSimulatorGPT2Meta/comments/entfg...