Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Cloud TPU VMs are generally available

cloud.google.com

Post reply on HN
Print view View original

41–50 of 53 posts

Re: Cloud TPU VMs are generally available

#41

Post by johndough »

echelon wrote: post #39

Do existing PyTorch models work out of the box with cloud TPU, or does it require some tweaking? Are there cost savings over traditional GPU workloads? When will GKE support TPU? And will there be preemptable instances?

Last time I checked, you had to change a few lines to use torch_xla.

If there are cost savings depends on your workload. If you are training huge models, it might be cost effective, but honestly most tasks I have seen would have been fine on a CPU.

The pricing page lists preemptible instances: https://cloud.google.com/tpu/pricing#single-device-pricing

Re: Cloud TPU VMs are generally available

#42

Post by jacquesm »

zak wrote: post #9

Founder of the Cloud TPU program here. If you'd like to experiment with TPU VMs for free and are willing to share your work with the world somehow (e.g. via publications or open-source projects), you can apply to participate in the TPU Research Cloud (TRC) program here: https://sites.research.google/trc/

Nice achievement and great to see you are still there after all that time to see it through to GA. Congrats! It must be a bit like seeing your child cycle to school alone for the first time :)

Re: Cloud TPU VMs are generally available

#43

Post by AnonMO »

mahastore wrote: post #4

First thing you should explain is what the hell a TPU is/

Google cloud is part of a company called alphabet which also own a little tool called google search you should try it. it’s very handy. Also if you don’t know what a tpu is the product probably is targeted at you

Re: Cloud TPU VMs are generally available

#44

Post by ChadNauseam »

zak wrote: post #7
mahastore wrote: post #4

First thing you should explain is what the hell a TPU is/

Here's a quick overview of TPUs from last year's Google I/O keynote: https://www.youtube.com/watch?v=XFFrahd05OM&t=1565s

The section on error-correcting qubits right after the bit you timestamped is really interesting too, I wonder what the status of that is

Re: Cloud TPU VMs are generally available

#46

Post by mirker »

zak wrote: post #9

Founder of the Cloud TPU program here. If you'd like to experiment with TPU VMs for free and are willing to share your work with the world somehow (e.g. via publications or open-source projects), you can apply to participate in the TPU Research Cloud (TRC) program here: https://sites.research.google/trc/

What are some of the things on the roadmap for the platform? Any immediate plans to close the command-line gap for TPU utilization, etc.?

My overall impression is TPUs are pretty awesome but the software stack is still a bit hard to use compared to mature GPU tools. I’d imagine it’s pretty hard for inexperienced users to use them.

Re: Cloud TPU VMs are generally available

#47

Post by Filligree »

zak wrote: post #32
Reubend wrote: post #31

Do these TPU VMs have preemptible/spot pricing? I'm just a hobbyist, but it always seemed annoying before that one would have to orchestrate VMs to spin up/down alongside spot instances of TPUs manually.

Yes, there are a couple of ways to use Cloud TPUs at lower priority and lower cost. If you are a hobbyist, I highly recommend trying out Cloud TPUs for free via the TPU Research Cloud: https://sites.research.google/trc/ If you are flexible, you (or your scripts) can access a lot of compute power at odd hours!

That’s for researchers, however. For hobbyists, I guess we just have colab?

Re: Cloud TPU VMs are generally available

#48

Post by zak »

Filligree wrote: post #47
zak wrote: post #32

Earlier quoted context omitted.

Yes, there are a couple of ways to use Cloud TPUs at lower priority and lower cost. If you are a hobbyist, I highly recommend trying out Cloud TPUs for free via the TPU Research Cloud: https://sites.research.google/trc/ If you are flexible, you (or your scripts) can access a lot of compute power at odd hours!

That’s for researchers, however. For hobbyists, I guess we just have colab?

I started the TRC program alongside the Cloud TPU program to make interesting amounts of ML compute available to a broad group of creative people, not only to academic researchers.

The TRC program welcomes hobbyists, artists, students, independent learners, technical writers, and a variety of others. We love it when the TPU Research Cloud enables people to do something that wouldn't have been possible otherwise.

I definitely recommend applying to the TRC program - please feel free to say directly that you are a hobbyist. The sign-up form is short, and it's likely that you can access a lot of compute if you are flexible and persistent.

Re: Cloud TPU VMs are generally available

#49

Post by zak »

jacquesm wrote: post #42
zak wrote: post #9

Founder of the Cloud TPU program here. If you'd like to experiment with TPU VMs for free and are willing to share your work with the world somehow (e.g. via publications or open-source projects), you can apply to participate in the TPU Research Cloud (TRC) program here: https://sites.research.google/trc/

Nice achievement and great to see you are still there after all that time to see it through to GA. Congrats! It must be a bit like seeing your child cycle to school alone for the first time :)

Thanks very much! We've come a long way, but there is always more interesting work required to keep up with the deep learning frontier and enable Cloud TPU customers and TRC users to expand it further.

Re: Cloud TPU VMs are generally available

#50

Post by zak »

rg111 wrote: post #45

Cloud TPUs are the ultimate form of vendor lock-in. I vehemently oppose using it for anything truly critical.

We try to make it easy to switch back and forth between Cloud TPUs and other hardware platforms using JAX, PyTorch, and TensorFlow. This is a difficult technical challenge, but the XLA compiler helps a lot, and switching is easier now than it has ever been.
Post reply on HN