Live data from Hacker News

Google Colab Pro is switching to compute credits

news.ycombinator.com

61–70 of 89 posts

Re: Google Colab Pro is switching to compute credits

#61
post #57

Earlier quoted context omitted.

Thanks a lot for collab, I have access to really powerful compute from my job yet I still find myself using collab a lot. The only problem I had with it was the very vague FAQ and the spotty resource allocations but the new faq is much better. Thanks again!

I hope it works better for you, please send feedback if it doesn't!

Colab is awesome, thank you.

I'll take the opportunity to request better editing ergonomics: the ability to connect from Jupyter/notebook-supporting editors and IDEs; the ability to open/edit .ipynb files from the local disk and/or Github without having to first put them on Google Drive.

Colab/Jupyter and friends are reinventing many wheels around editing code, and it would be nicer for them to support tools like Jupytext.

Re: Google Colab Pro is switching to compute credits

#62
Good. Hopefully this will reduce the randomness of type-of-GPU assignment on the Pro plan.

I fine-tuned a model on Colab Pro earlier this year and having to launch and quit 6 or 7 times to get a faster graphics card to ensure it completed within the time limit sucked.

Hope this will give more transparency into whether you are assigned a whole card or a virtual slice of one. Something I could never work out before!

Re: Google Colab Pro is switching to compute credits

#63
post #62

Good. Hopefully this will reduce the randomness of type-of-GPU assignment on the Pro plan. I fine-tuned a model on Colab Pro earlier this year and having to launch and quit 6 or 7 times to get a faster graphics card to ensure it completed within the time limit sucked. Hope this will give more transparency into whether you are assigned a whole card or a virtual slice of one. Something I could never work out before!

You're always getting the whole card, never a slice fwiw. We haven't found a GPU virtualization solution that has a strong enough security boundary between slices, so we keep you isolated.

And yes, hoping to give you more control over chip type too, stay tuned.

Re: Google Colab Pro is switching to compute credits

#64
post #57

Earlier quoted context omitted.

I hope it works better for you, please send feedback if it doesn't!

Colab is awesome, thank you. I'll take the opportunity to request better editing ergonomics: the ability to connect from Jupyter/notebook-supporting editors and IDEs; the ability to open/edit .ipynb files from the local disk and/or Github without having to first put them on Google Drive. Colab/Jupyter and friends are reinventing many wheels around editing code, and it would be nicer for them to support tools like Jup…

Will add to my list, thanks!

Re: Google Colab Pro is switching to compute credits

#65
At last! I love Colab but the vague promises around availability and quota made it impossible to recommend for my team to use professionally.

I even tried and failed to get it up and running with a Google cloud GPU recently, before just switching to Lambda which worked first time (but had since hit availability issues).

Re: Google Colab Pro is switching to compute credits

#66

I like the transparency, but this doesn't feel like the right way to do it. Computation should be free (or nearly free) if there's idle capacity, paid if Google is near capacity, and expensive/bidding if Google is above capacity. Flat compute units seem simple, but result in a lot of waste.

>Computation should be free (or nearly free) if there's idle capacity I guess nothing stops you from buying infra and offering it for "free (or nearly free)".

Agree. I just do not like restaurants which charge full price of burgers to me when there are clearly more patties on grill than customers at that time.

Re: Google Colab Pro is switching to compute credits

#67
post #63
post #62

Good. Hopefully this will reduce the randomness of type-of-GPU assignment on the Pro plan. I fine-tuned a model on Colab Pro earlier this year and having to launch and quit 6 or 7 times to get a faster graphics card to ensure it completed within the time limit sucked. Hope this will give more transparency into whether you are assigned a whole card or a virtual slice of one. Something I could never work out before!

You're always getting the whole card, never a slice fwiw. We haven't found a GPU virtualization solution that has a strong enough security boundary between slices, so we keep you isolated. And yes, hoping to give you more control over chip type too, stay tuned.

Thanks that's good to know.

I would have upgraded to Pro+ if I had confidence it would speed up the process, but the promises of what you get (beyond "runs in background") were/are so vague I couldn't tell if it was worth it. I was only using 50-80 hours a month and it sounded like a plan aimed at more usage.

Google was/is leaving money on the table with the old scheme.

Re: Google Colab Pro is switching to compute credits

#68
I really like the increase in transparency, I found it somewhat disturbing to pay for what feels like a random amount of stuff. How should I know if I need Pro or Pro+ if there is no estimate out there what either might get me. The update does not seem to change that though. I would love to have a distribution plotted of how much compute I might expect. Or at least Min/Average/Max run time until disconnect (rn. only Max is known).

Re: Google Colab Pro is switching to compute credits

#69
post #68

I really like the increase in transparency, I found it somewhat disturbing to pay for what feels like a random amount of stuff. How should I know if I need Pro or Pro+ if there is no estimate out there what either might get me. The update does not seem to change that though. I would love to have a distribution plotted of how much compute I might expect. Or at least Min/Average/Max run time until disconnect (rn. only…

They can benchmark a few architectures (ResNet50, BERT) and tell us how many times we can train a model on a specific level of subscription.

Re: Google Colab Pro is switching to compute credits

#70
post #59

Earlier quoted context omitted.

I've been using Colab for a few months as I dive into my personal ML education. Free at first, Pro tier now. It's very handy and I appreciate the product. I suppose my concern is this - Was Colab using "compute units" internally for limits already? Or was it "as available"? And if so, will my compute units allocation be what it has been or will it be decreasing? To be honest, when I read the email today, I assumed th…

Thanks! Colab has always had a notion of quota for all users. This notion was implemented in a hard to describe way, but we would throttle you as you used more compute without making it clear how or why. This change will make the quota allocation transparent and visible for paid users. Any limits we're using here are just to ensure you don't use more GPU $ than what you've paid for.

> Any limits we're using here are just to ensure you don't use more GPU $ than what you've paid for.

So naturally, any unused balance will be refunded, right Chris? You wouldn't want to start overcharging some users while fixing the "bug" or "loophole" that undercharged others.

Post reply on HN