Folding Forum

They denied even one T4 this morning.
Shall I try another region or how can I get one?

Ichbin3 wrote:They denied even one T4 this morning.
Shall I try another region or how can I get one?

Yeah, just get another region.
At times, I had 5 or 6 units, 2 per region, because they kept throttling one (I figured, at least the other one will work).

Ichbin3 Central and East would be my top choices, then West after that b/c West seems to get kicked off more than the others. All other regions after this will be more expensive.

Meelee why do u continue to use throttle-ing services? I know they're free, but do they even finish a WU without expiring? I think not

Knish wrote:Ichbin3 Central and East would be my top choices, then West after that b/c West seems to get kicked off more than the others.

No way to get a T4. Even no preemptible. Nowhere ...
P100 preeemtible are available, but expensive, Preemptible NVIDIA P4 GPUs are there too but I couldn't find the output of this for fah.

Cloud folders in particular should be aware that there is a vulnerability in the FAHControl GUI before version 7.6.20 that could allow your cloud instances to execute code on your GUI machine: viewtopic.php?f=108&t=36471

Ichbin3 wrote:Preemptible NVIDIA P4 GPUs are there too but I couldn't find the output of this for fah.

I'm not clear what you mean by "output." If you're looking for log entries, GPU entries are anything with "x22" i.e. Actually, let me back up where you said "They denied even one GPU" (sorry for going over things you already know, i just want to be thorough)
I take it as you're on steps for "Req Quota Increase." In order to be able to request, you must have upgraded to the paid tier. Then, under IAM and Quotas, the line item to request is "GPU (all regions)" and if that's where you got an email denying you, then I'm not sure what else you can do. Please give me feedback on how the GCP guide was for that part in bringing up the quota edit for GPU all regions because I'm looking at it now and unsure if it's different for me bc i've got older accounts. I wrote "just type in GPU" and select GPU all regions, but wondering if it's better to maybe type in "regions"?

Anyway if you get past that part, then when you pick your GPU not every subregion has all GPUs, so if you want a T4 in us east, then pick either east1-c or east1-d. But a P100 would be either east1-b or east1-c. A P4 are in these regions: central1-a/c, east-4a/b/c, and west2-b/c. But with the 90 day limit GCP now imposes before expiring your credits you'd need multiple P4's going and i've never been approved more than 1 GPU.

thx, gunnarre!

Knish wrote:you must have upgraded to the paid tier

Sure i did.

Knish wrote:I wrote "just type in GPU" and select GPU all regions, but wondering if it's better to maybe type in "regions"?

I also had to hit 'enter'. Before, after typing 'gpu' I did choose T4 and then requested in the region. My fault.
Now, after hiting enter, I requested any gpu for 'all regions', but the result is the same - denied.

Knish wrote: I'm not clear what you mean by "output."

I was looking for the PPD of the P4.

Ichbin3 wrote: I was looking for the PPD of the P4.

~300k ppd

Don't know why you're getting denied, sorry about that. maybe you can make multiple CPU instances if u feel like still pressing forward. the default quota is a sum of v32 in all regions, but a max of v24 in any 1 particular region, so u could make a 16core in central and another in east, for example. A 16core is ~100k ppd and its rate is < $0.11 per hour. Those two instances will use up the credits in about 50 days. Cheapest way to create these instances are to make series: E2, machine type: custom, cores:16, memory: 8. Don't forget to make them preemptible. Then FAH installation is even easier bc there's none of the file editting necessary since no need to set up a GPU. Still need to reboot the instance tho if u set up for remote monitoring. Remember to have your fahclient at home running 7.6.21 per the security announcement.

In Azure, the rates for P100 have gone up a bit while the rates for V100 have come down to where they're almost equal, so it makes sense to use the V100 given its ~4-6mil ppd. I haven't written that into the github b/c it seems to over complicate things and I don't know how long that's gonna last. But the details of doing so (at least what's different to set up the V100 over the P100) are written here: https://www.reddit.com/r/Folding/commen ... h_to_v100/

Have got this mostly up and running. I would suggest in GCP looking up "region" instead of "GPU" to get "All Regions".
I cannot get FAH starting, or connect to it via my desktop app. I did install version 7.6.21 instead though as that is the latest.
Should I install 7.6.13 instead? Any other ideas on what I could do?

thx; i'll make the appropriate changes. I like v .13 as per the guide b/c it looks a little bit easier to edit the config file as per the guide to get a GPU going in GCP. Ver .21 looked like it added some stuff.

On your home PC u should definitely have ver .21 installed.

I don't know what I'm doing wrong, got it denied again.

ichbin - ptchern is saying it's better to type in "regions" instead of GPU when you go to request a quota increase of 1 GPU. the line name for it that you click on should say "GPUs (all regions)"

... and after that if you're still getting an email from GCP saying denied, I don't really know what else you can do. Maybe you just got hit with bad luck

Knish wrote:ichbin - ptchern is saying it's better to type in "regions" instead of GPU when you go to request a quota increase of 1 GPU. the line name for it that you click on should say "GPUs (all regions)"

Did that. Still ...

Knish wrote:

Ichbin3 wrote: I was looking for the PPD of the P4.

~300k ppd

300k PPD for the Pi4?
Last time I ran a 6-core I5, it ran well below 100k PPD.
What sorcery is this?