I haven't seen to have any trouble getting a GPU, and almost always P100s.

[QUOTE=James Heinrich;534227]I haven't seen to have any trouble getting a GPU, and almost always P100s.[/QUOTE]

You never get the sign:

"Failed to assign backend.
No backend with GPU available. Would you like to use a runtime with no accelerator?"

With options "Cancel" and "Connect"? That is very lucky then, it happens all the time for me.

It's indeed harder and harder, I couldn't get a GPU today.
No problem with CPU sessions (mprime P-1)

[QUOTE=ATH;534233]You never get the sign:
"Failed to assign backend. No backend with GPU available. Would you like to use a runtime with no accelerator?"[/QUOTE]On rare occasion, maybe once every 3 days or so. But then I retry an hour or so later and usually no problem.

My personal experience the last few days is that I will get one GPU session, usually in the afternoon (PST), that lasts for an hour. Once I get a session and it expires/disconnects, I can not get another that day.

Still getting issues here :(

I've gotten maybe an hour's worth in the last 28 (x2 accounts, each on a different device).

To contribute to the empirical observations...

Yesterday 3/5 requests were satisfied, each for exactly 10 hours.

Thus far today again 3/5 successes, but only for a single hour each.

The GPU72_TF admin report is showing /no one/ is currently running...

Bummer... :sad:

Yesterday I tried 10-15 times during a period of about 12 hours until I got a GPU, but then I had it for over 6 hours, probably 10 hours or whatever the maximum is now.

I just tried reconnecting and succeeded on both accounts. So you should see me running now.

[QUOTE=James Heinrich;534289]So you should see me running now.[/QUOTE]

Yup... You and bayanne.

None of my requests were satisfied.

Edit: And you and bayanne both dropped off again...