Re: First experiments with gfx1100/gfx1101/gfx1102
Hi all,
On 2024-03-11 08:36, Christian Kastner wrote:
> I ran the first experiments with gfx1100 (via W7800), gfx1101 (via
> W7600), gfx1102 (via W7500) over the last few days.
Unfortunately, I needed to disable scheduling for gfx1100 and gfx1101 on
the prod CI. I also purged unfinished jobs.
The temporary host for these cards initially could handle the load, but
over the past few days queues kept growing, in part due to new failure
modes (eg: even rocrand was stuck for hours).
On top of that, I needed that host back, as it was my testbed for new CI
features, and I've got some work piled up that I need to test.
I'll see that I move these cards to another host over the weekend. They
will be connected to the test CI though, until failure modes are better
understood.
Best,
Christian
Reply to: