Draft: Introduce Alder Lake coverage
We have 12 acer-cbv514-1h-34uz-brya devices with an Alder Lake GPU available. Having as a reference the Tiger Lake coverage (acer-cp514-2h-{1130g7,1160g7}-volteer, 10 device each), this proposes to extend the coverage to this device type.
| TGL job | sharding | nightly | device | ADL job | sharding | nightly |
|---|---|---|---|---|---|---|
anv-tgl{,-full} |
5 | 4 | 1160g7 | => anv-adl{,-full}
|
6 | 5 |
anv-tgl-angle{,-full} |
1 | 1 | 1130g7 | => anv-adl-angle{,-full}
|
1 | 1 |
intel-tgl-skqp |
1 | 0 | 1130g7 | => intel-adl-skqp
|
1 | 0 |
zink-anv-tgl{,-full} |
1 | 3 | 1160g7 | => zink-anv-adl{,full}
|
1 | 2 |
zink-anv-tgl-traces |
1 | 0 | 1160g7 | => zink-anv-adl-traces
|
- | 0 |
In a nightly pipeline, TGL uses 7 1160g7 and 1 1130g7. In the other pipelines 7 1160g7 and 2 1130g7. The current plan for 12 brya boards is to use 8 in the nightly, and 10 in the other pipelines (which seems too much, and will produce a bottle neck that even @marge-bot could suffer if some people is testing something).
TODO: during the evolution of this MR, other jobs were proposed: iris-adl-{egl,piglit} and iris-adl-traces. With the current proposal, we have a high usage of the boards to be reduced, instead of increased this those 3 jobs more.
TODO: sharding to distribute on 12 devices
TODO: prepare the zink-anv-adl-traces job.
Depends on: !27185 (merged)