intel
[GHA] Uplift Linux IGC Dev RT version to igc-dev-f8ec694
@intel/dpcpp-devops-reviewers, it looks like no one really cares about igc-dev updates anymore. Can we disable GHA workflow for updating igc-dev?
Is it true that we don't care about igc-dev? @YuriPlyakhin @dkhaldi @jsji
If so I would really like to remove it from everything.
Is it true that we don't care about igc-dev? @YuriPlyakhin @dkhaldi @jsji
If so I would really like to remove it from everything.
The major user is joint-matrix team. If @YuriPlyakhin @dkhaldi don't care, then yes, we can remove it from pre-commit.
Is it true that we don't care about igc-dev? @YuriPlyakhin @dkhaldi @jsji If so I would really like to remove it from everything.
The major user is joint-matrix team. If @YuriPlyakhin @dkhaldi don't care, then yes, we can remove it from pre-commit.
I'm not currently actively working on Joint matrix, so I'll let @dkhaldi decide. It was useful to me, when I was actively adding JM features to IGC, and then had to wait for months before I saw tests passing in SYCLOS with release drivers...
What I see is that no one from @intel/sycl-matrix-reviewers, @intel/dpcpp-esimd-reviewers and @intel/dpcpp-devops-reviewers groups look at the CI failures in these PRs.
GHA script creates PRs each week trying to update the driver version, but AFAIK @YuriPlyakhin is the only beneficiary of this script. If he is not going to analyze CI failures, I suggest we use stable drivers for JM in CI and drop this workflow.
What I see is that no one from @intel/sycl-matrix-reviewers, @intel/dpcpp-esimd-reviewers and @intel/dpcpp-devops-reviewers groups look at the CI failures in these PRs.
GHA script creates PRs each week trying to update the driver version, but AFAIK @YuriPlyakhin is the only beneficiary of this script. If he is not going to analyze CI failures, I suggest we use stable drivers for JM in CI and drop this workflow.
As I wrote, I'm not currently using it. Also, I'm not planning to analyze CI failures for this workflow. I recommend you checking with @dkhaldi, regarding, if someone wants to take use/take it. Anyway, if you want my answer, I don't object regarding dropping this workflow.
I don't look at failures because I was told by IGC they won't investigate or fix issues until they are in shipped IGC releases
What I see is that no one from @intel/sycl-matrix-reviewers, @intel/dpcpp-esimd-reviewers and @intel/dpcpp-devops-reviewers groups look at the CI failures in these PRs. GHA script creates PRs each week trying to update the driver version, but AFAIK @YuriPlyakhin is the only beneficiary of this script. If he is not going to analyze CI failures, I suggest we use stable drivers for JM in CI and drop this workflow.
As I wrote, I'm not currently using it. Also, I'm not planning to analyze CI failures for this workflow. I recommend you checking with @dkhaldi, regarding, if someone wants to take use/take it. Anyway, if you want my answer, I don't object regarding dropping this workflow.
@dkhaldi, do you know if anyone is using IGC dev drivers and willing to handle issues uncovered by our CI?
What I see is that no one from @intel/sycl-matrix-reviewers, @intel/dpcpp-esimd-reviewers and @intel/dpcpp-devops-reviewers groups look at the CI failures in these PRs. GHA script creates PRs each week trying to update the driver version, but AFAIK @YuriPlyakhin is the only beneficiary of this script. If he is not going to analyze CI failures, I suggest we use stable drivers for JM in CI and drop this workflow.
As I wrote, I'm not currently using it. Also, I'm not planning to analyze CI failures for this workflow. I recommend you checking with @dkhaldi, regarding, if someone wants to take use/take it. Anyway, if you want my answer, I don't object regarding dropping this workflow.
@dkhaldi, do you know if anyone is using IGC dev drivers and willing to handle issues uncovered by our CI?
Yes, I look at CI failures in these PRs but ONLY for SYCL joint matrix tests. If there are failures, we set them to xfail and make sure we have a jira for them. If there are xpasses, it means the fix made it to IGC-dev so we remove the corresponding xfail.
There are often other failures and the IGC team won't investigate them until they're in a release. How would you feel it we changed dev-igc testing here to only run joint matrix tests where you can investigate any failures? @dkhaldi
There are often other failures and the IGC team won't investigate them until they're in a release. How would you feel it we changed dev-igc testing here to only run joint matrix tests where you can investigate any failures? @dkhaldi
I am okay with keeping dev-igc for joint matrix tests only if other components are fine without it.
Thanks.
Does the above sound okay with you guys as well @bader @aelovikov-intel?
Thanks.
Does the above sound okay with you guys as well @bader @aelovikov-intel?
I'm fine assuming that @dkhaldi commits to address CI issues.