Skip to content

Conversation

sarnex
Copy link
Contributor

@sarnex sarnex commented Dec 4, 2024

igc-dev is using libigc.so.2 and released igc is using libigc.so.1, and there's no symlink to libigc.so in the dev-igc packages, so we need to do it ourselves.

Also, fix the workflow so it runs on changes to the driver install script and XFAIL failing tests.

And manually update the igc-dev version since the one we are using here is too old and the artifacts are expired so we can't build the image. The new version also contains an important fix for the joint matrix team.

Closes: #16088
Closes: #16251

@sarnex sarnex changed the title [CI] Fix dev-igc Docker container [CI] Fix igc-dev Docker container and bump version Dec 16, 2024
@sarnex sarnex marked this pull request as ready for review December 16, 2024 21:57
@sarnex sarnex requested a review from a team as a code owner December 16, 2024 21:57
Copy link
Contributor

@jsji jsji left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM. Thanks.

Copy link
Contributor

@uditagarwal97 uditagarwal97 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@YixingZhang007
Copy link
Contributor

LGTM, and thanks for helping resolve the issue! The tests joint_matrix_bf16_fill_k_cache_arg_dim.cpp and joint_matrix_bf16_fill_k_cache_runtime_dim.cpp pass on PVC but fail on the DG2 machine. I believe this is a bug in IGC rather than the CI, so XFAIL: gpu-intel-dg2 is added to these tests.

// Waiting for the commit in IGC to be pulled into the driver to resolve the
// test.
// XFAIL: gpu
// XFAIL: !igc-dev || gpu-intel-dg2
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@YixingZhang007 , XFAIL-TRACKER jira pointed below is submitted against PVC issue, that you fixed. If test is passing on PVC and is failing on DG2, it should be different jira, since problem is different. Could you please update xfail tracker?

Copy link
Contributor

@YixingZhang007 YixingZhang007 Dec 16, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For sure! I will create a new Jira issue regarding this and update the tracker :)

@sarnex sarnex requested a review from YuriPlyakhin December 17, 2024 00:24
Copy link
Contributor

@YuriPlyakhin YuriPlyakhin left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

approved, assuming Yixing will update the xfail tracker in the next commit.

@sarnex sarnex merged commit da16cd5 into intel:sycl Dec 17, 2024
22 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

5 participants