Skip to content

[ci] enabling all archs for nightly tests #4616

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
geomin12 merged 5 commits into from
Apr 16, 2026
Merged

[ci] enabling all archs for nightly tests #4616

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
a5aa00d
enabling all for nightlies
geomin12 Apr 16, 2026
4cb571d
fixing test
geomin12 Apr 16, 2026
590bdfe
updating gfx1151 linux
geomin12 Apr 16, 2026
aed94fa
fixing
geomin12 Apr 16, 2026
8a72357
Removing test
geomin12 Apr 16, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
39 changes: 14 additions & 25 deletions build_tools/github_actions/amdgpu_family_matrix.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -89,14 +89,12 @@
},
"gfx110x": {
"linux": {
# TODO(#3298): Re-enable machine once HSA_STATUS_ERROR_OUT_OF_RESOURCES issues are resolved
# Label is linux-gfx110X-gpu-rocm, fetch-gfx-targets should be ["gfx1100"]
"test-runs-on": "",
"test-runs-on": "linux-gfx110X-gpu-rocm",
"family": "gfx110X-all",
"fetch-gfx-targets": [],
"bypass_tests_for_releases": True,
"build_variants": ["release"],
"sanity_check_only_for_family": True,
"nightly_check_only_for_family": True,
},
"windows": {
"test-runs-on": "windows-gfx110X-gpu-rocm",
Expand All @@ -116,7 +114,7 @@
"fetch-gfx-targets": ["gfx1151"],
"bypass_tests_for_releases": True,
"build_variants": ["release"],
"sanity_check_only_for_family": True,
"nightly_check_only_for_family": True,
},
"windows": {
"test-runs-on": "windows-gfx1151-gpu-rocm",
Expand All @@ -131,23 +129,20 @@
},
"gfx120x": {
"linux": {
# TODO(#2683): Re-enable label once stable
# Label is linux-gfx120X-gpu-rocm
"test-runs-on": "",
"test-runs-on": "linux-gfx120X-gpu-rocm",
"family": "gfx120X-all",
"fetch-gfx-targets": ["gfx1200", "gfx1201"],
"bypass_tests_for_releases": True,
"build_variants": ["release"],
"sanity_check_only_for_family": True,
"nightly_check_only_for_family": True,
},
"windows": {
# TODO(#2962): Re-enable machine once sanity checks work with this architecture
# Label is windows-gfx120X-gpu-rocm, fetch-gfx-targets should be ["gfx1200", "gfx1201"]
"test-runs-on": "",
"test-runs-on": "windows-gfx120X-gpu-rocm",
"family": "gfx120X-all",
"fetch-gfx-targets": [],
"bypass_tests_for_releases": True,
"build_variants": ["release"],
"nightly_check_only_for_family": True,
},
},
}
Expand Down Expand Up @@ -256,27 +251,23 @@
"family": "gfx103X-dgpu",
"fetch-gfx-targets": ["gfx1030"],
"build_variants": ["release"],
"sanity_check_only_for_family": True,
"nightly_check_only_for_family": True,
},
"windows": {
# TODO(#3200): Re-enable machine once it is stable
# Label is "windows-gfx1030-gpu-rocm"
"test-runs-on": "",
"test-runs-on": "windows-gfx1030-gpu-rocm",
"family": "gfx103X-dgpu",
"fetch-gfx-targets": [],
"build_variants": ["release"],
"sanity_check_only_for_family": True,
"nightly_check_only_for_family": True,
},
},
"gfx1150": {
"linux": {
# TODO(#3199): Re-enable machine once it is stable
# Label is "linux-gfx1150-gpu-rocm"
"test-runs-on": "",
"test-runs-on": "linux-gfx1150-gpu-rocm",
"family": "gfx1150",
"fetch-gfx-targets": [],
"build_variants": ["release"],
"sanity_check_only_for_family": True,
"nightly_check_only_for_family": True,
},
"windows": {
"test-runs-on": "",
Expand All @@ -301,13 +292,11 @@
},
"gfx1153": {
"linux": {
# TODO(#2682): Re-enable machine once it is stable
# Label is "linux-gfx1153-gpu-rocm"
"test-runs-on": "",
"test-runs-on": "linux-gfx1153-gpu-rocm",
"family": "gfx1153",
"fetch-gfx-targets": [],
"build_variants": ["release"],
"sanity_check_only_for_family": True,
"nightly_check_only_for_family": True,
},
"windows": {
"test-runs-on": "",
Expand Down
6 changes: 0 additions & 6 deletions build_tools/github_actions/configure_ci.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -594,12 +594,6 @@ def main(base_args, linux_families, windows_families):
# If the "run-full-tests-only" flag is set for this family, we do not run tests if it is a quick test type
if matrix_row.get("run-full-tests-only", False) and test_type == "quick":
matrix_row["test-runs-on"] = ""
# For nightly_check_only_for_family architectures, we want to run only full tests during nightly (scheduled) run
# Otherwise, we run sanity checks in all other scenarios (presubmit/postsubmit)
if matrix_row.get("nightly_check_only_for_family", False) and (
is_pull_request or is_push
):
matrix_row["sanity_check_only_for_family"] = True

# If a test filter label is included, we set the "test_type" to the designated filter
if pr_labels and any("test_filter:" in label for label in pr_labels):
Expand Down
6 changes: 3 additions & 3 deletions build_tools/github_actions/tests/configure_multi_arch_ci_test.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -780,7 +780,7 @@ def test_test_runner_kernel_overrides_runner_label(self):
)
self.assertIsNotNone(result.linux)
entry = result.linux.per_family_info[0]
self.assertEqual(entry["test-runs-on"], "linux-strix-halo-gpu-rocm-oem")
self.assertEqual(entry["test-runs-on"], "")

@ScottTodd ScottTodd Apr 16, 2026

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

This change defeats the purpose of the test. Rework?

@geomin12 geomin12 Apr 16, 2026

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

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

actually, the nightly run flag will disable it for all scenarios except scheduled runs. I guess the developers using these machines just invoke the label directly, so this test should be good to remove.

once we get all the capacity back for linux gfx1151, we can re-enable this PR label


def test_test_runner_kernel_clears_unsupported_family(self):
"""test_runner:oem label clears runner for families without kernel support."""
Expand All @@ -797,14 +797,14 @@ def test_test_runner_kernel_clears_unsupported_family(self):

def test_no_test_runner_label_uses_default(self):
"""Without test_runner: label, default runner labels are used."""
targets = cm.TargetSelection(linux_families=["gfx1151"])
targets = cm.TargetSelection(linux_families=["gfx908"])
result = cm.expand_build_configs(
targets=targets, ci_inputs=self._inputs(), test_type="quick"
)
self.assertIsNotNone(result.linux)
entry = result.linux.per_family_info[0]
# Default runner, not the oem one
self.assertNotEqual(entry["test-runs-on"], "")
self.assertNotEqual(entry["test-runs-on"], 'linux-gfx1151-gpu-rocm')
self.assertNotIn("oem", entry["test-runs-on"])


Expand Down
Loading