[systems-assistant]

Bumps ROCm/rocm-systems from ROCm/rocm-systems@186be9a to ROCm/rocm-systems@c1ab6df.

See full comparison here: ROCm/rocm-systems@186be9a...c1ab6df

[therock-pr-bot]

🤖 Bump PR detected (author @systems-assistant[bot]). All policy checks are auto-approved for automated dependency bumps.

[amd-chiranjeevi]

Failed jobs

Job	Failure signature	Existing tracking
Windows gfx1151 — hipblaslt (1/1)	hipBLASLt W4A16: 44 failed cases; recurring from #9738	rocm-libraries#14872
Windows gfx1151 — hipdnn-cudnn-samples (1/1)	Timeout: 26 sample-execution tests exceed 300 seconds; 74/100 tests pass. Newly failing versus #9738	No exact tracking issue found
Windows gfx1151 — rocfft(*) Shards: 1, 2	Timeout: both Test actions exceed 60 minutes while individual accuracy tests continue progressing. Shard 2 recurs; shard 1 newly fails	Broad timeout tracking: TheRock#9318. No exact issue found for this gfx1151 standard-suite timeout
Linux gfx90a — hip-tests (1/4)	Unit_hipMalloc3DArray_MaxTexture - unsigned char timeout; recurring from #9738. Earlier Unit_hipGraphStreamPool_InstantiateFootprint assertion passes now	rocm-systems#13733; fix PR #14247
Linux gfx110X-all — rocprofiler-sdk (1/1)	hip-graph-bubbles-k256-i20 and k256-i200 validators fail: 99.9th-percentile dispatch gaps of 562.78/556.88 µs exceed the 50 µs threshold. Different from #9738 attachment failures; timing assertions, not timeouts	rocm-systems#10876 matches the i20 test and threshold. No confirmed exact open tracker found
Linux gfx110X-all — rocsparse(*) Shards: 1, 3	Shard 1: 26 standard-suite and 8 device-test failures. Shard 3: standard-suite timeout and 8 device-test failures. Pivot, solver, bitmap and preconditioner assertions; both jobs newly fail versus #9738	Related failure families: rocm-libraries#11297, rocm-libraries#14873. Exact gfx110X coverage unconfirmed
Linux gfx120X-all — rocalution (1/1)	Stall/timeout: no logged progress after lu_float/5, then Test action exceeds 30 minutes. Newly failing; different test from #9738's gfx1151 CR stall	No exact tracking issue found
Linux gfx120X-all — rocsparse (1/3)	Null-address GPU fault in csr2hyb_kernel, plus GPU fault in csrgemm_bitmap_mark_kernel; standard/device suites abort. Same fault family as #9738, in a different shard	Related tracking: rocm-libraries#11297, rocm-libraries#14873

Queued jobs

(*) groups all shards into one row. Earlier signatures below refer to the previously checked PR #9738 results, not completed failures in the current #9739 attempt. Queued jobs have no current result and no assigned runner.

Job	Earlier failure signature	Existing tracking
Linux gfx1151 — hipblaslt(*) Shards: 1, 2, 3, 4, 5, 6	Previously all six shards failed hipBLASLt W4A16: NO solution found!	rocm-libraries#14872
Linux gfx1151 — rocsparse(*) Shards: 1, 2, 3	Previously triangular-solve/pivot/numerical failures, standard-suite timeout and device-suite GPU faults/aborts	rocm-libraries#11297, rocm-libraries#14873
Linux gfx1151 — rocprofiler-sdk (1/1)	Previously late-attach copy counts, HSA signal counts, zero counters/empty ATT output and Missing detailed profiling! aborts	rocm-systems#14454, TheRock#6476. New bump includes test fixes rocm-systems#14503 for copy-volume and signal-count validation; result pending
Linux gfx1151 — rocalution (1/1)	Previously inversion assertion, then no progress after cr_double/4 and 30-minute Test action timeout	rocm-libraries#14873 covers related solver assertions. No exact CR-stall issue found
Linux gfx1151 — hip-tests (3/4)	Previously passed in #9738	No failure signature to match to a tracking issue
Windows gfx110X-all — hipsparse (3/3)	Previously passed in #9738	No failure signature to match to a tracking issue

commit diff b/w #9738 and #9739 - ROCm/rocm-systems@4e65d8d...c1ab6df

[amd-chiranjeevi]

Asan is green

Tests on gfx950, gfx94x, win-gfx110x and all sanity checks are green.

updated the known issues on the other gfxs in above comment. ( not able to restart the timeouts, since queue is high > 4h for linux-gfx1151, win-gfx110x, will track those failures ( if genuine ) in nightly test_artifacts results )

[systems-assistant]

TheRock Submodule Bump Activity

No upstream PR found for these commits

The following 1 commit(s) were included in this bump but have no associated pull request on ROCm/rocm-systems:

• 0179365