pytorch - 💡(How to fix) Fix AArch64 OSDC G3 Runner incorrectly reports cpu capability SVE128

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…

Code Example

2026-05-18T05:03:42.8707292Z Current runner version: '2.334.0'
2026-05-18T05:03:42.8712614Z Runner name: 'mt-l-arm64g3-16-62-6cxd4-runner-wcjdh'
2026-05-18T05:03:42.8713383Z Runner group name: 'default'
2026-05-18T05:03:42.8714281Z Machine name: 'mt-l-arm64g3-16-62-6cxd4-runner-wcjdh'
...
2026-05-18T05:04:00.3749597Z Complete job name: linux-jammy-aarch64-py3.10 / test-osdc (default, 2, 5, mt-l-arm64g3-16-62)
...
2026-05-18T05:06:22.8815689Z PyTorch built with:
2026-05-18T05:06:22.8816026Z   - GCC 13.4
2026-05-18T05:06:22.8816264Z   - C++ Version: 202002
2026-05-18T05:06:22.8816634Z   - Intel(R) MKL-DNN v3.11.2 (Git Hash 03c022d3ffdcee958cfacbe720048e725fdf644c)
2026-05-18T05:06:22.8817013Z   - OpenMP 201511 (a.k.a. OpenMP 4.5)
2026-05-18T05:06:22.8817333Z   - LAPACK is enabled (usually provided by MKL)
2026-05-18T05:06:22.8817605Z   - NNPACK is enabled
2026-05-18T05:06:22.8817837Z   - CPU capability usage: SVE128

---

PyTorch built with:
  - CPU capability usage <CPU_CAPABILITY>
RAW_BUFFERClick to expand / collapse

🐛 Describe the bug

In debugging https://github.com/pytorch/pytorch/issues/184212 I have noticed that the Graviton 3 OSDC runner is incorrectly reporting cpu capability SVE128

2026-05-18T05:03:42.8707292Z Current runner version: '2.334.0'
2026-05-18T05:03:42.8712614Z Runner name: 'mt-l-arm64g3-16-62-6cxd4-runner-wcjdh'
2026-05-18T05:03:42.8713383Z Runner group name: 'default'
2026-05-18T05:03:42.8714281Z Machine name: 'mt-l-arm64g3-16-62-6cxd4-runner-wcjdh'
...
2026-05-18T05:04:00.3749597Z Complete job name: linux-jammy-aarch64-py3.10 / test-osdc (default, 2, 5, mt-l-arm64g3-16-62)
...
2026-05-18T05:06:22.8815689Z PyTorch built with:
2026-05-18T05:06:22.8816026Z   - GCC 13.4
2026-05-18T05:06:22.8816264Z   - C++ Version: 202002
2026-05-18T05:06:22.8816634Z   - Intel(R) MKL-DNN v3.11.2 (Git Hash 03c022d3ffdcee958cfacbe720048e725fdf644c)
2026-05-18T05:06:22.8817013Z   - OpenMP 201511 (a.k.a. OpenMP 4.5)
2026-05-18T05:06:22.8817333Z   - LAPACK is enabled (usually provided by MKL)
2026-05-18T05:06:22.8817605Z   - NNPACK is enabled
2026-05-18T05:06:22.8817837Z   - CPU capability usage: SVE128

Expected Output of print(torch.__config__.show()) on mt-l-arm64g3-16-62 runner

CPU capability usage: SVE256

Side note:

PyTorch built with:
  - CPU capability usage <CPU_CAPABILITY>

is slightly misleading since the cpu capability is actually a runtime variable that can change depending on the runtime cpu variant.

I will raise an issue in https://github.com/pytorch/test-infra

Versions

2.13.0.dev20260517+cpu

cc @snadampal @milpuz01 @aditew01 @nikhil-arm @fadara01 @nWEIdia

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

pytorch - 💡(How to fix) Fix AArch64 OSDC G3 Runner incorrectly reports cpu capability SVE128