[GH-ISSUE #20] [ROCm 7.14] Pin PyTorch to 2.11 to avoid high CPU usage #4

New issue

Closed

opened 2026-10-26 22:22:47 +00:00 by bsakaguchi · 1 comment

bsakaguchi commented

2026-10-26 22:22:47 +00:00

Owner

Originally created by @LckHot on GitHub (Jul 30, 2026).
Original GitHub issue: https://github.com/Comfy-Org/ComfyUI-Standalone-Environments/issues/20

Hi, thank you for maintaining this repository.

To avoid the issue reported in ROCm issue #6406, could you please consider pinning the PyTorch version for the ROCm 7.14 environment to PyTorch 2.11 instead of 2.12?

According to the report, PyTorch 2.12 may cause abnormally high CPU usage after the first GPU operation, while downgrading to PyTorch 2.11 resolves the issue.

It may also be necessary to pin compatible versions of torchvision and torchaudio.

Thank you for your consideration.

Originally created by @LckHot on GitHub (Jul 30, 2026). Original GitHub issue: https://github.com/Comfy-Org/ComfyUI-Standalone-Environments/issues/20 Hi, thank you for maintaining this repository. To avoid the issue reported in [ROCm issue #6406](https://github.com/ROCm/ROCm/issues/6406), could you please consider pinning the PyTorch version for the ROCm 7.14 environment to **PyTorch 2.11** instead of 2.12? According to the report, PyTorch 2.12 may cause abnormally high CPU usage after the first GPU operation, while downgrading to PyTorch 2.11 resolves the issue. It may also be necessary to pin compatible versions of `torchvision` and `torchaudio`. Thank you for your consideration.

bsakaguchi closed this issue

2026-10-26 22:22:47 +00:00

bsakaguchi commented

2026-10-26 22:22:48 +00:00

Author

Owner

@coderabbitai[bot] commented on GitHub (Jul 30, 2026):

🔗 Related PRs

Comfy-Org/ComfyUI-Standalone-Environments#11 - Move AMD builds to ROCm 7.14 multi-arch wheels; add amd-multi-arch-index manifest kind [merged]
Comfy-Org/ComfyUI-Standalone-Environments#18 - Bump all vendor bundles to torch 2.12.x [merged]

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!

@coderabbitai[bot] commented on GitHub (Jul 30, 2026):  <details> <summary>🔗 Related PRs</summary> Comfy-Org/ComfyUI-Standalone-Environments#11 - Move AMD builds to ROCm 7.14 multi-arch wheels; add amd-multi-arch-index manifest kind [merged] Comfy-Org/ComfyUI-Standalone-Environments#18 - Bump all vendor bundles to torch 2.12.x [merged] </details> --- <details> <summary> 🧪 Issue enrichment is currently in open beta.</summary> You can configure auto-planning by selecting labels in the issue_enrichment configuration. To disable automatic issue enrichment, add the following to your `.coderabbit.yaml`: ```yaml issue_enrichment: auto_enrich: enabled: false ``` </details> 💬 Have feedback or questions? Drop into our [discord](https://discord.gg/coderabbit)!

bsakaguchi referenced this issue

2026-10-26 22:22:53 +00:00

[PR #4] [MERGED] Add linux-intel-xpu standalone build #9