Skip to content

Update ROCM version to 7.1 from 6.3 - #8867

Closed
heathen711 wants to merge 8 commits into
invoke-ai:mainfrom
heathen711:enhancement/heathen711/issue_8655
Closed

Update ROCM version to 7.1 from 6.3#8867
heathen711 wants to merge 8 commits into
invoke-ai:mainfrom
heathen711:enhancement/heathen711/issue_8655

Conversation

@heathen711

@heathen711 heathen711 commented Feb 10, 2026 •
edited by joshistoast
Loading

Copy link
Copy Markdown
Contributor

Summary

Update ROCM version to 7.1 from 6.3

Related Issues / Discussions

Part 1 of issue #8655
Resolves #9130

QA Instructions

Only affects ROCM users, just need to make sure it runs.

Checklist

  • Test docker image (need to use workflow)
  • Test manual install

Dockers build: https://github.com/heathen711/InvokeAI/actions/runs/21847900570

heathen711 and others added 4 commits February 9, 2026 19:42
@heathen711
Upgrade ROCm support from 6.3 to 7.1
60450b7 
- Update PyTorch to 2.10.0+rocm7.1 (from 2.7.1+rocm6.3)
- Update TorchVision to 0.25.0+rocm7.1 (from 0.22.1+rocm6.3)
- Update PyTorch ROCm index URL to rocm7.1
- Update Docker ROCm runtime to 7.1.1 (from 6.3.4)

ROCm 7.1 adds support for 3 new GPU architectures:
- gfx950: CDNA3+ (Instinct MI350, MI325X)
- gfx1150: RDNA 3.5 (Strix Point APU)
- gfx1151: RDNA 3.5 (Strix Halo APU)

All 11 previously supported architectures remain compatible.
Total supported architectures: 14 (11 existing + 3 new)

Relates to invoke-ai#8655
@heathen711 @claude
Fix uv sync for ROCm 7.1 by explicitly declaring triton-rocm
3ba1196 
The uv package manager couldn't resolve triton-rocm as a transitive
dependency of torch from the PyTorch custom index. This commit adds
triton-rocm==3.6.0 explicitly to the rocm extras with a Linux-only
platform marker.

Changes:
- Added triton-rocm==3.6.0; sys_platform == 'linux' to rocm extras
- Added triton-rocm source configuration pointing to torch-rocm index
- Already committed: torch version constraint relaxation (>=2.7.0,<3.0)
- Already committed: platform environment restrictions for ROCm

Verified with uv sync --extra rocm:
- torch==2.10.0+rocm7.1 installed correctly
- torchvision==0.25.0+rocm7.1 installed correctly
- triton-rocm==3.6.0 resolved successfully
- ROCm 7.1.25424 detected on AMD Radeon PRO V620
- All 8 model architectures accessible
- GPU compute operations verified

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
@heathen711
Update uv lock and remove old dockerfile
776cd58
@heathen711 @claude
Update installation docs for ROCm 7.1
45a60ba 
Changed torch backend from rocm6.3 to rocm7.1 in manual installation
instructions for Invoke v5.12 and later.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
@github-actions github-actions Bot added docker Root docs PRs that change docs python-deps PRs that change python dependencies labels Feb 10, 2026
@heathen711
heathen711 marked this pull request as ready for review February 10, 2026 14:55
@BieHDC BieHDC mentioned this pull request Feb 18, 2026
Closed
1 task
@lstein lstein added the 6.14.0 label Mar 7, 2026
@lstein lstein moved this from 6.14.x to 6.13.x in Invoke - Community Roadmap Mar 7, 2026
@lstein lstein added v6.13.x and removed 6.14.0 labels Mar 7, 2026
@lstein lstein moved this from 6.13.x Theme: MODELS to 6.14.x Theme: LIBRARY UPDATES in Invoke - Community Roadmap Mar 24, 2026
@lstein lstein added 6.14.0 and removed v6.13.x labels Apr 30, 2026
@lstein lstein assigned joshistoast and unassigned blessedcoolant May 11, 2026
@lstein lstein moved this from 6.14.x Theme: LIBRARY UPDATES to 6.13.5 LIBRARY UPDATES in Invoke - Community Roadmap May 12, 2026
@lstein lstein added 6.13.5 Library Updates and removed 6.14.0 labels May 12, 2026
@joshistoast

joshistoast commented May 13, 2026 •
edited
Loading

Copy link
Copy Markdown
Collaborator

Had a guy test rocm 7.2, he wasn't having an easy time getting this particular PR working due to an unrelated technical issue, however invoke DOES work with a simple rocm update.

The speeds were not good, but it does work, we can address rocm speed at a later date probably.

@joshistoast

joshistoast commented May 13, 2026

Copy link
Copy Markdown
Collaborator

Fix the merge conflict and I'll approve

heathen711 @qwencoder
Merge main into enhancement/heathen711/issue_8655
ed54fde 
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
heathen711 and others added 2 commits May 13, 2026 16:24
heathen711 @qwencoder
docs: update ROCm torch backend from 6.3 to 7.1
14a9c99 
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
heathen711 @qwencoder
fix: regenerate uv.lock with ROCm 7.1 dependencies
24672e4 
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
@heathen711

heathen711 commented May 13, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor Author

@joshistoast my code should not have caused these failure AFAIK. Wonder if there is a pinning that isn't resolving the same for MPS and for typegen as I regen the uv.lock? Asking before I spend time deep diving if it's an obvious fix to you.

Typo*

@keturn keturn mentioned this pull request May 23, 2026
Draft
5 tasks
@lstein lstein mentioned this pull request May 25, 2026
Closed
@lstein

lstein commented Jun 17, 2026

Copy link
Copy Markdown
Collaborator

@heathen711 I’d like to get this ROCM update rolled into the 6.13.5 release. Your PR is currently failing several checks and has some easily-fixed conflicts with main. Could you address these and leave a comment when ready for re-review? Thanks.

@lstein lstein mentioned this pull request Jun 23, 2026
Closed
5 tasks
@heathen711

heathen711 commented Jun 23, 2026

Copy link
Copy Markdown
Contributor Author

Apologies, I was on vacation all of last week. I see your fork, closing this.

@heathen711 heathen711 closed this Jun 23, 2026
@lstein lstein mentioned this pull request Jun 23, 2026
Merged
lstein added a commit that referenced this pull request Jun 24, 2026
@lstein @heathen711
@claude heathen711 @qwencoder
Update ROCm to 7.1 (supersedes #8867) (#9297)
f6e8460 
* Upgrade ROCm support from 6.3 to 7.1

- Update PyTorch to 2.10.0+rocm7.1 (from 2.7.1+rocm6.3)
- Update TorchVision to 0.25.0+rocm7.1 (from 0.22.1+rocm6.3)
- Update PyTorch ROCm index URL to rocm7.1
- Update Docker ROCm runtime to 7.1.1 (from 6.3.4)

ROCm 7.1 adds support for 3 new GPU architectures:
- gfx950: CDNA3+ (Instinct MI350, MI325X)
- gfx1150: RDNA 3.5 (Strix Point APU)
- gfx1151: RDNA 3.5 (Strix Halo APU)

All 11 previously supported architectures remain compatible.
Total supported architectures: 14 (11 existing + 3 new)

Relates to #8655

* Fix uv sync for ROCm 7.1 by explicitly declaring triton-rocm

The uv package manager couldn't resolve triton-rocm as a transitive
dependency of torch from the PyTorch custom index. This commit adds
triton-rocm==3.6.0 explicitly to the rocm extras with a Linux-only
platform marker.

Changes:
- Added triton-rocm==3.6.0; sys_platform == 'linux' to rocm extras
- Added triton-rocm source configuration pointing to torch-rocm index
- Already committed: torch version constraint relaxation (>=2.7.0,<3.0)
- Already committed: platform environment restrictions for ROCm

Verified with uv sync --extra rocm:
- torch==2.10.0+rocm7.1 installed correctly
- torchvision==0.25.0+rocm7.1 installed correctly
- triton-rocm==3.6.0 resolved successfully
- ROCm 7.1.25424 detected on AMD Radeon PRO V620
- All 8 model architectures accessible
- GPU compute operations verified

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

* Update uv lock and remove old dockerfile

* docs: update ROCm torch backend from 6.3 to 7.1

Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>

* fix: regenerate uv.lock with ROCm 7.1 dependencies

Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>

* remove redundant inclusion of torch-triton wheel && confirm no aarch64-linux support

* chore(frontend): openapi and typegen

* chore: drop spurious typegen changes from ROCm PR

The ROCm 7.1 update is dependency-only and does not alter any backend
route or model, so schema.ts/openapi.json should match main. The earlier
regeneration introduced spurious diffs (removed @example blocks,
reformatted JSDoc) from a local tooling/backend mismatch, failing the
typegen-checks "compare files" step. Restore both files to main.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* fix(deps): keep macOS on torch 2.7.x; gate rocm extra to linux

Loosening the base torch pin to <3.0 (so the rocm extra could use
2.10.0+rocm7.1) let macOS resolve torch 2.12.0, whose MPS backend fails
with OOM on GitHub's hosted macOS runners (no usable Metal GPU),
breaking the py3.x macos-default pytest jobs.

Split the base pin by platform so macOS stays on 2.7.x (matching main)
while linux/win still allow >=2.10, and gate the rocm extra's
torch/torchvision to sys_platform == 'linux' (ROCm is x86_64-linux-only)
so non-linux resolution is unaffected.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* chore(frontend): regenerate schema for pydantic 2.13 docstring descriptions

The dependency update bumps pydantic 2.11.7 -> 2.13.4, which now emits a
plain class docstring as the JSON-schema description. CacheStats gains
"Collect statistics on cache performance." in both openapi.json and
schema.ts. This is the sole schema delta from the bump and is what the
openapi-checks/typegen-checks jobs regenerate.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Heathen711 <Heathen711@me.com>
Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>
Co-authored-by: heathen711 <heathen711@example.com>
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
dunkeroni pushed a commit to dunkeroni/InvokeAI that referenced this pull request Jun 29, 2026
@lstein @heathen711
@claude heathen711 @qwencoder @dunkeroni
Update ROCm to 7.1 (supersedes invoke-ai#8867) (invoke-ai#9297)
177e7df 
* Upgrade ROCm support from 6.3 to 7.1

- Update PyTorch to 2.10.0+rocm7.1 (from 2.7.1+rocm6.3)
- Update TorchVision to 0.25.0+rocm7.1 (from 0.22.1+rocm6.3)
- Update PyTorch ROCm index URL to rocm7.1
- Update Docker ROCm runtime to 7.1.1 (from 6.3.4)

ROCm 7.1 adds support for 3 new GPU architectures:
- gfx950: CDNA3+ (Instinct MI350, MI325X)
- gfx1150: RDNA 3.5 (Strix Point APU)
- gfx1151: RDNA 3.5 (Strix Halo APU)

All 11 previously supported architectures remain compatible.
Total supported architectures: 14 (11 existing + 3 new)

Relates to invoke-ai#8655

* Fix uv sync for ROCm 7.1 by explicitly declaring triton-rocm

The uv package manager couldn't resolve triton-rocm as a transitive
dependency of torch from the PyTorch custom index. This commit adds
triton-rocm==3.6.0 explicitly to the rocm extras with a Linux-only
platform marker.

Changes:
- Added triton-rocm==3.6.0; sys_platform == 'linux' to rocm extras
- Added triton-rocm source configuration pointing to torch-rocm index
- Already committed: torch version constraint relaxation (>=2.7.0,<3.0)
- Already committed: platform environment restrictions for ROCm

Verified with uv sync --extra rocm:
- torch==2.10.0+rocm7.1 installed correctly
- torchvision==0.25.0+rocm7.1 installed correctly
- triton-rocm==3.6.0 resolved successfully
- ROCm 7.1.25424 detected on AMD Radeon PRO V620
- All 8 model architectures accessible
- GPU compute operations verified

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

* Update uv lock and remove old dockerfile

* docs: update ROCm torch backend from 6.3 to 7.1

Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>

* fix: regenerate uv.lock with ROCm 7.1 dependencies

Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>

* remove redundant inclusion of torch-triton wheel && confirm no aarch64-linux support

* chore(frontend): openapi and typegen

* chore: drop spurious typegen changes from ROCm PR

The ROCm 7.1 update is dependency-only and does not alter any backend
route or model, so schema.ts/openapi.json should match main. The earlier
regeneration introduced spurious diffs (removed @example blocks,
reformatted JSDoc) from a local tooling/backend mismatch, failing the
typegen-checks "compare files" step. Restore both files to main.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* fix(deps): keep macOS on torch 2.7.x; gate rocm extra to linux

Loosening the base torch pin to <3.0 (so the rocm extra could use
2.10.0+rocm7.1) let macOS resolve torch 2.12.0, whose MPS backend fails
with OOM on GitHub's hosted macOS runners (no usable Metal GPU),
breaking the py3.x macos-default pytest jobs.

Split the base pin by platform so macOS stays on 2.7.x (matching main)
while linux/win still allow >=2.10, and gate the rocm extra's
torch/torchvision to sys_platform == 'linux' (ROCm is x86_64-linux-only)
so non-linux resolution is unaffected.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* chore(frontend): regenerate schema for pydantic 2.13 docstring descriptions

The dependency update bumps pydantic 2.11.7 -> 2.13.4, which now emits a
plain class docstring as the JSON-schema description. CacheStats gains
"Collect statistics on cache performance." in both openapi.json and
schema.ts. This is the sole schema delta from the bump and is what the
openapi-checks/typegen-checks jobs regenerate.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Heathen711 <Heathen711@me.com>
Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>
Co-authored-by: heathen711 <heathen711@example.com>
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@lstein lstein Awaiting requested review from lstein lstein is a code owner
@blessedcoolant blessedcoolant Awaiting requested review from blessedcoolant blessedcoolant is a code owner

Assignees

@joshistoast joshistoast

Labels

6.13.5 Library Updates docker docs PRs that change docs python-deps PRs that change python dependencies Root

Projects

Invoke - Community Roadmap
Status: 6.13.5 LIBRARY UPDATES

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

[enhancement]: Rocm 7.2

4 participants

@heathen711 @joshistoast @lstein @blessedcoolant