Gate nvfp4 compute on CUDA 13+ to avoid cuBLAS failures/OOM on older builds

[Alan5168]

Summary

• supports_nvfp4_compute() only checks vendor + compute capability (SM ≥ 10), so Blackwell GPUs on torch builds compiled against CUDA < 13 (e.g. cu128) still take the native NVFP4 path and fail at matmul time — CUBLAS_STATUS_NOT_SUPPORTED from cublasLtMatmulAlgoGetHeuristic, or large VRAM inflation / OOM. Multiple reports in Native NVFP4 (Blackwell) Loading Failure on RTX 5090 - Wan 2.2/Flux2Dev/LTX2 Workflow #11864.
• This adds a torch CUDA build-version gate (major ≥ 13) after the compute-capability check, following the file's existing torch_version_numeric parsing idiom; a missing/unparseable torch.version.cuda is treated as unsupported. Affected setups degrade to the regular quantized-storage path instead of erroring.
• An equivalent local patch was field-reported in Native NVFP4 (Blackwell) Loading Failure on RTX 5090 - Wan 2.2/Flux2Dev/LTX2 Workflow #11864 to resolve the OOM on RTX 50-series + cu128.

Test plan

• python -m py_compile comfy/model_management.py
• RTX 5070 Ti (SM 12.0) + torch 2.11.0+cu130: patched check returns True — no regression; NVFP4 Wan2.2 checkpoints load and sample on the native path on this stack
• Same GPU with torch.version.cuda simulated as "12.8": returns False, gating the Native NVFP4 (Blackwell) Loading Failure on RTX 5090 - Wan 2.2/Flux2Dev/LTX2 Workflow #11864 failure mode
• Confirmation from a cu128 + Blackwell user that the OOM / cuBLAS error is gone

🤖 Generated with Claude Code

[github-actions]

✅ All contributors have signed the CLA. Thank you! This PR is ready to be merged.
_{Posted by the CLA Assistant Lite bot.}

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: fb3ed40f-d5fc-45ad-92e5-3acea30bf389

📥 Commits

Reviewing files that changed from the base of the PR and between 7c59a07 and e33d5b9.

📒 Files selected for processing (1)

• comfy/model_management.py

📜 Recent review details

⚠️ CI failures not shown inline (2)

GitHub Actions: CLA Assistant / 0_cla-assistant.txt: Gate nvfp4 compute on CUDA 13+ to avoid cuBLAS failures/OOM on older builds

Conclusion: failure

View job details

##[group]Run contributor-assistant/github-action@ca4a40a7d1004f18d9960b404b97e5f30a505a08
 with:
   path-to-document: https://github.com/Comfy-Org/comfy-cla/blob/main/comfyui_icla.md
   remote-organization-name: comfy-org
   remote-repository-name: comfy-cla
   path-to-signatures: signatures/cla.json
   branch: main
   allowlist: action@github.com,actions-user,ampagent,claude,comfy-pr-bot,GitHub Action,github-actions,github-actions[bot],Glary Bot,Glary-Bot,*[bot]
   custom-notsigned-prcomment: 🎉 Thank you for your contribution, we really appreciate it! 🎉
Like many open source projects, we require contributors to sign our [Contributor License Agreement (CLA)](https://github.com/Comfy-Org/comfy-cla/blob/main/comfyui_icla.md). A CLA makes the ownership of contributions explicit, so contributors and the project share a clear understanding of how the code can be used. By signing, you:
- Confirm that you own your contribution.
- Keep the right to reuse your own code.
- Grant us a copyright license to include and share it within our projects.
CLAs are standard practice across major open source projects including those under the Apache Software Foundation and the Linux Foundation. Ours is based on the Apache Software Foundation's CLA. Most importantly, it would enable us to relicense the project under a more permissive license in the future, giving the project and its community greater flexibility.
✍ **To sign, please post a new comment on this PR with exactly the following text:** ✍
   custom-pr-sign-comment: I have read and agree to the Contributor License Agreement
   custom-allsigned-prcomment: ✅ All contributors have signed the CLA. Thank you! This PR is ready to be merged.
   use-dco-flag: false
   lock-pullrequest-aftermerge: true
   suggest-recheck: true
 env:
   GITHUB_***REDACTED***
   PERSONAL_ACCESS_***REDACTED***
 ##[endgroup]
 CLA Assistant GitHub Action bot has started the process
 (node:1936) [DEP0040] DeprecationWarning: The `punycode` module is deprec...

---

GitHub Actions: CLA Assistant / cla-assistant: Gate nvfp4 compute on CUDA 13+ to avoid cuBLAS failures/OOM on older builds

Conclusion: failure

View job details

##[group]Run contributor-assistant/github-action@ca4a40a7d1004f18d9960b404b97e5f30a505a08
 with:
   path-to-document: https://github.com/Comfy-Org/comfy-cla/blob/main/comfyui_icla.md
   remote-organization-name: comfy-org
   remote-repository-name: comfy-cla
   path-to-signatures: signatures/cla.json
   branch: main
   allowlist: action@github.com,actions-user,ampagent,claude,comfy-pr-bot,GitHub Action,github-actions,github-actions[bot],Glary Bot,Glary-Bot,*[bot]
   custom-notsigned-prcomment: 🎉 Thank you for your contribution, we really appreciate it! 🎉
Like many open source projects, we require contributors to sign our [Contributor License Agreement (CLA)](https://github.com/Comfy-Org/comfy-cla/blob/main/comfyui_icla.md). A CLA makes the ownership of contributions explicit, so contributors and the project share a clear understanding of how the code can be used. By signing, you:
- Confirm that you own your contribution.
- Keep the right to reuse your own code.
- Grant us a copyright license to include and share it within our projects.
CLAs are standard practice across major open source projects including those under the Apache Software Foundation and the Linux Foundation. Ours is based on the Apache Software Foundation's CLA. Most importantly, it would enable us to relicense the project under a more permissive license in the future, giving the project and its community greater flexibility.
✍ **To sign, please post a new comment on this PR with exactly the following text:** ✍
   custom-pr-sign-comment: I have read and agree to the Contributor License Agreement
   custom-allsigned-prcomment: ✅ All contributors have signed the CLA. Thank you! This PR is ready to be merged.
   use-dco-flag: false
   lock-pullrequest-aftermerge: true
   suggest-recheck: true
 env:
   GITHUB_***REDACTED***
   PERSONAL_ACCESS_***REDACTED***
 ##[endgroup]
 CLA Assistant GitHub Action bot has started the process
 (node:1936) [DEP0040] DeprecationWarning: The `punycode` module is deprec...

🧰 Additional context used

📓 Path-based instructions (5)

**/*

📄 CodeRabbit inference engine (AGENTS.md)

**/*: Keep changes small, direct, and limited to the narrowest code path and fewest files necessary; prefer practical fixes over broad architectural work.
Prefer fewer dependencies and do not add a ComfyUI dependency unless absolutely necessary.
Remove obsolete code, dead branches, unused options, debug prints, and unnecessary compatibility paths.
Preserve existing APIs, node names, model-loading behavior, file layout, and workflow compatibility unless replacement is explicit.
Keep architectural layers focused; do not leak UI, API, workflow, queue, persistence, telemetry, model-loading, node, or execution concerns into unrelated layers.
Shared core modules should depend only on lower-level primitives and their own domain concepts; higher-level concepts belong at callers, adapters, services, or boundaries.
Pass only the narrowest data needed across boundaries and keep identity mapping, persistence, history, telemetry, response shaping, and UI state in their owning layers.
Before touching many files, identify the smallest owner layer; use caller-side mappings, adapters, events, or narrow interfaces instead of exposing private concepts across layers.
Core ComfyUI code must not make unsolicited internet requests or add uploads, telemetry, analytics, tracking, reporting, update checks, remote configuration, licensing checks, or similar outbound paths.
Model downloading is allowed only when explicitly authorized by the user, limited to the requested artifact, and free of telemetry, tracking, unrelated metadata, or background activity.
Warning and info messages should be short and actionable; documentation and README changes should be concise, factual, and tied to changed behavior.
Use short direct commit subjects such as Fix ..., Add ..., Support ..., Remove ..., or Update ...; keep PR descriptions short and state the problem, behavior change, and tests.
Prefer one coherent behavioral change per commit and prioritize crashes, incorrect dtype/device behavior, m...

Files:

• comfy/model_management.py

**/*.py

📄 CodeRabbit inference engine (AGENTS.md)

**/*.py: Keep state and capability flags on the object that owns the behavior; use explicit parent-owned fields rather than probing children with getattr for parent control flow.
Keep public methods aligned with caller contracts; preserve arguments, parameter order, return types, side effects, and error behavior unless all affected interfaces are intentionally updated.
Do not add unused compatibility parameters, flags, attributes, constructor options, or model-specific options to shared helpers.
Normalize third-party and upstream return conventions at integration boundaries so core code receives the expected type and shape.
Avoid caller-side unwrapping such as out = out[0] unless the called interface documents that return structure.
Do not add torch.no_grad, torch.inference_mode, or inference-mode wrappers; only disable a globally enabled inference mode when a training path requires gradients.
Do not add freeze, unfreeze, or trainability toggles to ComfyUI model classes.
Remove training-only behavior such as dropout from inference models while preserving checkpoint and state-dict compatibility, using nn.Identity when necessary to retain slots.
Keep imports at module scope; use inline imports only for established optional-backend probes or import-cycle avoidance.
Use try/except only for optional dependency, platform, or backend detection with a useful fallback, and prefer specific exception types.
Do not add code for unsupported pinned library versions or obsolete PyTorch workarounds; unsupported formats, quantization metadata, and bad states should fail clearly.
Match local file style and keep comments sparse, useful, and non-obvious; remove comments that merely restate code.

Files:

• comfy/model_management.py

comfy/**/*.py

📄 CodeRabbit inference engine (AGENTS.md)

comfy/**/*.py: Treat dtype, device placement, VRAM use, offloading, and memory behavior as correctness concerns; use existing ComfyUI quantization, memory, offload, and optimized-operation helpers.
Prefer shared optimized kernels, backend dispatchers, and documented interfaces over duplicate handwritten operations; treat selected backend callables as opaque.
Do not duplicate operations with custom float32-upcasting inference kernels; use generic ComfyUI operations or native PyTorch operations.
If a model constructor has an operations parameter, assume it is never None; do not add fallback torch operations.
Avoid unnecessary parameters in model, block, and operation constructors or forwards; reuse existing model classes, blocks, operations, and helpers.
Model detectors must inspect only the first dimension of linear weights, guard every dereferenced state-dict key, and order specific signatures before broad fallbacks.
Avoid einops in core inference code; use native tensor operations such as reshape, view, permute, transpose, flatten, and related methods.
Keep metadata, counters, shape calculations, indices, split boundaries, and control-flow values as Python values rather than tensors.
Avoid unnecessary casts and transfers; preserve intended compute and storage dtypes, tensor shapes, and backend result contracts.
Keep model-native latent layout handling inside the model or latent-format owner rather than reshaping it in nodes or caller-side adapters.
DiT models must pad every patchified target or reference input with comfy.ldm.common_dit.pad_to_patch_size and crop only the target output back to its original dimensions.
Do not add defensive shape, configuration, or dtype casts that merely obscure a clear tensor-operation failure; validate only at meaningful boundaries.
Raw parameters not owned by an operation should be cast at use with comfy.ops.cast_to_input or comfy.model_management.cast_to; model constructors should not contain dtype workaro...

Files:

• comfy/model_management.py

**

⚙️ CodeRabbit configuration file

**: IMPORTANT: Only comment on issues directly introduced by this PR's code changes.
Treat AGENTS.md as mandatory repository policy, not optional style guidance.
Flag PR changes that violate AGENTS.md even when the code is otherwise functional.
In particular, enforce architecture boundaries, dtype/device/memory rules,
interface contracts, import style, no unnecessary try/except blocks, no inline
imports, no outbound internet paths in core ComfyUI, and narrow scoped fixes.
Prefer direct findings over suggestions when a rule is violated. Only ignore
AGENTS.md when it clearly conflicts with a newer explicit maintainer instruction
in the PR.
Do NOT flag pre-existing issues in code that was merely moved, re-indented,
de-indented, or reformatted without logic changes. If code appears in the diff
only due to whitespace or structural reformatting (e.g., removing a with: block),
treat it as unchanged. Contributors should not feel obligated to address
pre-existing issues outside the scope of their contribution.

Files:

• comfy/model_management.py

comfy/**

⚙️ CodeRabbit configuration file

comfy/**: Core ML/diffusion engine. Focus on:

• Backward compatibility (breaking changes affect all custom nodes)
• Memory management and GPU resource handling
• Performance implications in hot paths
• Thread safety for concurrent execution

Files:

• comfy/model_management.py

🧠 Learnings (2)

📚 Learning: 2026-02-21T14:01:41.482Z

Learnt from: pythongosssss
Repo: Comfy-Org/ComfyUI PR: 12555
File: comfy_extras/nodes_glsl.py:719-724
Timestamp: 2026-02-21T14:01:41.482Z
Learning: In PyOpenGL, bare Python scalars can be accepted for 1-element array parameters by NumberHandler. This means you can pass an int/float directly to OpenGL texture deletion (e.g., glDeleteTextures(tex)) without wrapping in a list. Verify function-specific expectations and ensure types match what the OpenGL call expects; use explicit lists only when the API requires an array.

Applied to files:

• comfy/model_management.py

📚 Learning: 2026-05-13T12:31:45.069Z

Learnt from: rattus128
Repo: Comfy-Org/ComfyUI PR: 13802
File: comfy/pinned_memory.py:19-30
Timestamp: 2026-05-13T12:31:45.069Z
Learning: When reviewing code that uses comfy/pinned_memory.py’s `HostBuffer.extend(size=..., reallocate=...)`: by default (`reallocate` is not True / False), `extend(size=...)` is a *relative increment* that grows the buffer by `size` bytes—so slicing like `[offset:offset+size]` after `hostbuf.extend(size=size)` is correct and the argument should not be rewritten to `offset + size`. Only in the single-segment reallocation mode (`reallocate=True`, e.g., as used by `resize_pin_buffer()` in `comfy/model_management.py`) should `size` be treated as an *absolute target* and the call/arguments should be checked accordingly.

Applied to files:

• comfy/model_management.py

---

📝 Walkthrough

Walkthrough

Updated the FP8 compute capability check to parse the installed CUDA runtime’s major version. The function now returns false when the version is below CUDA 13 or cannot be parsed.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title matches the main change: gating NVFP4 compute on CUDA 13+ to avoid failures on older builds.
Description check	✅ Passed	The description accurately explains the CUDA 13+ gate, the failure mode it addresses, and the test plan.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[Alan5168]

I have read and agree to the Contributor License Agreement