std::bad_alloc

### Custom Node Testing

- [x] I have tried disabling custom nodes and the issue persists (see [how to disable custom nodes](https://docs.comfy.org/troubleshooting/custom-node-issues#step-1%3A-test-with-all-custom-nodes-disabled) if you need help)

### Expected Behavior

no clue how to troubleshoot this, rocm working fine, pytorch working fine. tried every flag that made any sense but always get this memory error. This is a Ryzen 7 255 with 780m GPU and "theRock" with rocm 7rc1 and ubuntu 25

### Actual Behavior

```
python main.py --disable-all-custom-nodes
Checkpoint files will always be loaded safely.
Total VRAM 14673 MB, total RAM 29345 MB
pytorch version: 2.7.1+rocm7.0.0rc20250827
AMD arch: gfx1100
ROCm version: (7, 1)
Set vram state to: NORMAL_VRAM
Device: cuda:0 AMD Radeon Graphics : native
Using pytorch attention
terminate called after throwing an instance of 'std::bad_alloc'
what(): std::bad_alloc
Aborted (core dumped)

python main.py --disable-all-custom-nodes --cpu
Checkpoint files will always be loaded safely.
Total VRAM 29345 MB, total RAM 29345 MB
pytorch version: 2.7.1+rocm7.0.0rc20250827
Set vram state to: DISABLED
Device: cpu
Using sub quadratic optimization for attention, if you have memory or speed issues try using: --use-split-cross-attention
terminate called after throwing an instance of 'std::bad_alloc'
what(): std::bad_alloc
Aborted (core dumped)
```

### Steps to Reproduce

do the log above

### Debug Logs

```powershell
what flags to debug should i run?
```

### Other

_No response_