[PR #23] Small fix on init.py #38

New issue

Open

opened 2026-10-26 22:28:18 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-26 22:28:18 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/Comfy-Org/ComfyUI_Nvidia_VFX_Nodes/pull/23
Author: @overallbit
Created: 3/21/2026
Status: 🔄 Open

Base: main ← Head: patch-1

📝 Commits (1)

f2937f8 Update init.py

📊 Changes

1 file changed (+2 additions, -2 deletions)

View changed files

📝 __init__.py (+2 -2)

📄 Description

output tensor was being allocated on CPU (RAM) instead of CUDA (VRAM) because it inherited the device from the input tensor, which ComfyUI provides on CPU. This caused a per-frame GPU→CPU copy inside the loop. Output buffer now allocates directly on CUDA; result is moved back to the original device once after all frames are done.

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/Comfy-Org/ComfyUI_Nvidia_VFX_Nodes/pull/23 **Author:** [@overallbit](https://github.com/overallbit) **Created:** 3/21/2026 **Status:** 🔄 Open **Base:** `main` ← **Head:** `patch-1` --- ### 📝 Commits (1) - [`f2937f8`](https://github.com/Comfy-Org/ComfyUI_Nvidia_VFX_Nodes/commit/f2937f89bd3f59f2efe11ed34e4e070310d4d687) Update __init__.py ### 📊 Changes **1 file changed** (+2 additions, -2 deletions) <details> <summary>View changed files</summary> 📝 `__init__.py` (+2 -2) </details> ### 📄 Description output tensor was being allocated on CPU (RAM) instead of CUDA (VRAM) because it inherited the device from the input tensor, which ComfyUI provides on CPU. This caused a per-frame GPU→CPU copy inside the loop. Output buffer now allocates directly on CUDA; result is moved back to the original device once after all frames are done. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi added the

pull-request

label

2026-10-26 22:28:18 +00:00

bsakaguchi referenced this issue

2026-10-26 22:28:20 +00:00

[PR #38] Reduce memory pressure and fix DLPack lifetime in RTX VSR #42