From 9100e873bdb5643f4548d185124690b2a959dc54 Mon Sep 17 00:00:00 2001 From: lin-bot23 Date: Wed, 22 Jul 2026 10:14:08 +0800 Subject: [PATCH 1/2] fix: explain 'Torch not compiled with CUDA enabled' error for Intel GPU users MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Intel GPU users see 'Torch not compiled with CUDA enabled' and mistakenly follow the NVIDIA CUDA reinstallation steps. Added a note explaining this error is expected for Intel hardware — they should verify XPU availability instead. Closes #1298 --- troubleshooting/overview.mdx | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/troubleshooting/overview.mdx b/troubleshooting/overview.mdx index 62b3e7952..de7129c04 100644 --- a/troubleshooting/overview.mdx +++ b/troubleshooting/overview.mdx @@ -338,6 +338,17 @@ python main.py --force-fp16 --cpu ### Intel GPU Issues +**"Torch not compiled with CUDA enabled" error:** + +If you see this error with an Intel GPU, this is **expected behavior** — Intel GPUs use the XPU backend, not CUDA. Do not follow the NVIDIA CUDA reinstallation steps above; they will not fix this for Intel hardware. + +To verify your setup is working correctly: +```bash +python -c "import torch; print('XPU available:', torch.xpu.is_available())" +``` + +If `torch.xpu.is_available()` returns `True`, ComfyUI will automatically use the XPU backend. If it returns `False`, install the XPU-enabled PyTorch using one of the options below. + **Option 1: Native PyTorch XPU support (Windows/Linux):** ```bash # Install PyTorch nightly with XPU support From 89efa529758fea6606a21b7e4f7d837505641f87 Mon Sep 17 00:00:00 2001 From: lin-bot23 Date: Wed, 22 Jul 2026 13:05:11 +0800 Subject: [PATCH 2/2] =?UTF-8?q?fix:=20reword=20Intel=20GPU=20CUDA=20error?= =?UTF-8?q?=20entry=20=E2=80=94=20remove=20em=20dash=20and=20soften=20abso?= =?UTF-8?q?lute=20language?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- troubleshooting/overview.mdx | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/troubleshooting/overview.mdx b/troubleshooting/overview.mdx index de7129c04..c933165d5 100644 --- a/troubleshooting/overview.mdx +++ b/troubleshooting/overview.mdx @@ -340,7 +340,7 @@ python main.py --force-fp16 --cpu **"Torch not compiled with CUDA enabled" error:** -If you see this error with an Intel GPU, this is **expected behavior** — Intel GPUs use the XPU backend, not CUDA. Do not follow the NVIDIA CUDA reinstallation steps above; they will not fix this for Intel hardware. +On Intel GPU systems this error does not indicate a CUDA problem and does not require reinstalling PyTorch. Intel GPUs use the XPU backend instead of CUDA. Do not follow the NVIDIA CUDA reinstallation steps above; they do not apply to Intel hardware. To verify your setup is working correctly: ```bash