Frequent Recompile using TorchcompileModel Node even if the resolution is fixed

### Expected Behavior

Here is my workflow:
[Problem Demostrate.json](https://github.com/user-attachments/files/18364282/Problem.Demostrate.json)
I only enabled my personal node for Gemini prompt generator, and no other custom node at all,
https://github.com/LiJT/ComfyUI-Gemini-Prompt-Generator-JT
If you have your own favourite prompt generator like ollama or others that use external LLM to generate prompt, please just use yours, The problem remains the same, I have also tried using the LayerStyle Gemini prompt generator, still the very same result
As you can see from the image above, its the screenshot of the workflow.
After each run, the resolution is totally fixed, only the prompt is changed from the prompt node, but the whole workflow will start to recompile again and again for each generation, these unknown reason troubles me a lot.

### Actual Behavior

The torch compile will Frequent recompile after each prompt generation received from the LLM. it is soooooo annoying because the process getting extremely slow for each generation
Initially I thought the problem was caused by this node:
https://github.com/chengzeyi/Comfy-WaveSpeed/issues/18#issuecomment-2580709264
But after I tried using the official node getting the very same error, there must be something wrong with settings somewhere but i cannot find where

### Steps to Reproduce

Use any Flux text2image workflow, add the ComfyUI built in TorchCompileModel node after "Load Diffusion Model" Node, leave the setting as default "Inductor", then, in your Clip text encode node, use any prompt generator node and plug the output to its input text area, and just run and queue 5 times, you will see that the whole workflow will recompile 5 times after each generation for unknown reason!!!

### Debug Logs

```powershell
E:\ComfyUI_windows_portable>.\python_embeded\python.exe -s ComfyUI\main.py --windows-standalone-build
Adding extra search path checkpoints E:/SD-webui-forge/webui\models/Stable-diffusion
Adding extra search path configs E:/SD-webui-forge/webui\models/Stable-diffusion
Adding extra search path vae E:/SD-webui-forge/webui\models/VAE
Adding extra search path loras E:/SD-webui-forge/webui\models/Lora
Adding extra search path loras E:/SD-webui-forge/webui\models/LyCORIS
Adding extra search path upscale_models E:/SD-webui-forge/webui\models/ESRGAN
Adding extra search path upscale_models E:/SD-webui-forge/webui\models/RealESRGAN
Adding extra search path upscale_models E:/SD-webui-forge/webui\models/SwinIR
Adding extra search path upscale_models E:/SD-webui-forge/webui\models\upscale_models
Adding extra search path embeddings E:/SD-webui-forge/webui\embeddings
Adding extra search path hypernetworks E:/SD-webui-forge/webui\models/hypernetworks
Adding extra search path controlnet E:/SD-webui-forge/webui\models/ControlNet
Adding extra search path clip E:/SD-webui-forge/webui\models/clip/
Adding extra search path clip_vision E:/SD-webui-forge/webui\models/clip_vision
Adding extra search path ipadapter E:/SD-webui-forge/webui\models/ipadapter
Adding extra search path blip E:/SD-webui-forge/webui\models/BLIP
Adding extra search path instantid E:/SD-webui-forge/webui\models/instantid
Adding extra search path insightface E:/SD-webui-forge/webui\models/insightface
[START] Security scan
[DONE] Security scan

ComfyUI-Manager: installing dependencies done.
** ComfyUI startup time: 2025-01-09 18:01:19.474442
** Platform: Windows
** Python version: 3.12.7 (tags/v3.12.7:0b05ead, Oct 1 2024, 03:06:41) [MSC v.1941 64 bit (AMD64)]
** Python executable: E:\ComfyUI_windows_portable\python_embeded\python.exe
** ComfyUI Path: E:\ComfyUI_windows_portable\ComfyUI
** Log path: E:\ComfyUI_windows_portable\ComfyUI\user\comfyui.log

Prestartup times for custom nodes:
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Easy-Use
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\rgthree-comfy
2.3 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Manager

Total VRAM 24563 MB, total RAM 65312 MB
pytorch version: 2.5.1+cu124
xformers version: 0.0.28.post3
Set vram state to: NORMAL_VRAM
Device: cuda:0 NVIDIA GeForce RTX 4090 : cudaMallocAsync
Using xformers attention
[Prompt Server] web root: E:\ComfyUI_windows_portable\ComfyUI\web
[Crystools INFO] Crystools version: 1.21.0
[Crystools INFO] CPU: Intel(R) Core(TM) i9-14900K - Arch: AMD64 - OS: Windows 10
[Crystools INFO] Pynvml (Nvidia) initialized.
[Crystools INFO] GPU/s:
[Crystools INFO] 0) NVIDIA GeForce RTX 4090
[Crystools INFO] NVIDIA Driver: 566.36
Flow - 0.5.2: Downloading and Upading Flows
Flow - 0.5.2: Flows have been updated successfully.
[ComfyUI-Easy-Use] server: v1.2.6 Loaded
[ComfyUI-Easy-Use] web root: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Easy-Use\web_version/v2 Loaded
E:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\timm\models\layers_init_.py:48: FutureWarning: Importing from timm.models.layers is deprecated, please import via timm.layers
warnings.warn(f"Importing from {name} is deprecated, please import via timm.layers", FutureWarning)
E:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\timm\models\registry.py:4: FutureWarning: Importing from timm.models.registry is deprecated, please import via timm.models
warnings.warn(f"Importing from {name} is deprecated, please import via timm.models", FutureWarning)
E:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\timm\models\helpers.py:7: FutureWarning: Importing from timm.models.helpers is deprecated, please import via timm.models
warnings.warn(f"Importing from {name} is deprecated, please import via timm.models", FutureWarning)

Loading: ComfyUI-Impact-Pack (V8.2)
[Impact Pack] Wildcards loading done.

Loading: ComfyUI-Impact-Subpack (V1.2.6)
[Impact Subpack] ultralytics_bbox: E:\ComfyUI_windows_portable\ComfyUI\models\ultralytics\bbox
[Impact Subpack] ultralytics_segm: E:\ComfyUI_windows_portable\ComfyUI\models\ultralytics\segm

Loading: ComfyUI-Inspire-Pack (V1.9.1)
Total VRAM 24563 MB, total RAM 65312 MB
pytorch version: 2.5.1+cu124
xformers version: 0.0.28.post3
Set vram state to: NORMAL_VRAM
Device: cuda:0 NVIDIA GeForce RTX 4090 : cudaMallocAsync

Loading: ComfyUI-Manager (V3.4)
ComfyUI Version: v0.3.10-41-g2307ff67 | Released on '2025-01-08'
[SD Prompt Reader] Node version: 1.3.4
[SD Prompt Reader] Core version: 1.3.5
[ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/model-list.json
[ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/github-stats.json
[ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/extension-node-map.json
[ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json
E:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\albumentations_init_.py:24: UserWarning: A new version of Albumentations is available: 2.0.0 (you have 1.4.22). Upgrade using: pip install -U albumentations. To disable automatic update checks, set the environment variable NO_ALBUMENTATIONS_UPDATE to 1.
check_for_updates()
[ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/alter-list.json
Nvidia APEX normalization not installed, using PyTorch LayerNorm
(pysssss:WD14Tagger) [DEBUG] Available ORT providers: TensorrtExecutionProvider, CUDAExecutionProvider, CPUExecutionProvider
(pysssss:WD14Tagger) [DEBUG] Using ORT providers: CUDAExecutionProvider, CPUExecutionProvider
Workspace manager - Openning file hash dict
🦄🦄Loading: Workspace Manager (V2.1.0)
[comfyui_controlnet_aux] | INFO -> Using ckpts path: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_controlnet_aux\ckpts
[comfyui_controlnet_aux] | INFO -> Using symlinks: False
[comfyui_controlnet_aux] | INFO -> Using ort providers: ['CUDAExecutionProvider', 'DirectMLExecutionProvider', 'OpenVINOExecutionProvider', 'ROCMExecutionProvider', 'CPUExecutionProvider', 'CoreMLExecutionProvider']
DWPose: Onnxruntime with acceleration providers detected

[START] ComfyUI AlekPet Nodes v1.0.39
Node -> ChatGLMNode: ChatGLM4TranslateCLIPTextEncodeNode, ChatGLM4TranslateTextNode, ChatGLM4InstructNode, ChatGLM4InstructMediaNode [Loading]
Node -> GoogleTranslateNode: GoogleTranslateCLIPTextEncodeNode, GoogleTranslateTextNode [Loading] Node -> ArgosTranslateNode: ArgosTranslateCLIPTextEncodeNode, ArgosTranslateTextNode [Loading]

Node -> DeepTranslatorNode: DeepTranslatorCLIPTextEncodeNode, DeepTranslatorTextNode [Loading]
Node -> ExtrasNode: PreviewTextNode, HexToHueNode, ColorsCorrectNode [Loading]
Node -> PoseNode: PoseNode [Loading]
Node -> IDENode: IDENode [Loading]
Node -> PainterNode: PainterNode [Loading]

[END] ComfyUI AlekPet Nodes
LOADED 6 FONTS
** Comfly Loaded : fly, just fly

███╗ ███╗ █████╗ ██████╗██╗ ██╗██╗███╗ ██╗███████╗
████╗ ████║██╔══██╗██╔════╝██║ ██║██║████╗ ██║██╔════╝
██╔████╔██║███████║██║ ███████║██║██╔██╗ ██║█████╗
██║╚██╔╝██║██╔══██║██║ ██╔══██║██║██║╚██╗██║██╔══╝
██║ ╚═╝ ██║██║ ██║╚██████╗██║ ██║██║██║ ╚████║███████╗
╚═╝ ╚═╝╚═╝ ╚═╝ ╚═════╝╚═╝ ╚═╝╚═╝╚═╝ ╚═══╝╚══════╝

██████╗ ███████╗██╗ ██╗ ██╗███████╗██╗ ██████╗ ███╗ ██╗███████╗
██╔══██╗██╔════╝██║ ██║ ██║██╔════╝██║██╔═══██╗████╗ ██║██╔════╝
██║ ██║█████╗ ██║ ██║ ██║███████╗██║██║ ██║██╔██╗ ██║███████╗
██║ ██║██╔══╝ ██║ ██║ ██║╚════██║██║██║ ██║██║╚██╗██║╚════██║
██████╔╝███████╗███████╗╚██████╔╝███████║██║╚██████╔╝██║ ╚████║███████║
╚═════╝ ╚══════╝╚══════╝ ╚═════╝ ╚══════╝╚═╝ ╚═════╝ ╚═╝ ╚═══╝╚══════╝

[tinyterraNodes] Loaded
All packages from requirements.txt are installed and up to date.
llama-cpp installed
All packages from requirements.txt are installed and up to date.
[comfy_mtb] | INFO -> loaded 94 nodes successfuly
[comfy_mtb] | INFO -> Some nodes (2) could not be loaded. This can be ignored, but go to http://127.0.0.1:8188/mtb if you want more information.

[rgthree-comfy] Loaded 42 epic nodes. 🎉

WAS Node Suite: OpenCV Python FFMPEG support is enabled
WAS Node Suite: ffmpeg_bin_path is set to: C:\ffmpeg\ffmpeg.exe
WAS Node Suite: Finished. Loaded 220 nodes successfully.

"The best revenge is massive success." - Frank Sinatra
Import times for custom nodes:
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-NodeAligner
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\websocket_image_save.py
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\cg-use-everywhere
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Inpaint-CropAndStitch
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-mxToolkit
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_AdvancedRefluxControl
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-photoshop
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUi_NNLatentUpscale
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\cg-image-picker
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-42lux
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-WD14-Tagger
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Image-Saver
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-inpaint-nodes
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_IPAdapter_plus
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_FluxMod
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\wavespeed
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-LTXTricks
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_UltimateSDUpscale
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyMath
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_essentials
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-DepthAnythingV2
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Custom-Scripts
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\Comfyui_Comfly
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Frame-Interpolation
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-ppm
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Fluxtapoz
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-GGUF
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Advanced-ControlNet
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-prompt-reader-node
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-JDCN
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\rgthree-comfy
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-HunyuanVideoWrapper
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-workspace-manager
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-KJNodes
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-MingNodes
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Inspire-Pack
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_tinyterraNodes
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Impact-Pack
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_controlnet_aux
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfy_mtb
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_LayerStyle_Advance
0.0 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-FluxTrainer
0.1 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_LayerStyle
0.1 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Crystools
0.1 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_VLM_nodes
0.1 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-CogVideoXWrapper
0.2 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Detail-Daemon
0.2 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Gemini-Prompt-Generator-JT
0.3 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Easy-Use
0.4 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-GIMM-VFI
0.5 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Manager
0.6 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-BrushNet
0.9 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-PuLID-Flux-Enhanced
0.9 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-VideoHelperSuite
1.1 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\was-node-suite-comfyui
1.3 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_Fill-Nodes
1.4 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_Custom_Nodes_AlekPet
2.1 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Impact-Subpack
2.7 seconds: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-disty-Flow

Starting server

To see the GUI go to: http://127.0.0.1:8188
FETCH DATA from: E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Manager\extension-node-map.json [DONE]
[comfy_mtb] | INFO -> Found multiple match, we will pick the last E:/SD-webui-forge/webui\models\upscale_models
['E:\ComfyUI_windows_portable\ComfyUI\models\upscale_models', 'E:/SD-webui-forge/webui\models/ESRGAN', 'E:/SD-webui-forge/webui\models/RealESRGAN', 'E:/SD-webui-forge/webui\models/SwinIR', 'E:/SD-webui-forge/webui\models\upscale_models']
got prompt
Using xformers attention in VAE
Using xformers attention in VAE
VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16
----INPUT----
Generate me a prompt for image generator. The theme of the prompt is a black cat. You already created those prompts: []. Make sure you generate original prompt. Think about it step by step and make some internal critique. You only need to output generated prompt and nothing else You must keep the length of your generated prompt around 200 words.
----OUTPUT----
A majestic black cat, sleek and muscular, perched atop a crumbling gothic tower silhouetted against a blood-red sunset. Rain slicks its fur, highlighting its emerald eyes that gleam with an unsettling intelligence. The wind whips around it, causing its fur to stand on end, revealing a subtle skeletal structure beneath. The tower is overgrown with ivy, its stones worn and weathered by centuries of storms. In the background, a swirling vortex of dark clouds threatens to engulf the scene. The overall style should be hyperrealistic, with a focus on texture and detail. The cat's expression is one of aloof contemplation, almost regal in its indifference to the chaotic elements around it. Render the scene with dramatic chiaroscuro lighting, emphasizing the contrast between light and shadow. Include subtle magical elements – perhaps a faint, ethereal glow emanating from the cat's eyes, or wisps of mystical energy swirling around the tower. The color palette should be predominantly dark and moody, with splashes of vibrant red from the sunset and emerald green from the cat's eyes. The image should evoke a sense of mystery, power, and ancient magic, suggesting a creature both beautiful and dangerous, a guardian of secrets hidden within the crumbling stone. The texture should be incredibly detailed, showing every strand of fur, every crack in the stone, every drop of rain.
model weight dtype torch.float8_e4m3fn, manual cast: torch.bfloat16
model_type FLUX
Requested to load FluxClipModel_
loaded completely 9.5367431640625e+25 4777.53759765625 True
CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cuda:0, dtype: torch.float16
clip missing: ['text_projection.weight']
Requested to load Flux
loaded completely 9.5367431640625e+25 11350.067443847656 True
0%| | 0/28 [00:00pause
```

### Other

Please tell me where I did wrong.... I followed the instruction here to enable Triton and everything working as intended
https://purz.notion.site/Get-Windows-Triton-working-for-Mochi-6a0c055e21c84cfba7f1dd628e624e97
https://github.com/woct0rdho/triton-windows?tab=readme-ov-file

I just want ComfyUI will only compile once for a fixed resolution settings, and let me use the external prompt generator at the same time.. But this bothers me for a quite long time