[Bug]: Generation just hangs for ever before last step

[Mozoloa]

Is there an existing issue for this?

• I have searched the existing issues and checked the recent builds/commits

What happened?

Since the update 1.1, very often when I do batches of images, one of them will hang at one of the latest steps and never complete.

Clicking interrupt does nothing, so does skip and reloading the UI doesn't help, the whole UI is stuck and it seems that no other functionality works. The console shows the total progress this way (I'm generating 100 batches of one 512x512 images ) :

100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:02<00:00,  6.99it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:03<00:00,  6.44it/s]
Total progress:   3%|█▉                                                              | 60/2000 [00:11<04:26,  7.27it/s]

I can't do anything but start the whole thing

Steps to reproduce the problem

1. Go to TXT2IMG or IMG2IMG
2. Do a large batch of images
3. At some point the generation will hang and nothing will work anymore

What should have happened?

The generation should have continued like it did before

Commit where the problem happens

c3eced2

What platforms do you use to access the UI ?

Windows 11, RTX3090

What browsers do you use to access the UI ?

Brave

Command Line Arguments

--ckpt-dir 'G:\AI\Models\Stable-diffusion\Checkpoints' --xformers --embeddings-dir 'G:\AI\Models\Stable-diffusion\Embeddings' --lora-dir 'G:\AI\Models\Stable-diffusion\Lora

OR

--ckpt-dir 'G:\AI\Models\Stable-diffusion\Checkpoints' --otp-sdp-attention --embeddings-dir 'G:\AI\Models\Stable-diffusion\Embeddings' --lora-dir 'G:\AI\Models\Stable-diffusion\Lora

List of extensions

ControlNet v1.1.134
Image browser

Console logs

venv "G:\AI\Image Gen\A1111\stable-diffusion-webui\venv\Scripts\Python.exe"
Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug  1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)]
Commit hash: c3eced22fc7b9da4fbb2f55f2d53a7e5e511cfbd
Installing xformers
Collecting xformers==0.0.17
  Using cached xformers-0.0.17-cp310-cp310-win_amd64.whl (112.6 MB)
Installing collected packages: xformers
Successfully installed xformers-0.0.16
Installing requirements


Installing ImageReward requirement for image browser

Launching Web UI with arguments: --autolaunch --ckpt-dir G:\AI\Models\Stable-diffusion\Checkpoints --xformers --embeddings-dir G:\AI\Models\Stable-diffusion\Embeddings --lora-dir G:\AI\Models\Stable-diffusion\Lora --reinstall-xformers
ControlNet v1.1.134
ControlNet v1.1.134
Loading weights [3dcc66eccf] from G:\AI\Models\Stable-diffusion\Checkpoints\Men\Saruman.ckpt
Creating model from config: G:\AI\Image Gen\A1111\stable-diffusion-webui\configs\v1-inference.yaml
LatentDiffusion: Running in eps-prediction mode
DiffusionWrapper has 859.52 M params.
Loading VAE weights specified in settings: G:\AI\Image Gen\A1111\stable-diffusion-webui\models\VAE\NewVAE.vae.pt
Applying xformers cross attention optimization.
Textual inversion embeddings loaded(15): bad-artist, bad-artist-anime, bad-hands-5, bad-image-v2-39000, bad-picture-chill-75v, bad_prompt, bad_prompt_version2, badhandv4, charturnerv2, easynegative, HyperStylizeV6, ng_deepnegative_v1_75t, pureerosface_v1, ulzzang-6500, ulzzang-6500-v1.1
Textual inversion embeddings skipped(4): 21charturnerv2, nartfixer, nfixer, nrealfixer
Model loaded in 7.2s (load weights from disk: 2.5s, create model: 0.4s, apply weights to model: 0.4s, apply half(): 0.3s, load VAE: 0.5s, move model to device: 0.6s, load textual inversion embeddings: 2.5s).
Running on local URL:  http://127.0.0.1:7860

To create a public link, set `share=True` in `launch()`.
Startup time: 19.8s (import torch: 2.7s, import gradio: 2.2s, import ldm: 1.0s, other imports: 2.4s, list SD models: 0.4s, setup codeformer: 0.1s, load scripts: 1.8s, load SD checkpoint: 7.2s, create ui: 1.2s, gradio launch: 0.7s).
Loading weights [c6bbc15e32] from G:\AI\Models\Stable-diffusion\Checkpoints\0\1.5-inpainting.ckpt
Creating model from config: G:\AI\Image Gen\A1111\stable-diffusion-webui\configs\v1-inpainting-inference.yaml
LatentInpaintDiffusion: Running in eps-prediction mode
DiffusionWrapper has 859.54 M params.
Loading VAE weights specified in settings: G:\AI\Image Gen\A1111\stable-diffusion-webui\models\VAE\NewVAE.vae.pt
Applying xformers cross attention optimization.
Model loaded in 2.0s (create model: 0.4s, apply weights to model: 0.4s, apply half(): 0.3s, load VAE: 0.2s, move model to device: 0.6s).
Running DDIM Sampling with 19 timesteps
Decoding image: 100%|██████████████████████████████████████████████████████████████████| 19/19 [00:02<00:00,  9.21it/s]
Running DDIM Sampling with 19 timesteps                                              | 18/2000 [00:01<03:04, 10.77it/s]
Decoding image: 100%|██████████████████████████████████████████████████████████████████| 19/19 [00:01<00:00, 13.87it/s]
Running DDIM Sampling with 19 timesteps                                              | 38/2000 [00:04<02:31, 12.94it/s]
Decoding image: 100%|██████████████████████████████████████████████████████████████████| 19/19 [00:01<00:00, 12.92it/s]
Running DDIM Sampling with 19 timesteps                                              | 56/2000 [00:07<02:37, 12.31it/s]
Decoding image: 100%|██████████████████████████████████████████████████████████████████| 19/19 [00:01<00:00, 13.33it/s]
Running DDIM Sampling with 19 timesteps                                              | 76/2000 [00:10<02:29, 12.88it/s]
Decoding image: 100%|██████████████████████████████████████████████████████████████████| 19/19 [00:01<00:00, 12.03it/s]
Running DDIM Sampling with 19 timesteps                                              | 94/2000 [00:13<03:02, 10.43it/s]
Decoding image: 100%|██████████████████████████████████████████████████████████████████| 19/19 [00:01<00:00, 13.91it/s]
Running DDIM Sampling with 19 timesteps                                             | 113/2000 [00:15<02:33, 12.31it/s]
Decoding image: 100%|██████████████████████████████████████████████████████████████████| 19/19 [00:01<00:00, 13.84it/s]
Running DDIM Sampling with 19 timesteps                                             | 133/2000 [00:18<02:23, 13.03it/s]
Decoding image:  21%|██████████████                                                     | 4/19 [00:00<00:01, 11.32it/s]
Total progress:   7%|████▎                                                          | 137/2000 [00:21<04:56,  6.28it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:02<00:00,  6.90it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:02<00:00,  6.94it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:02<00:00,  7.14it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:03<00:00,  6.42it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:02<00:00,  6.81it/s]
  0%|                                                                                           | 0/20 [00:00<?, ?it/s]
Total progress:   5%|███▏                                                           | 101/2000 [00:23<07:14,  4.37it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:02<00:00,  7.10it/s]
 75%|█████████████████████████████████████████████████████████████▌                    | 15/20 [00:02<00:00,  6.22it/s]
Total progress:   2%|█▏                                                              | 36/2000 [00:07<06:58,  4.69it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:03<00:00,  6.17it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:02<00:00,  6.89it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:02<00:00,  7.07it/s]
 10%|████████▎                                                                          | 2/20 [00:00<00:03,  4.87it/s]
Total progress:   3%|██                                                              | 63/2000 [00:14<07:18,  4.42it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:02<00:00,  7.57it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:02<00:00,  6.99it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:03<00:00,  6.44it/s]
Total progress:   3%|█▉                                                              | 60/2000 [00:11<04:26,  7.27it/s]

Additional information

I remember that at some point it hanged but got unstuck somehow and I got an error which I don't remember but it did say to use --no-half-vae, I haven't tested that and never needed that before on torch 1.13.1 for tens of thousands of gens. I'm exclusively using the new 840000 mse VAE

[Mozoloa]

New information: I've tried --no-half-vae and it doesn't change anything. Also the hanging seems to also happen when I try to interrupt some gens, still no information in the console

[begon123]

It began to arise more and more often, rebooting sdiffused and chrome no longer helps.
Gives 1-2 generations and again an error.

Please help!

p.s windows 11, rtx 3060 (last drivers)

Sorry, I have vladmandic/automatic - the bug report is not for you, but the error is exactly the same and I have not found a similar one anywhere.

[VRArt1]

I've also been having this issue since one of the recent updates.

[Chem1ce]

Im also having the same issue.

[Mozoloa]

This has to do with cu118 or Torch 2.0, I reverted to 1.13.1+cu117 and I never get it

[ChenNdG]

This has to do with cu118 or Torch 2.0, I reverted to 1.13.1+cu117 and I never get it

so you back to 1.0 version of A1111 ?
Or you just use 1.13.1+cu117 in A1111 v1.1 ?

[Mozoloa]

This has to do with cu118 or Torch 2.0, I reverted to 1.13.1+cu117 and I never get it

so you back to 1.0 version of A1111 ?
Or you just use 1.13.1+cu117 in A1111 v1.1 ?

The UI still works with 1.13.1, I changed a line in launch.py that talked about torch to change it back to 1.13.1+cu117 but I don't know exactly how anymore since I'm on my phone, and added --reinstall-torch to the command line arguments, but tbh now I just renamed venv to venv2 so I have both versiond of torch at the ready by just renaming venv

[ChenNdG]

This has to do with cu118 or Torch 2.0, I reverted to 1.13.1+cu117 and I never get it

so you back to 1.0 version of A1111 ?
Or you just use 1.13.1+cu117 in A1111 v1.1 ?

The UI still works with 1.13.1, I changed a line in launch.py that talked about torch to change it back to 1.13.1+cu117 but I don't know exactly how anymore since I'm on my phone, and added --reinstall-torch to the command line arguments, but tbh now I just renamed venv to venv2 so I have both versiond of torch at the ready by just renaming venv

if I understand correctly, I have to change this line here :

[Mozoloa]

Found the commit that changed it : d5063e0 just revert the change from launch.py

[ChenNdG]

Found the commit that changed it : d5063e0 just revert the change from launch.py

Thanks !

[halr9000]

Having same issue, can't reliably reproduce, it just happens when it happens, and there's no hints in the console for troubleshooting.

[quasiblob]

I get this too, I have checked out a commit from around this date, can't remember which one as today I noticed I had for some reason changed to latest commit, so I had to do a checkout again, this one doesn't seem to be stopping when batch generating images, at least it hasn't so far: