Need little assist for setting in Vlad's webui

[josh5671]

Kinda use automatic1111's webui from day 1 it appear and switching to Vlad's webui today.
(have 0 knowledge on python, might be not sure what i am talking about, please correct me if i am wrong)

I am running 3070 with Torch2+CU118 on 1111, with --no-half-vae --opt-sdp-no-mem-attention
I able to generate 3 batch size 30 step 480x640 and 20 step highres 960x1280 (i might be doing it wrong, but so far i always doing this)

On Vlad i unable to generate with same setting.
It just straight out of vram.

╭───────────────────────────────────────── Traceback (most recent call last) ──────────────────────────────────────────╮
│ E:\automatic\modules\call_queue.py:58 in f │
│ │
│ 57 │ │ │ │ pr.enable() │
│ ❱ 58 │ │ │ res = list(func(*args, **kwargs)) │
│ 59 │ │ │ if shared.cmd_opts.profile: │
│ │
│ E:\automatic\modules\call_queue.py:38 in f │
│ │
│ 37 │ │ │ try: │
│ ❱ 38 │ │ │ │ res = func(*args, **kwargs) │
│ 39 │ │ │ finally: │
│ │
│ ... 13 frames hidden ... │
│ │
│ E:\automatic\repositories\stable-diffusion-stability-ai\ldm\modules\diffusionmodules\model.py:132 in forward │
│ │
│ 131 │ │ h = self.norm1(h) │
│ ❱ 132 │ │ h = nonlinearity(h) │
│ 133 │ │ h = self.conv1(h) │
│ │
│ E:\automatic\venv\lib\site-packages\torch\nn\functional.py:2059 in silu │
│ │
│ 2058 │ │ return torch._C.nn.silu(input) │
│ ❱ 2059 │ return torch._C._nn.silu(input) │
│ 2060 │
╰──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
OutOfMemoryError: CUDA out of memory. Tried to allocate 1.76 GiB (GPU 0; 8.00 GiB total capacity; 5.95 GiB already
allocated; 0 bytes free; 6.52 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting
max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

And time to time i have seen it try to allocate 10gb vram while my gpu only have 8gb.

PS: any extra detail i gonna to provide so you guys able to provide asist?

[vladmandic]

--no-half-vae and --opt-sdp-no-mem-attention don't exist here, you need to set values via UI settings.

[josh5671]

thanks for almost instantly reply.

yes, the no-half-vae and opt-sdp-no-mem-attention has been set in Cuda Setting
the cuda memory run out still there.

[vladmandic]

this may be just borderline - you were running very near the limit and torch 2.0 is just a little bit more memory hungry. the allocation values do not seem horrible on their own.

[josh5671]

I see. kinda agree i touch the hard borderline of your of your webui.
For now will switch back to 1111's webui (just copy the model,extension and output back), will come back a week or weeks later, since Vlad you literally update it daily or between days.
Maybe my next try, all will be working fine.