Computer Impulse Buy / AI Ganbang / Soaking Discussion

please clone poony

how long would this take?

1 Like

Ok!

Tue Jul 8 14:36:35 2025
±----------------------------------------------------------------------------------------+
| NVIDIA-SMI 576.80 Driver Version: 576.80 CUDA Version: 12.9 |
|-----------------------------------------±-----------------------±---------------------+
| GPU Name Driver-Model | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 NVIDIA GeForce RTX 5090 WDDM | 00000000:01:00.0 On | N/A |
| 32% 35C P0 71W / 575W | 3624MiB / 32607MiB | 0% Default |
| | | N/A |
±----------------------------------------±-----------------------±---------------------+

±----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| 0 N/A N/A 2372 C+G …s\Mozilla Firefox\firefox.exe N/A |
| 0 N/A N/A 3172 C+G …s\Mozilla Firefox\firefox.exe N/A |
| 0 N/A N/A 9536 C+G …ntrolPanel\SystemSettings.exe N/A |
| 0 N/A N/A 10248 C+G …yApp\MicrosoftSecurityApp.exe N/A |
| 0 N/A N/A 12036 C+G …erControlCenter_v2\Killer.exe N/A |
| 0 N/A N/A 12472 C+G C:\Windows\explorer.exe N/A |
| 0 N/A N/A 13004 C+G …0.3296.93\msedgewebview2.exe N/A |
| 0 N/A N/A 13124 C+G …indows\System32\ShellHost.exe N/A |
| 0 N/A N/A 14460 C+G …y\StartMenuExperienceHost.exe N/A |
| 0 N/A N/A 18308 C+G …2txyewy\CrossDeviceResume.exe N/A |
| 0 N/A N/A 19352 C+G …_cw5n1h2txyewy\SearchHost.exe N/A |
| 0 N/A N/A 19356 C+G …App_cw5n1h2txyewy\LockApp.exe N/A |
| 0 N/A N/A 23124 C+G …em32\ApplicationFrameHost.exe N/A |
| 0 N/A N/A 25136 C+G …0.3296.93\msedgewebview2.exe N/A |
| 0 N/A N/A 25396 C+G …are\IntelGraphicsSoftware.exe N/A |
| 0 N/A N/A 26456 C+G …8bbwe\PhoneExperienceHost.exe N/A |
| 0 N/A N/A 26560 C+G …\SubAgent\AlienFXSubAgent.exe N/A |
| 0 N/A N/A 28096 C+G …yb3d8bbwe\WindowsTerminal.exe N/A |
| 0 N/A N/A 29356 C+G …xyewy\ShellExperienceHost.exe N/A |
| 0 N/A N/A 31500 C+G …SnippingTool\SnippingTool.exe N/A |
±----------------------------------------------------------------------------------------+

I’m growing a new poon clone in her cheek. Isn’t filler blob!

Will hatch out on 1000th play of kinderpoon

1 Like

true ur too up-to-date last version pytorch support 12.8 https://pytorch.org/get-started/locally/
pozz downgrade :sunglasses:

1 Like

Wait, am I…

in the future?

usually unstable version is horrible so we fight a bit n then we juzt downgrade

then

pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128

gud luck :sunglasses:

1 Like

porny poony?

Will try these things after I procrastinate a bit. :sunglasses:

Thank you!

1 Like

usually unstable version is horrible so we fight a bit n then we juzt downgrade

then

pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128

gud luck :sunglasses:

1 Like

All these squiggles make me think I’m speaking in tongues to a computer god but probably need to learn to raw dog this stuff without comfyui.

comfyui too complicated for moi :sunglasses:

1 Like

Installing python sounds like something _____ do by the bins.

1 Like

backdoor installation an extra $5

1 Like

idk more research is needed but that shit getting pricey plus she’s a bitch so idk y’all.

Maybe the backdoor services are indeed a mere $5 more.

@knar I think its working. I do dis.

Here’s the updated, end-to-end recap from where we left off:

  1. Upgraded to CUDA 12.8 PyTorch wheels

    • Installed torch==2.7.1+cu128, torchvision==0.22.1+cu128, torchaudio==2.7.1+cu128.
  2. Rebuilt the OpenMMLab stack for that exact build

    • Uninstalled old mmcv-full, mmengine, xformers.
    • Installed pure-Python mmengine.
    • Installed (or built) mmcv-full against CUDA 12.8 until mmcv.__version__ reported 2.2.0 with CUDA support.
    • Built xformers from source (but it only exposed float16/bfloat16 kernels).
  3. Verified core imports

    python -c "import torch, torchvision, mmcv, mmengine; print(torch.__version__, torch.cuda.is_available(), torchvision.__version__, mmcv.__version__)"
    

    → All loaded, CUDA available.

  4. First GPU test run

    • Launched test_DLoRAL.py… -i inputs -o outputs.
    • Crashed inside .enable_xformers_memory_efficient_attention() because no float32 xFormers operator existed.
  5. Disabled the xFormers call

    • Commented out or wrapped

      self.unet.enable_xformers_memory_efficient_attention()
      

      in src/DLoRAL_model.py so it falls back to the standard attention kernels.

It’s spitting out as a trillion .pngs damn. Gonna have to poop them back into video. But still interesting as fuck.

png= poon nude graphic

In :tm: dreams

1 Like

damn… it failed cuz things are pointed to wrong things. Man! this way more labor intensive than just squinting when there is bad video.

Haha I dun haff explicit dreams

Um…. Maybe Martyrz needs an upscale?
Gosh I need to rewatch it…. Def one of the greatest fucked up films ever yo.

Shot on Cooke S4, makes sense.
Them bitches high end as fuck…. And put on a super 16 format for that grain grit on da negative, it’s not post-processing.
Ooooo damn

Ok I do this and it go now.