-
Both the nvjpeg encoder and decoder define a `get_device_context`, `nvenc` does its own thing in `init_module`..
It's all unnecessarily messy.
Also, the cuda context object could be cached and r…
-
-
I'm cautiously updating cuda from the package managers version (11.5) to the required minimum (11.8+ ... I think I'm getting 12.3.2...).
I'm first removing cuda like this:
```sudo apt-get --purg…
-
### Version
0.3.0
### Describe the bug.
When encoding an image from an nparray to jpeg - result is incorrect, image is offset to the righ, colors are 'separated'.
The exact image, if saved to jp…
-
### 🐛 Describe the bug
Steps to reproduce:
1) Download TorchVision 0.11.1 windows wheel from https://download.pytorch.org/whl/cu102/torchvision/
2) Unpack the binary
3) Run `dumpbin /dependents im…
-
**Describe the bug**
Threaded Lock Contention while threaded-clean thread is cleaning.
So that NVENC failed and fallback (see below)
Reason why I notice that:
See https://github.com/orgs/Xpra-or…
-
See #2052 and b7647530b151640888f71e3c4e8c2499afd77bc3
We should combine the scaling, re-striding and format conversion.
Also support `swscale` since Debian still does not package `libyuv`.
Aft…
-
I think that another good Sponsor feature would be to enable thumbnails generation with Nvidia GPU.
Even if your server's many cores, it's still more efficient to use a GPU which has more and more…
-
the problem is that can not load DLL"nppi64_80". when i try to debug,the JPEGCompression will trigger exception
-
i cannot build aur python-torchvision-rocm in manjaro because
Emitting ninja build file /var/tmp/pamac-build-chameleon/python-torchvision-rocm/src/vision-0.13.1/build/temp.linux-x86_64-3.10/build.n…