You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This is not very scientific, as it was just a simple prompt passed and I did not make any captures for proof yet.
But I chose to install Linux (not WSL or virtualization, but a full blown Linux installation on my NVMe).
I used a specific model and a prompt consisting of 6 parameters.
On Linux Mint v21.1, generation (after priming the model) took 8 seconds with 20 steps standard settings.
With face restoration and 35 steps, it took 16 seconds.
On Windows 11, the first trial took 5 seconds.
With face restoration and 35 steps, took 9 seconds.
For this trial, the same extensions were installed as well as the same Pyhton module versions and CUDA versions.
This is pretty crazy.
BTW, I use a Geforce 3060 OC with the same OC settings mounted in an ASUS Prime H670-PLUS d4 with an i7-13700K processor and 32 gigabytes of DDR-4400 RAM.
The Linux nVidia drivers were binaries provided by nVidia (the runfile) and CUDA was also installed from there, not from apt.
What do you think? I think we need to do some more intensive testing here.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
-
This is not very scientific, as it was just a simple prompt passed and I did not make any captures for proof yet.
But I chose to install Linux (not WSL or virtualization, but a full blown Linux installation on my NVMe).
I used a specific model and a prompt consisting of 6 parameters.
On Linux Mint v21.1, generation (after priming the model) took 8 seconds with 20 steps standard settings.
With face restoration and 35 steps, it took 16 seconds.
On Windows 11, the first trial took 5 seconds.
With face restoration and 35 steps, took 9 seconds.
For this trial, the same extensions were installed as well as the same Pyhton module versions and CUDA versions.
This is pretty crazy.
BTW, I use a Geforce 3060 OC with the same OC settings mounted in an ASUS Prime H670-PLUS d4 with an i7-13700K processor and 32 gigabytes of DDR-4400 RAM.
The Linux nVidia drivers were binaries provided by nVidia (the runfile) and CUDA was also installed from there, not from apt.
What do you think? I think we need to do some more intensive testing here.
Beta Was this translation helpful? Give feedback.
All reactions