-
Notifications
You must be signed in to change notification settings - Fork 7
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Request for Benchmarks #10
Comments
Unfortunately, I think my computer's performance is not good enough to test benchmarking Only Visual Studio 2022 and cmd are on In my case, it would be better to lower the .onnx specifications. **Additionally #9 updated the best combination in my environment Also, I tested rife_cuda, and it seems that it can be used if it is the same as the existing my 60fps setting Best setting results for my environmentrife_cuda testcustom file ( i5 13600k + 4070 ti ) |
NVIDIA GeForce RTX 4090 |
7950x(No PBO) + 5600Mhz DDR5(2x) + RTX4080
|
Are you referring to 480x360 2x (2x_SuperUltraCompact) 1520 fps vs 1840 fps? Framerate is extremely high here so the difference is likely due to CPU and other factors, not GPU itself. This benchmark runs in under 1 second so there's huge variability. I tried a few runs and it's giving me results ranging from 1450 to 1750. |
Yes of course CPU and RAM speed would affect the speed. And you can check the wiki page to see more output. |
Thanks to everyone who has been sharing benchmarks, please keep them coming. Based on some of those results it seems like it might be useful to have another model with performance in between UltraCompact and SuperUltraCompact. For example the 3070 and 3070 Ti are barely unable to run UltraCompact on 23.796 fps 1080p video smoothly in mpv but they can run SuperUltraCompact easily. I'll look into it to see if it can be done easily. |
Based on 3070 and 3070 ti, it would be good to have the following rules. If the 3070 is aiming to play 1080p60 I think it would be better to design the model assuming resize from the start. Since the 1080p60 video comes out at 52.65 fps even with SuperUltraCompact anyway, you should also think about the situation of resizing. I recommend designing assuming 1080p is resized to 720p. [720p-3070]
If you lower the performance slightly on UltraCompact, it will work smoothly at 60fps. The following up-scaling methods are likely to be possible in 3070. 30fps 60fps |
@the-database, I think it would be useful to add 6x 480x360 (or 8x) benchmarks for those people who are going to upscale old/low resolution anime into 4K. I personally have 4K screen so I might encounter such situations. Perhaps you're expecting to upscale 4x and then rely on more simple upscaling algorithms? |
NVIDIA GeForce RTX 3070 TI |
NVIDIA GeForce RTX 4070 TI
|
RTX 3060 12GB
|
Intel i5-10500H
|
i5 12600k
i5 12600k
|
NVIDIA GeForce RTX 2080 Ti
|
I'm collecting benchmarks for different hardware configurations on the wiki here: https://github.com/the-database/mpv-upscale-2x_animejanai/wiki/Benchmarks
If you would like to contribute your benchmarks, please try downloading the prerelease archive 1.0.1 here: https://github.com/the-database/mpv-upscale-2x_animejanai/releases/tag/1.0.1
To run the benchmark, extract the release archive and then run the bat file at:
mpv-upscale-2x_animejanai\portable_config\shaders\animejanai_v2_benchmark_all.bat
. If running the bat just opens and closes the console Window immediately, an error has occurred. If that happens, please open a command window, navigate to thempv-upscale-2x_animejanai\portable_config\shaders
directory, and then run this command:.\animejanai_v2_benchmark_all.bat
. Any errors that occurred should be printed to the console, please share them here.The text was updated successfully, but these errors were encountered: