You are not logged in. Please login or register.
SmoothVideo Project → Using SVP → New RIFE filter - 3x faster AI interpolation possible in SVP!!!
aloola
> I just want to report that I got 30fps on 4k transcoding
and you're capped by the encoder speed
try h264 instead
or set preset=fast in 'advanced' mode, cause it's set to "slow" by default
aloola
> I just want to report that I got 30fps on 4k transcodingand you're capped by the encoder speed
try h264 instead
it did 10% better 
it did 10% better
Why not try even faster settings?
ultrafast
superfast
veryfast
faster
Chainik wrote:aloola
> I just want to report that I got 30fps on 4k transcodingand you're capped by the encoder speed
try h264 insteadit did 10% better
do you have discord ? please show me your profile setup .. you said 3 gpu threads, SVP allowes 1,2 and 4
i tried to replicate what you did .. advanced encoding with preset fast:
23:53:34.822: ===== Starting mpv ======
23:53:34.822: Command line: C:\Program Files (x86)\SVP 4\mpv64\mpv.exe D:/The Northman (2022).mkv --o=D://The Northman (2022).SVP.temporary.mkv --no-audio --no-sub --no-sub-auto --input-ipc-server=mpvencodepipe --input-media-keys=no --no-msg-color --vf=vapoursynth:[C:\Users\patri\AppData\Roaming\SVP4\scripts\ffff.py]:4:12 --of=matroska --ovc=h264_nvenc --ovcopts=b=10000000,preset=slow,profile=high,rc=vbr,maxrate=13000000,bufsize=20000000,time_base=1000/47952,threads=12
23:53:34.879: (+) Video --vid=1 (hevc 3840x2160 23.976fps)
23:53:34.879: Audio --aid=1 --alang=deu (*) 'Surround 7.1' (truehd 8ch 48000Hz)
23:53:34.879: Audio --aid=2 --alang=eng 'Surround 7.1' (truehd 8ch 48000Hz)
23:53:34.879: Audio --aid=3 --alang=eng 'Stereo' (ac3 2ch 48000Hz)
23:53:34.879: Subs --sid=1 --slang=deu (hdmv_pgs_subtitle)
23:53:34.879: Subs --sid=2 --slang=deu (hdmv_pgs_subtitle)
23:53:34.879: Subs --sid=3 --slang=deu (hdmv_pgs_subtitle)
23:53:34.879: Subs --sid=4 --slang=eng (hdmv_pgs_subtitle)
23:53:34.879: Subs --sid=5 --slang=eng (hdmv_pgs_subtitle)
23:53:34.879: Subs --sid=6 --slang=eng (hdmv_pgs_subtitle)
23:53:34.879: File tags:
23:53:34.879: Title: The Northman (2022)
23:53:35.954: vstrt: TensorRT version mismatch, built with 2135 but loaded with 2136; continue but fingers crossed...
23:53:36.859: CUDA lazy loading is not enabled. Enabling it can significantly reduce device memory usage. See `CUDA_MODULE_LOADING` in https://docs.nvidia.com/cuda/cuda-c-pro … l#env-vars
23:53:36.941: CUDA lazy loading is not enabled. Enabling it can significantly reduce device memory usage. See `CUDA_MODULE_LOADING` in https://docs.nvidia.com/cuda/cuda-c-pro … l#env-vars
23:53:39.511: [autoconvert] Converting yuv420p10 -> p010
23:53:39.521: VO: [lavc] 3840x2160 p010
23:53:39.521: [vo/lavc] Opening encoder: NVIDIA NVENC H.264 encoder [h264_nvenc]
23:53:39.725: [ffmpeg] h264_nvenc: 10 bit encode not supported
23:53:39.749: [ffmpeg] h264_nvenc: No capable devices found
23:53:39.749: [vo/lavc] Could not initialize encoder.
23:53:39.749: Could not initialize video chain.
23:53:39.947: Video: no video
23:53:39.947: [encode] no data written to target file
23:53:39.947: Exiting... (Interrupted by error)
23:53:40.128: (!!!) Intermediate file may be broken: D:\\The Northman (2022).SVP.temporary.mkv
23:53:40.128: ===== mpv exited with code 2 =====
*edit* error found .. movies was 10bit and ... this does not work. so i have to use convert to 8bit or .. use h265 instead, see below
Thanks, will invest later tonight how to apply this
I hope a solution can be found. Looking at what the 3070Ti card can do, I'm very curious to see what will be achieved with the 4090. Bottlenecks will probably appear somewhere and if they can be identified then the potential for performance gains is huge. You are blazing a new trail, the next ones after you will find it easier 
Transcoding is completely different to MPV playback ... Transcoding H265 with profile fast is utilizing my CPU 100% .. playback not.
Tested this out with an RTX 4090 and seem to be getting around 115fps on a 1080p video. So much better than the default implementation. Used to only get around 80fps with the default
It is good that there are more 4090 card owners on this forum. It will be easier to compare results
80 to 115 is same result for me .. i would be happy to have a discord evening even curious with the rtx3070ti results shown above ... and 3 gpu threads as i cannot select odd numbers 
80 to 115 is same result for me .. i would be happy to have a discord evening
you can edit to 3 threads here
encoder log
@dlr5668 he also has 3070ti, we may need him to test to compare the results.
Did you rename the profile to the name of the Ride default profile Name ? Generic 4.4 profile is odd as the 4.6 is used in vs-mlrt
Did you rename the profile to the name of the Ride default profile Name ? Generic 4.4 profile is odd as the 4.6 is used in vs-mlrt
like I said before, that model does nothing with TRT mode on.trt model Default is 4.6.
you can try to switch to model 4.4, edit base.py file 
It seems you run the SVP basic RIFE and not vs-mlrt as introduced here: https://www.svp-team.com/forum/viewtopi … 483#p81483
correct ?
anyhow, i tried the basic Rife implementation and nothing changed with 3 threads ..
which nvidia driver version are you using ?
i run windows 11 with the latest driver from january 5th, 528.02
mine also 528.02
It seems you run the SVP basic RIFE and not vs-mlrt as introduced here: https://www.svp-team.com/forum/viewtopi … 483#p81483
I follow that instruction. so vs-mlrt
proof I run trt ver. %AppData%\Roaming\SVP4\cache\Program Files (x86)\SVP 4\rife\models\rife
here is benchmark with TRT off, it's way slower
also make sure these files in the right place
2. replace generate.js, base.py in SVP 4\script; restart SVP to be sure
maybe you copy these files to SVP 4\MPV64
Maybe an indication that TRT is not used in my setup .. are you in discord or TeamViewer etc tomorrow? Going to bed now
mine also 528.02
It seems you run the SVP basic RIFE and not vs-mlrt as introduced here: https://www.svp-team.com/forum/viewtopi … 483#p81483
I follow that instruction. so vs-mlrt
proof I run trt ver. %AppData%\Roaming\SVP4\cache\Program Files (x86)\SVP 4\rife\models\rifehere is benchmark with TRT off, it's way slower
also make sure these files in the right place
2. replace generate.js, base.py in SVP 4\script; restart SVP to be sure
maybe you copy these files to SVP 4\MPV64
I followed all of that as well. I just did everything again to make sure and I get the same result. up to around 115fps. Changing to 4.4 doesn't change anything either. All of my other specs are the same as yours I think. 5900X, 32Gb 3600mhzCL14
Everything I've seen suggested so far, hasn't changed the speed. It always gets capped at around 60-75% usage like this for the entire duration
Transcoding with TRT ON or OFF seems to have close performance when we look to FPS, but with TRT ON the power consumption is about 25% less than TRT OFF. So, I think there is a margin for optimizations.
TRT ON
TRT OFF
That one - https://github.com/AmusementClub/vs-mlrt - is MUUUUCH better
-------------
Let's try this:0. nothing to do with Python
1. download --> https://www.svp-team.com/files/temp/rife-trt-0701-1.7z <--, unpack into SVP 4\rife (you should already have this folder), so vstrt.dll must be in the root, i.e. SVP 4\rife\vstrt.dll
2. replace generate.js, base.py in SVP 4\script; restart SVP to be sure
3. menu -> Applications settings -> Additional options -> All settings, go to 'User defined options'
Title: TensortRT; Script name: rife_trt; Other values by default - "FRC profile", "ON or OFF"
Click 'Add option'
Go to the RIFE video profile, see the added TensortRT on/off swtich.
Make a copy of the RIFE video profile, one with TensorRT ON and another with OFF.Pros:
- no Pytorch!
- works in a real-time
- even faster (?) than vsrife [could be just because of some TensorRT options]
Cons:
- even slower (?) first-time initialization [could be just because of some TensorRT options]
- a nasty command-line window will pop-up for every new video resolution
I'm getting 150fps transcoding with 1080p content with this (4090), this is really good! I'm using 4 gpu threads.
Real time is still slow and choppy but I haven't taken the time to check that everything's properly set-up, I will do it tomorrow.
I see no performance change with the TensorRT setting - so i am still thinking it is not running with Tensorcores
if it creates cache files - then it's working
i posted some benchmark data with a sample script: https://github.com/AmusementClub/vs-mlr … nt-4630465
numbers should be valid as someone else with RTX4080 is half performance
the parameter peak_performance = True increases performance by ~60%
thanks for the parameter feedback@github
i posted some benchmark data with a sample script: https://github.com/AmusementClub/vs-mlr … nt-4630465
numbers should be valid as someone else with RTX4080 is half performancethe parameter peak_performance = True increases performance by ~60%
thanks for the parameter feedback@github
peak_performance is just for the bench, right ?
Is it safe to install the new RIFE stuff for SVP now? Chainik was saying that real-time was not really working, but I see you guys have been troubleshooting over the weekend. Would like to know if real-time interpolation works with the latest updates.
Depends on your resolution .. 1080p runs without any problem .. 2k as well ... 4k is still not fast enough
Depends on your resolution .. 1080p runs without any problem .. 2k as well ... 4k is still not fast enough
I have an RTX 3070 145 W and only have a 1080p TV.
=== RIFE / PyTorch+TensorRT installation ===
!!! UPDATED 07-01-2023 !!!
- replaced base.py : fp16 isntead of fp32
+ added fix for vsrife
============!!! Advanced users only, if you don't know what this is all about - you don't need it !!!
0. Update SVP via online updater (this will update Python to ver. 3.10.9)
1. Dependencies installation:
--------------
Download TensorRT libs: https://github.com/HolyWu/vs-rife/relea … 2_win64.7z
unpack CUDA-blah-blah.7z to SVP 4\mpv64run cmd as Administrator
cd SVP 4\mpv64
python get-pip.py python -m pip install torch torchvision --extra-index-url https://download.pytorch.org/whl/cu117 python -m pip install tensorrt-8.5.2.2-cp310-none-win_amd64.whl python -m pip install torch-tensorrt-fx-only tqdm python -m pip install -U --no-deps vsrife python -m vsrifeA quick fix for vsrife's "slowmo" bug: replace __init__.py in SVP 4\mpv64\Lib\site-packages\vsrife\
2. SVP modification
---------------
Replace SVP 4\script\generate.js and base.py with the attached onesmenu -> Applications settings -> Additional options -> All settings, go to 'User defined options'
Title: TensortRT; Script name: rife_trt; Other values by default - "FRC profile", "ON or OFF"
Click 'Add option'Go to the RIFE video profile, see the added TensortRT on/off swtich.
Make a copy of the RIFE video profile, one with TensorRT ON and another with OFF.----------------
For a given video resolution the first-time start is super-slow (while it generates TRT net in a cache folder --> %APPDATA%\SVP4\cache\trt), so better do it in a "transcoding" mode.
Do I follow this guide? Or is there a more updated one?
SmoothVideo Project → Using SVP → New RIFE filter - 3x faster AI interpolation possible in SVP!!!
Powered by PunBB, supported by Informer Technologies, Inc.