The Parallel Video Processor is a utility that allows you to create a cluster of PCs and use it to collaboratively encode videos with x264 and/or x265.
That means that if you have more than one PCs at home or a few friends with PCs, you can collaboratively encode a video; that of course means that the video will be encoded much faster. More PCs in the cluster means more processing power and a faster encode.
The official webpage for it is located at: http://parallel-video-processor.blogspot.com. There's also a quickstart guide there (at the left hand corner, under Pages, Quickstart).
You can download the latest release from my Dropbox folder.
A few gotchas before you start:
a) Always check my website for the MD5 and SHA1 checksums of the pvp.exe file before running it.
b) It has the following dependencies that you should already have on your PC before running it: FFMPEG, FFProbe, x264 and/or x265.
c) PvP will need access to your network, the ability to create, edit, delete files and execute external programs (FFMPEG, FFProbe, x264 and x265). That means that you'll need to put it in your firewall's trusted applications. Otherwise it will be extremely slow (especially during sanity checks) and it will probably not work very well.
d) If your video file is an avisynth script, make sure that each node gets a different copy of the script. That is because when an avisynth script is detected, PvP will try to edit it so that the node processes the amount of data that it's supposed to (usually it tries to do so through FFMPEG's -ss and -t or through x264/x265 --seek and --frames). So if for example you are using an avisynth script over a network share, you should just create as many copies of the scripts as the number of nodes and assign a unique script per node.
e) If you are using benchmarking to sort out how much processing each node should perform, ALWAYS convert it to weights before proceeding to the actual encoding. Benchmarking is somewhat volatile. A system might receive a score of 13 one time and the next 12 or 14. That means that if your encoding has failed, there is a chance that the calculation of video processing will not be the same as the time before, resulting in a period of video missing or it being duplicate.
f) Using too many x264 processes on the same system (i.e. having more than one nodes on one PC) without using the FFMPEG pipe can result to the x264 encoder getting stuck. I haven't had too much time to work on this, but it appears that the culprit here is DirectShow.
g) It can run on any PC with an operating system that is Windows XP or newer.
- Increases the speed of your encodes
- It can save you money; 5 cheapo laptops in collaboration can put expensive PCs to shame
- 4 different cluster node weight methods: Pre-set, Manual, Equal, Benchmark
- Autocomplete for x264, x265, FFMPEG and any previously-entered settings that the user might need to tweak (such as a cluster node IP address, username, password, etc.)
- Video settings autodetection
- Security: Cluster nodes require authentication and will only accept connections from specific IPs
- Recovery mode: If some nodes have failed, crashed or had a temporary network disruption, they can individually be set into recovery mode and re-encode just a part of the video instead of having to redo everything from the start
- A Swiss Army knife for Video Encoding: You can create a cluster of nodes on the same PC. This can be used for a number of reasons; the most popular being bypassing the OS memory restrictions per process (typically 2GB per process for 32-bit windows and 4GB for 64-bit windows); Another reason would be that you can save a lot of time just by using it. After you've used it once and saved your settings, all you need to do to start another encode with the same settings would be to just load your settings and go (in theory, that's just 2 button presses); Autocomplete for x264 and x265 settings goes a long way too in saving your time instead of just typing away; Yet another reason is to facilitate x265 encoding. Just select your input video file and PvP will do the rest. No need for your video file to be YUV or Y4M.
Please let me know of any bugs, issues, questions, feature requests, etc.
+ Reply to Thread
Results 1 to 1 of 1