RipBot264 v1.18.3 - Simple and easy to use GUI -> IPOD . PSP . CONSOLES . BLURAY

[Atak_Snajpera]

You must use x86 version of KLNMeansCL due to installed AviSynth x86.
Run http://developer.download.nvidia.com/compute/DevZone/OpenCL/Projects/oclDeviceQuery.zip to see if hd5500 has proper support for OpenCL

[kensiko35]

Hey Atak, I was shown your application from another forum guy. This is pretty amazing. I used HandBrake for a while now simply because it is much faster than many others, this is what really matters to me, speed !

I didn't try your application yet, I don't have a second powerful computer at home. But I was wandering if I should buy a Intel 8 cores CPU or get another Intel quad core. My concern is about bandwidth, I'm on powerline Ethernet here all over my house. It would be great if I could put the other computer in the garage, but I will not get more than 2 MB per second (bytes, not bites). It is pretty reliable though, unlike wifi.:thanks:

[Atak_Snajpera]

It would be great if I could put the other computer in the garage, but I will not get more than 2 MB per second (bytes, not bites). It is pretty reliable though, unlike wifi.
DE mode works fine even via Wi-Fi . If you can stream (watch) video file without any fps drops via your network then it should be all ok.

I used HandBrake for a while now simply because it is much faster than many others, this is what really matters to me, speed !
Encoding preset determines speed.

[kensiko35]

DE mode works fine even via Wi-Fi . If you can stream (watch) video file without any fps drops via your network then it should be all ok.


Encoding preset determines speed.
Thanks.

So now as I see your app can use the gpu to encode ? I have a hd7800, would it be much faster than my 2500k @ 4.5 ghz ?

[dv8r]

I have 4 nodes on ethernet and I will put my laptop online via wifi at times. The wifi usually takes about 5-10 seconds longer to load the file. Not too shabby at all. Not when you consider the amount of time you're saving having multiple machines working on a file. It's a great program. I'm doing a x265 encode now at 45 fps with 1 ancient and 3 older machines. Can't beat that.

[omegaeel]

I love this program, been using it for over a year. But lately I've been having an issue with the error text "Source file could not be indexed". It only comes up with DE enabled.

I've had this happen with v1.19.3, with mp4s and mkvs, Windows 10 and Windows 7. I've tried switching to a different Temp folder through Ripbot264.ini, but this seems to be ignored. I'm running it as Administrator, I can access the shared Temp from other PCs on the LAN.

What am I missing?

[Atak_Snajpera]

I love this program, been using it for over a year. But lately I've been having an issue with the error text "Source file could not be indexed". It only comes up with DE enabled.

I've had this happen with v1.19.3, with mp4s and mkvs, Windows 10 and Windows 7. I've tried switching to a different Temp folder through Ripbot264.ini, but this seems to be ignored. I'm running it as Administrator, I can access the shared Temp from other PCs on the LAN.

What am I missing?
When you get that message run this in command console and see what happens. (post screenshot)

Change X to correct drive letter

"\\[YOUR-PC-NAME]\Ripbot264temp\Tools\AviSynth plugins\ffms\ffmsindex.exe" -f -k "X:\temp\Ripbot264temp\Job1\video.mkv"'

So now as I see your app can use the gpu to encode ?
GPU is only used for denoising. Encoder runs purely on CPU.

[omegaeel]

When you get that message run this in command console and see what happens. (post screenshot)

Change X to correct drive letter

"\\[YOUR-PC-NAME]\Ripbot264temp\Tools\AviSynth plugins\ffms\ffmsindex.exe" -f -k "X:\temp\Ripbot264temp\Job1\video.mkv"'


GPU is only used for denoising. Encoder runs purely on CPU.

Screen cap: https://www.dropbox.com/s/g5obpoqsyjf81gw/screencap.png?dl=0

The command did create "video.mkv.ffindex", I attempted to rerun the encode and this time a window popped up saying it could not open "info.txt", even though I saw it was there when I ran ffmsindex. In the directory "info.txt" was now gone.

[guest]

I see that Khanattila has posted an updated version.

[Atak_Snajpera]

Screen cap: https://www.dropbox.com/s/g5obpoqsyjf81gw/screencap.png?dl=0

The command did create "video.mkv.ffindex", I attempted to rerun the encode and this time a window popped up saying it could not open "info.txt", even though I saw it was there when I ran ffmsindex. In the directory "info.txt" was now gone.

I asked you to use specific NETWORK path ("\\[YOUR-PC-NAME]\...). You used local path.

[bcn_246]

Just wondered, when using networked encoding is the resulting quality any lower than single machine? How exactly is the encode spread (I assume it is by segmenting the video).

[Atak_Snajpera]

Just wondered, when using networked encoding is the resulting quality any lower than single machine?
In CQ mode quality is the same. In 2-pass mode should be also almost the same due to added Bitrate Distribution Optimization
https://www.doom9.org/showthread.php?p=1729404#post1729404

[billqs]

Can someone tell me where the logs are kept. I'm having trouble backing up one of my discs. Previously, I've just had to adjust something in the .ini program and it would work, but I've had no success with it. I keep getting a Demux error.

[Atak_Snajpera]

go to job folder and locate demuxlog.txt

[billqs]

go to job folder and locate demuxlog.txt

I've attached the demuxlog.txt. Looks like there may be trouble reading the source, unless that points to something else.

Thanks for all your help! I used your program for several years and it's been a lifesaver!

EDIT: I tried cleaning the media even though it appears clean. I also tried another blu-ray drive in my PC, still getting same results.
EDIT: For some reason it didn't show the log I had uploaded so here it is:
eac3to v3.31
command line: "C:\Program Files (x86)\Ripbot 264\Tools\eac3to\eac3to.exe" "D:\" 1) 2: "E:\Temp\RipBot264temp\job1\video.mkv" -seekToIFrames 3: "E:\Temp\RipBot264temp\job1\audio_1_English.core.dts" -core 5: "E:\Temp\RipBot264temp\job1\5_subtitles_English_1080.sup" 6: "E:\Temp\RipBot264temp\job1\6_subtitles_Spanish_1080.sup" 1: "E:\Temp\RipBot264temp\job1\chapters.txt" -progressnumbers -log="E:\Temp\RipBot264temp\job1\demuxlog.txt"
------------------------------------------------------------------------------
M2TS, 1 video track, 2 audio tracks, 2 subtitle tracks, 1:36:53, 24p /1.001
1: Chapters, 24 chapters
2: h264/AVC, 1080p24 /1.001 (16:9)
3: DTS Master Audio, English, 5.1 channels, 24 bits, 48kHz
(core: DTS, 5.1 channels, 1509kbps, 48kHz)
4: AC3, English, 2.0 channels, 224kbps, 48kHz
5: Subtitle (PGS), English
6: Subtitle (PGS), Spanish
Creating file "E:\Temp\RipBot264temp\job1\chapters.txt"...
[s05] Extracting subtitle track number 5...
[s06] Extracting subtitle track number 6...
[v02] Extracting video track number 2...
[a03] Extracting audio track number 3...
[a03] Extracting DTS core...
[v02] Muxing video to Matroska...
[a03] Creating file "E:\Temp\RipBot264temp\job1\audio_1_English.core.dts"...
[s05] Creating file "E:\Temp\RipBot264temp\job1\5_subtitles_English_1080.sup"...
[s06] Creating file "E:\Temp\RipBot264temp\job1\6_subtitles_Spanish_1080.sup"...
Reading the source file failed. <ERROR>
Aborted at file position 4269801472. <ERROR>

[omegaeel]

I asked you to use specific NETWORK path ("\\[YOUR-PC-NAME]\...). You used local path.

It doesn't work. The Ripbot264temp network share points to the directory with the job1 files, not the shared tools. Ripbot264.ini still lists StoreTempFilesin=TEMP

[kwangkwang]

Hi
I am using 20 cores 40 threads (2 x E5-2660v2) to perform x265 10bit.
Unlike x264 where all cores are fully utilised, the maximum number of cores use for x265 are 16 cores. Increasing number of encoding servers has no effect to use the remaining idle 4 cores.

Is there something that I need to tweak for x265 to use all cores or is it a limit that x265 use max 16 cores?

[mdchaser]

Hi
I am using 20 cores 40 threads (2 x E5-2660v2) to perform x265 10bit.
Unlike x264 where all cores are fully utilised, the maximum number of cores use for x265 are 16 cores. Increasing number of encoding servers has no effect to use the remaining idle 4 cores.

Is there something that I need to tweak for x265 to use all cores or is it a limit that x265 use max 16 cores?

I may be reading this wrong but it sounds like you are trying to use two NUMA nodes on the same machine to encode? If that is the case I would recommend loading two VMs, giving them each 10 or 20 cores and set them to not cross NUMA boundaries. You will get noticeably better performance and should be able to more fully utilize your hardware. Good luck! And might I ask which settings you are using for x265? I'm using CRF 18 with the goal of getting near perfect encodes but I think the scale is different than it was with x264.

[Atak_Snajpera]

Fixed: EncodingClient couldn't copy tools to temp folder if user chose "StoreTempFilesIn=TEMP"

[Ma]

I am using 20 cores 40 threads (2 x E5-2660v2) to perform x265 10bit.
Unlike x264 where all cores are fully utilised, the maximum number of cores use for x265 are 16 cores.

It could be related to commit 10983 https://bitbucket.org/multicoreware/x265/commits/e1adac00dce8e5641cbe9aec3d50a72261c308d9

x265 team tested this patch only on Linux systems. On Windows it is big slow-down change -- on 2 NUMA system with 48 logical cores up to 50% maximum speed.

There are 2 simple solutions:
1) please add '--pools 20,20' option to x265 command line,
2) you can test x265 builds from page www.msystem.waw.pl/x265 (not clean builds which are without changes) -- if you take x265 binaries from yellow background (VS 2015, version >= 1.9+15) you can run it without any changes in options.

This problem is under investigation by x265 team, so please report back what you observe.

[Balthazar2k4]

I haven't used RipBot in ages, but is there some way to stop it from converting TrueHD to FLAC? I only need RipBot to encode video and leave all of the audio alone.

[kwangkwang]

I may be reading this wrong but it sounds like you are trying to use two NUMA nodes on the same machine to encode? If that is the case I would recommend loading two VMs, giving them each 10 or 20 cores and set them to not cross NUMA boundaries. You will get noticeably better performance and should be able to more fully utilize your hardware. Good luck! And might I ask which settings you are using for x265? I'm using CRF 18 with the goal of getting near perfect encodes but I think the scale is different than it was with x264.

Thanks for the idea.

I just sold off the setup. Going to replacing them with 2 machines each running E3-1275Lv3 and E3-1260Lv5 respectively. Might build a third for third DE.

I am using X265 10 bit in CRF17 after comparing a sample scene color histogram with the original. I am not very techy into the tweaking of x26X parameter. Port over most of what x264 parameter had. It is just a single sample scene compare and the histogram matching to the original as close as possible.:)

[kwangkwang]

I haven't used RipBot in ages, but is there some way to stop it from converting TrueHD to FLAC? I only need RipBot to encode video and leave all of the audio alone.

Don't do audio?
mkvmerge the original file audio and new video later.

[billqs]

Well, I don't know why Ripbot wouldn't rip my disc, but Handbrake did without a problem. I much prefer Ripbot, but every so often I get a film Ripbot just won't rip, which is crazy since I believe both program use the same tools.

[dv8r]

Hey Atak, I've got the wake on LAN working nicely now that I figured out the MAC address formats. One issue I'm seeing is ripbot brings all machines online great on the first video but if I have several videos in a batch after the first finishes it will shut down my nodes (as I want it to) but when the next video starts my nodes stay offline. I have it set to put the nodes to sleep after an encode but is there a way to wake them when the next video starts creating chunks? I could set the shutdown method to no action but it is a nice feature to have them shut down as soon as the encodes are done. Thanks

[Atak_Snajpera]

ripbot shutdowns servers only if ALL JOBS in queue are done. Maybe you should disable some power options in windows. Something tells me that windows shutdowns your servers not main pc.

[Ronski]

I haven't used RipBot in ages, but is there some way to stop it from converting TrueHD to FLAC? I only need RipBot to encode video and leave all of the audio alone.

This has caught me out a few times, I now extract just the video with MKV Toolnix, then after RipBot I use it again to merge the audio/subtitles/chapters back in.

[omegaeel]

Fixed: EncodingClient couldn't copy tools to temp folder if user chose "StoreTempFilesIn=TEMP"

Fix worked as far the "Source file could not be indexed" error.

Now a window pops up after indexing and says "Cannot open file" referring to "info.txt"

Screencap: https://www.dropbox.com/s/hcdxmlxeczrb13b/screencap2.png?dl=0

[guest]

Screencap: https://www.dropbox.com/s/hcdxmlxeczrb13b/screencap2.png?dl=0

Not that again :(

[dv8r]

I've went through all my power settings 100 times. They're all the same and I've read a ton on the correct settings for these nodes. I have all the power saver settings off. Crazy thing is 3 of my nodes will wake everytime but one won't. Exact same computers, settings, OS. Bios settings the same etc. I've been watching it like a hawk trying to figure out exactly what the problem is. It's not a huge deal but I'm away from home a lot so I do stuff remotely with teamviewer. Sucks when I'm away and can't get a node to Wake. As for why the nodes shut down immediately after a video is encoded and then won't awake when the next video starts I'm not sure why this is. Maybe it's a Window's deal. Good thing is as long as they're all running the app works great.

[Atak_Snajpera]

I've went through all my power settings 100 times. They're all the same and I've read a ton on the correct settings for these nodes. I have all the power saver settings off. Crazy thing is 3 of my nodes will wake everytime but one won't. Exact same computers, settings, OS. Bios settings the same etc. I've been watching it like a hawk trying to figure out exactly what the problem is. It's not a huge deal but I'm away from home a lot so I do stuff remotely with teamviewer. Sucks when I'm away and can't get a node to Wake. As for why the nodes shut down immediately after a video is encoded and then won't awake when the next video starts I'm not sure why this is. Maybe it's a Window's deal. Good thing is as long as they're all running the app works great.

You should see this when ripbot shutdowns servers
http://i.cubeupload.com/f5SSST.png

Like I said before this only happens if all jobs in queue are done.

If servers shutdown automatically without any command from main pc this would mean that operating system decided to do that for you.
Keep in mind that Wake-ON-Lan works only if pc is phisically connected with router. You can't wake up device which was connected via wi-fi.

[dv8r]

Thanks Atak, I've never seen that on mine. My nodes are on a wired connection. I'll keep going through my network adapter settings and BIOS settings. It has to be something in those. There are a lot of settings that affect wake on LAN I'm finding.

[Ronski]

I once had a pc that used to go to sleep whilst ripbot was running with four cores loaded 100%, never did figure it out.

[guest]

I have an interesting scenario that needs to be solved.
As a way of using KNLMeansCL, and "old fashioned" CUDA support for other programs on the one PC, I have installed a GTX 580,and a GTX 650.
The 580 is for the "old" stuff, and the 650 is to "trick" KNLMeansCL into using the 650, as it's supported.
But alas, it doesn't work...UNLESS the 580 is disabled.
I'm wondering if its something to do with the 580 being the Primary Card in the system !!
Would a BIOS setting or re positioning them on the motherboard would change their "priority"

Any suggestions ???

[Atak_Snajpera]

You can select which device you want to use with this switch. 0 = first device , 1 = second , 2 - third and so on.
http://i.cubeupload.com/P1Hipq.png

[guest]

You can select which device you want to use with this switch. 0 = first device , 1 = second , 2 - third and so on.
http://i.cubeupload.com/P1Hipq.png

Hi Atak, thanks for that, but I could not get it to work without still having to disable the 580 in Device Manager.

However, it does seem to work with the latest version of KNLMeansCL v.0.7.5, (from the other forum).

I downloaded it a few days ago, and included in the download were some "different" opencl.dll's that have to copied into Windows folders, and also a newer version of MS Visual C++ 2015 (links also provided on the other forum)......

Anyway, seems to work for me, so now 2 more questions :-

When using DE with KNLMeansCL, and this particular PC is "helping", what GPU is it going to choose ??, or would it also need that switch ??

And adding that switch, would be required for encoding from that PC, is there a way to make that switch the default setting ??

[dv8r]

I wonder how well opencl on the newer intel processors stacks up to using the gpu. I'm sure the performance is much lower than a graphics card. I've been looking into it since my slimline desktops don't have many options for upgrading gpu due to power requirements. It'd be easier to just upgrade mobo and a processor that supports opencl than trying to get a compatible card and big enough power supply in them. Anyone tried opencl with a newer i7? Either way I don't really understand why knlmeanscl can't be compiled to be backwards compatible with all opencl compliant cards. It makes it confusing with what works and what doesn't. I'm sure it makes it hard to implement into ripbot as well.

[guest]

I wonder how well opencl on the newer intel processors stacks up to using the gpu. I'm sure the performance is much lower than a graphics card. I've been looking into it since my slimline desktops don't have many options for upgrading gpu due to power requirements. It'd be easier to just upgrade mobo and a processor that supports opencl than trying to get a compatible card and big enough power supply in them. Anyone tried opencl with a newer i7? Either way I don't really understand why knlmeanscl can't be compiled to be backwards compatible with all opencl compliant cards. It makes it confusing with what works and what doesn't. I'm sure it makes it hard to implement into ripbot as well.

I'd also like to know how to get an Intel processor to "help"...I do have a couple that support it, but as yet haven't been able to get it to work.

I installed the latest opencl runtime drivers for Intel CPU's (x64), but not sure whether I need to have the x86 drivers as well...and then what ???

I did my first serious run of KNLMeansCL v.0.7.5 today, using a couple of recently acquired GPU's...the best one being an AMD R9 290, and a HD6850, and a GTX650...had them going with DE, and the HD6850 & GTX650 were performing almost identically, but about a 1/4 of what the R9 290 was able to assist.

Now as a comparison, I did a movie with KNLMeansCL, which took about 35 minutes, and then the same movie with as similar settings as I could do, with HQDN3D and it only took about 8 minutes, I am yet to watch both encodes on TV, so as to compare.

And then the same movie with using HQDN3D and x265 with Very Slow, and it was going to take over 4 hours, even with DE on 3 reasonable powerful PC's.

Anyway early days, lot's of fiddle ahead.

We need MORE DE servers, Atak !!!! :)

[Atak_Snajpera]

GTX 650 - 1.4 TFlops
HD6850 - 1.5 TFlops
R290 - 4.8 TFlops

I'm not surprised that r290 is much faster in knlmeanscl. Also keep in mind that that we have here VLIW vs newer GCN architecture.

[guest]

GTX 650 - 1.4 TFlops
HD6850 - 1.5 TFlops
R290 - 4.8 TFlops

I'm not surprised that r290 is much faster in knlmeanscl. Also keep in mind that that we have here VLIW vs newer GCN architecture.

That's interesting info, Atak...where did you get that from ??

What about a GTX 680 ?

Dunno what that other stuff means :(

[Atak_Snajpera]

That's interesting info, Atak...where did you get that from ??

What about a GTX 680 ?

Dunno what that other stuff means :(

From uncle google.

[guest]

So you have a Xeon or 2, have you enabled open cl ?

If so, how, if not, why not ???

Then maybe KNLMeansCL could be used with CPU as well.

[apostolis21]

Hey, I'm trying to use distributed encoding and it keeps encoding the first chunk again and again, any help?

[dv8r]

Good info there Atak, I can't seem to find anything on opencl intel performance, as far as actual numbers. Burt, my bro uses nlmeans a lot with handbrake and one thing that stands out to me with his stuff is the file size will get so low on some movies that I can't believe the quality is as good with such a small filesize. He has more patience than me, and the new high dollar dual xeons. Even then it's SLOW on cpu only. I'm pretty sure xeons don't have any graphics built in so no opencl on them that I know of. Most dual xeon mobo's i've been around have terrible built in graphics on board. I have a older dual xeon machine myself but my nodes with newer i5's puts out the same fps or better. You guys keep up the good work.

[guest]

I'm pretty sure xeons don't have any graphics built in so no opencl on them that I know of.

I'm pretty sure from what I've read about it recently, that Xeon's do support OpenCL, and they don't use the graphics to achieve that.

I have an E5 2697 V2 12 core, that I have yet to experiment with.

I'm hoping Atak comes back with some info, as I know he has a Xeon system (might even be a duallie).

[Atak_Snajpera]

Even old Q6600 supports opencl if you have amd drivers. Running knlmeanscl on cpu is practically useless due to extreme slooooooooow speeeeeeeeddddd...

[Wishbringer]

let's better say:
Q6600 has no graphics core,
Xeon E5-2697v2 has no graphics core.
So they can't support OpenCL native with graphics core parallelism.
They can emulate graphic cores with AMD drivers (in software layer), but it's slow.
It's like play witcher3 with softwarerendering instead of using a decent card.

Even an old 80486 could emulate OpenCL, if someone is willing to write software layer,
but if it is usefull is another question.
Better stick to filters that don't use OpenCL on systems without a natively supporting hardware.

[apostolis21]

Hey, is distributed encoding superseeded or not used anymore? No one seems to know anything on this matter, not even Atak :rolleyes:

There is no ts or mpg support and I wonder why? Almost every recording software uses mpg for the old mpeg2 streams and ts for the new mpeg4 ones.

I have win7 at both computers, save java version and both have avisynth 2.6. Shared folders are both accessible from the two computers and I am using the classic 1.17.5 version of ripbot.

I would appreciate any insight here cause I really want to use this feature even if I have to mux my ts to an mp4 every time before the encoding process

Kind Regards

[guest]

Xeon E5-2697v2 has no graphics core.

They can emulate graphic cores with AMD drivers (in software layer), but it's slow.


So how do you do that ??

[Atak_Snajpera]

I am using the classic 1.17.5 version of ripbot.
This version was released in 2014-01-13 ! First upgrade to latest version and then we will talk ok :)

They can emulate graphic cores with AMD drivers (in software layer), but it's slow.
I do not think this is emulation of physical cores. I get pretty decent GFlops in FlopsCL benchmark on Xeon CPU.
http://i.cubeupload.com/a3YzjP.png

Just for testing purposes I've installed Intel OpenCL driver (http://registrationcenter.intel.com/irc_nas/5198/opencl_runtime_15.1_x64_setup.msi) to see if I can get better results in FlopsCL.
They are just terribad. First of all FlopsCL crashes if you select problem size larger than ~2600 blocks. Also Intel Platform is 2x slower than AMD! Nice joke AMD platform works better on Intel CPU :) I suspect that Intel Platform does not use AVX and relies only on SSE2.
http://i.cubeupload.com/e80ihI.png