Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
On Sun, 9 Jan 2005, Steven M. Schultz wrote: > On Sun, 9 Jan 2005, Trent Piepho wrote: > > > I thought that Dik meant the effective -q never went below 8, even though > > it go have gone all the way to 4 and still be within the target bitrate. > > If that's happening it's a Bug. It might start out at 8 but > with 'cbr' I've seen the encoder vary (and over a wide range) the > effective -q in order to keep close to the specified bitrate. Even if the CBR mode didn't have this bug, it's still fundamentally at a disadvantage vs VBR. The reality is that a given stream will have some parts which are more complex and need more bits to encode than other parts. In VBR mode the encoder can use fewer bits for the easy parts and more bits for the hard parts. In CBR mode the encoder is forced to waste bits on the easy parts when the extra bits result in little improvement, and then it is forced to drop quality on the hard parts when using some extra bits would result in a large improvement. --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
On Sun, 9 Jan 2005, Steven M. Schultz wrote: On Sun, 9 Jan 2005, Trent Piepho wrote: I thought that Dik meant the effective -q never went below 8, even though it go have gone all the way to 4 and still be within the target bitrate. Exactly. If that's happening it's a Bug. It might start out at 8 but with 'cbr' I've seen the encoder vary (and over a wide range) the effective -q in order to keep close to the specified bitrate. I use very short DV photo slideshows for testing things, because I can generate them whenever I want. These slideshows which are static except from the transitions. When I don't specify the -q option, mpeg2enc will stick to effective -q 8 for at least a few seconds, removing many details from the image. So, the explanation is that mpeg2enc needs a few seconds of video material to 'get going' and everything will get better later on in the video? I tried the anytovcd script by Nicolas Boos and it works beautifully. So I guess I'd better just use a two-pass VBR method. Cheers, Dik --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
On Sun, 9 Jan 2005, Trent Piepho wrote: > I thought that Dik meant the effective -q never went below 8, even though > it go have gone all the way to 4 and still be within the target bitrate. If that's happening it's a Bug. It might start out at 8 but with 'cbr' I've seen the encoder vary (and over a wide range) the effective -q in order to keep close to the specified bitrate. But I don't worry about it - most of the material I'm working with is not the greatest quality to begin with and I've never had 'golden eyeballs' ;) Cheers, Steven Schultz --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
On Sun, 9 Jan 2005, Steven M. Schultz wrote: > On Sun, 9 Jan 2005, Dik Takken wrote: > > bandwidth. Why does the CBR encoder not use -q 4 to keep the bitrate > > as high as the requested bitrate? > > Because -q enabled VBR mode? ;) In 'cbr' mode the '-q' setting > is ignored because the encoder is controlling it and I have seen > the effective -q go down to 3 or less. > > With CBR the encoder will vary the 'quality' - look at the "quant=" > lines that get logged and you'll see the effective "-q" as the > encoder is running. I thought that Dik meant the effective -q never went below 8, even though it go have gone all the way to 4 and still be within the target bitrate. --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
On Sun, 9 Jan 2005, Dik Takken wrote: bandwidth. Why does the CBR encoder not use -q 4 to keep the bitrate as high as the requested bitrate? Because -q enabled VBR mode? ;) In 'cbr' mode the '-q' setting is ignored because the encoder is controlling it and I have seen the effective -q go down to 3 or less. With CBR the encoder will vary the 'quality' - look at the "quant=" lines that get logged and you'll see the effective "-q" as the encoder is running. The point I was trying to make is that transcode is not doing anything you couldn't do with at most 1 test encoding of a representative section of the data. Steven Schultz --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
On Fri, 7 Jan 2005, Steven M. Schultz wrote: On Fri, 7 Jan 2005, Nicolas Boos wrote: Le Thu, 6 Jan 2005 20:27:26 +0100 (CET) Dik Takken <[EMAIL PROTECTED]> écrivait : Currently, I need to run mpeg2enc 4 or even 5 times in order to find the optimal quantiser number (the -q option) that yields the best quality and uses all available bitrate. That's why I'm currently Or simply use --cbr and let the encoder vary the -q automatically. --cbr, is that a new CVS option? I use the last release version, so I assume that omitting the -q option enables CBR mode. Specify the bitrate that will generate the desired size and the encoder will do the rest for you. Replace "-q N" with "--cbr" to use the specified bitrate, the encoder will automatically vary the effective -q for you without 4 or 5 passes. CBR mode is *very* cautious with using the available bitrate. When there is little motion, the CBR encoder uses only a tiny fraction of the available bitrate, keeping the -q as high as 8. The result is that lots of detail that could easily fit in the available bandwidth is discarded. When I switch to VBR, I see that I can easily encode the same video with -q 4, resulting in an amazingly detailed MPEG that does not exceed the specified bandwidth. Why does the CBR encoder not use -q 4 to keep the bitrate as high as the requested bitrate? Cheers, Dik
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
Le Fri, 7 Jan 2005 19:43:30 +0100 Nicolas Boos <[EMAIL PROTECTED]> écrivait : [...] > You should try the (pseudo) two passes mode of the anytovcd.sh script > from the current cvs. For example, like that for a dvd output : > >$ anytovcd.sh -i lavrec.avi -r -p dvd [-f light -t temporal] > > The script try to find the good quantization value so the output > size never get higher than required (4700 mbytes here). More precisely, it runs first pass at quantization value 1 and bitrate limited (cbr like) then calculate average quant value from a pseudo statfile (the mpeg2enc output). The second pass runs vbr with the computed quantization value and the max. bitrate allowed for the desired output type e.g. 7500 for DVDs. Cheers, Nicolas --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
On Sat, 8 Jan 2005, Dik Takken wrote: > > Maybe I did something wrong and mpeg2enc indeed works exactly as transcode > does. I will check this again soon. What 'mpeg2enc' command does transcode generate for you? transcode isn't doing the encoding itself as I recall, it runs mpeg2enc as a subprocess. I think transcode was/is using "-f 3" (generic MPEG-2) rather than "-f 8" (DVD defaults). The 'generic' form used 'cbr' where the DVD setting enabled 'vbr'. That was changed to be 'vbr' for both unless '--cbr' (or -u) is given. That's what I had in mind when I suggested using the cbr option. Is transcode doing any extra/different filtering than when you run mpeg2enc yourself? If so that would be another area where difference would come in. It should be possible to use the same options ('-f 3 -b ' and so on that transcode would generate, but that's just a guess. At that point the output should be the same (one would hope ;)). Cheers, Steven Schultz --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
On Fri, 7 Jan 2005, Steven M. Schultz wrote: Currently, I need to run mpeg2enc 4 or even 5 times in order to find the optimal quantiser number (the -q option) that yields the best quality and uses all available bitrate. That's why I'm currently Or simply use --cbr and let the encoder vary the -q automatically. Last time I checked this, the encoder didn't want to use the full bitrate that it was allowed to use. The used quantiser numbers were also quite high. Transcode produced bigger output files by default, which looked more crysp also. all nicely within the specified bitrate limits. I could only persuade mpeg2enc to use the same bitrate by manually specifying the -q value that yields good results, which can only be found by trial and error. Or educated guess, but my scripts can't do that, unfortunately... :) Maybe I did something wrong and mpeg2enc indeed works exactly as transcode does. I will check this again soon. Specify the bitrate that will generate the desired size and the encoder will do the rest for you. Replace "-q N" with "--cbr" to use the specified bitrate, the encoder will automatically vary the effective -q for you without 4 or 5 passes. No need to use transcode (unless you want to ;)) - I didn't know it had a magic quality option Experience should build after after a while though - shouldn't take more than 1 or 2 test encodings (representative portion of the entire video) to get the parameters "good enough". Or just set -q down to 3 and be done with it :) Cheers, Steven Schultz --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users - --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
On Fri, 7 Jan 2005, Steven M. Schultz wrote: > It does sound though like you're aiming at a specified size and if so > then you'll need to hold the bitrate constant and vary the quality. > Vbr works the other way - holds the quality constant but varies the > bitrate ;) Obviously, CBR mode doesn't hold the instantaneous bitrate constant, by making every frame exactly the same size. Rather, it holds the bitrate constant over some short interval. Probably a hundred frames as most. When trying to fit something on a DVD or CD, all you care about is that the bitrate is constant over the *entire stream*. You could easily have higher than average bitrate for several thousand frames, followed by a lower than average for several thousand more. --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
On Fri, 7 Jan 2005, Steven M. Schultz wrote: > > Dik Takken <[EMAIL PROTECTED]> écrivait : > > > > > quality and uses all available bitrate. That's why I'm currently > > > forced to use transcode for fully automated mpeg2 encoding. Transcode > > > automatically finds all optimal encoder settings and uses all > > > available bitrate. The thought came to me after I sent my previous suggestion... Transcode is NOT automatically finding optimal encoder settings at all. All transcode is doing is either 1) relying on the old (undocumented) behaviour of "cbr" for generic MPEG-2 or 2) specifying "--cbr". MPEG2 encoding is always vbr now unless cbr is specifically requested. You can achieve the exact same thing that transcode is doing by replacing "-q N" with "--cbr" (if you're using the cvs version of mjpegtools). It does sound though like you're aiming at a specified size and if so then you'll need to hold the bitrate constant and vary the quality. Vbr works the other way - holds the quality constant but varies the bitrate ;) Cheers, Steven Schultz --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
On Fri, 7 Jan 2005, Nicolas Boos wrote: > Le Thu, 6 Jan 2005 20:27:26 +0100 (CET) > Dik Takken <[EMAIL PROTECTED]> écrivait : > > > Currently, I need to run mpeg2enc 4 or even 5 times in order to find > > the optimal quantiser number (the -q option) that yields the best > > quality and uses all available bitrate. That's why I'm currently Or simply use --cbr and let the encoder vary the -q automatically. Specify the bitrate that will generate the desired size and the encoder will do the rest for you. Replace "-q N" with "--cbr" to use the specified bitrate, the encoder will automatically vary the effective -q for you without 4 or 5 passes. No need to use transcode (unless you want to ;)) - I didn't know it had a magic quality option Experience should build after after a while though - shouldn't take more than 1 or 2 test encodings (representative portion of the entire video) to get the parameters "good enough". Or just set -q down to 3 and be done with it :) Cheers, Steven Schultz --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
Le Thu, 6 Jan 2005 20:27:26 +0100 (CET) Dik Takken <[EMAIL PROTECTED]> écrivait : Hi Dik, [...] > Currently, I need to run mpeg2enc 4 or even 5 times in order to find > the optimal quantiser number (the -q option) that yields the best > quality and uses all available bitrate. That's why I'm currently > forced to use transcode for fully automated mpeg2 encoding. Transcode > automatically finds all optimal encoder settings and uses all > available bitrate. You should try the (pseudo) two passes mode of the anytovcd.sh script from the current cvs. For example, like that for a dvd output : $ anytovcd.sh -i lavrec.avi -r -p dvd [-f light -t temporal] The script try to find the good quantization value so the output size never get higher than required (4700 mbytes here). Cheers, Nicolas --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
On Tue, 4 Jan 2005, Steven Boswell II wrote: there's currently a coding frenzy around mpeg2enc and the denoisers. While this coding fenzy is going on, I would like to know if there is any interest in implementing a per-GOP multi-pass encoding mode. Currently, I need to run mpeg2enc 4 or even 5 times in order to find the optimal quantiser number (the -q option) that yields the best quality and uses all available bitrate. That's why I'm currently forced to use transcode for fully automated mpeg2 encoding. Transcode automatically finds all optimal encoder settings and uses all available bitrate. For manual encoding, I prefer mpeg2enc, but for automated encoding, it just isn't usable. Cheers, Dik --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
On Tue, Jan 04, 2005 at 10:13:28PM -0800, Steven Boswell II wrote: > I'm testing it on a dual-processor Athlon MP 2800+ > machine. Just FYI, HP has a collection of mid-sized test systems available via the Internet, running various operating systems and CPU types. I see this includes 4-way Itanium2 and Xeon (which shows up as 8-way from userspace) systems running Linux. Various compilers including gcc and Intel's icc are provided. Access is free -- even if you're not planning on buying one yourself -- since these are partly intended for developers whose end-users might have this sort of hardware. System list and more information here: http://www.testdrive.hp.com/current.shtml You wouldn't normally have 100% of the machine to yourself (they have a more restricted testing program for that situation). If you start running long test jobs that eat 100% CPU time, they might take issue with it. But for short tests to make sure things are at least working properly and quickly, on hardware you might not otherwise have available, you might take a look. Note that they only provide unencrypted telnet and ftp for connections. No support for ssh, rsync, X11 display-back, etc. -Dave Dodge --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
On Wed, 5 Jan 2005, Steven Boswell II wrote: > But how much better is it filling up both CPUs? > Do you have any sense of how much less overall > idle time you have now? The load average is slightly above 2 now instead of the 1.8 range. The estimaged time to complete the encoding is about 15 hours. The previous run took a bit over 17 so it looks like an overall speedup in the neighborhood of 15% (+/- a little ;) was obtained by using '-p 2' > Also, does the idle CPU percentage change if you > modify the run priorities (done with "renice" Haven't tried that - and OS/X probably does things slightly differently ;) Cheers, Steven Schultz --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
>>it's nice to see the main thread (the one >>denoising intensity) running near 100% CPU >>usage. > >That's how y4mdenoise has always behaved - >but now it's using part of the 2nd cpu as well :) But how much better is it filling up both CPUs? Do you have any sense of how much less overall idle time you have now? Also, does the idle CPU percentage change if you modify the run priorities (done with "renice" under most Unices)? Maybe it's paranoia, but I swear there's a lot less idle CPU time if y4mdenoise is running at -10 and the rest of the pipeline is running at 0, instead of my usual setup where everything runs at 19. And I was pretty sure process scheduling didn't work like that. Steven Boswell ulatekh at yahoo dot com __ Do you Yahoo!? Meet the all-new My Yahoo! - Try it today! http://my.yahoo.com --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
On Tue, 4 Jan 2005, Steven Boswell II wrote: > I don't know when we developers will officially > release a new version of mjpegtools -- there's :) > version in order to use y4mdenoise. But if you do, I > just finished writing the first dual-processor version! Your timing is _excellent_. The rendering session just completed early this morning and I was just about to start the encoding when I saw your announcement of the "-p" option. > reading/writing of frames into separate threads. "-p > 2" moves color-denoising into a separate thread. So 'top' on OS/X shows the number of threads being used by a process and with "y4mdenoise -p 2" there are 4 threads. Wheee! > than it was. And it's nice to see the main thread > (the one denoising intensity) running near 100% CPU usage. That's how y4mdenoise has always behaved - but now it's using part of the 2nd cpu as well :) It's working nicely so far - thanks! Cheers, Steven Schultz --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
Re: [Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
Hallo > You control it with the new "-p" option. "-p 0" > preserves the old behavior. "-p 1" moves the > reading/writing of frames into separate threads. "-p > 2" moves color-denoising into a separate thread. So > far, that's as parallel as y4mdenoise gets. I can > make it more parallel by making it denoise separate > slices of the same frame in different threads, but > that's going to be a LOT of work, so I figure I'd > write this first & get it out there. BTW: On a dual CPU I was able to fill up both CPU's by starting 2 encoding commands. > I'm testing it on a dual-processor Athlon MP 2800+ > machine. Even running the massive "near-perfection > with videotapes" pipeline I posted to this list a few > days ago, I can't quite fill up the processors -- > total idle time hovers around 20%. But it's better > than it was. And it's nice to see the main thread > (the one denoising intensity) running near 100% CPU > usage. When I run it on my Athlon MP (2.13GHz) I get that output after 900 frames: ps ax -m -L 11405 - pts/4- 0:13 lav2yuv john.eli - 11405 -S+ 0:13 - 11406 - pts/4- 0:01 buffer -b 2048 - 11406 -S+ 0:01 - 11407 - pts/4- 12:00 lt-y4mdenoise -p 2 - 11407 -Rl+ 12:00 - - 11424 -Sl+0:01 - - 11425 -Sl+0:01 - - 11426 -Sl+2:41 - 11408 - pts/4- 0:25 mpeg2enc -f 8 -4 1 -2 1 -q 5 -P -N 1.0 -M 2 -E -10 -D 10 -a 2 -o test_p1.m2v - 11408 -Sl+0:25 - - 11423 -Sl+0:03 - - 11427 -Sl+0:31 - - 11428 -Sl+0:29 - Is this something you expected ? auf hoffentlich bald, Berni the Chaos of Woodquarter Email: [EMAIL PROTECTED] www: http://www.lysator.liu.se/~gz/bernhard --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
[Mjpeg-users] y4mdenoise (in CVS) now does 2 processors!
I don't know when we developers will officially release a new version of mjpegtools -- there's currently a coding frenzy around mpeg2enc and the denoisers. So you'll still have to get the latest CVS version in order to use y4mdenoise. But if you do, I just finished writing the first dual-processor version! You control it with the new "-p" option. "-p 0" preserves the old behavior. "-p 1" moves the reading/writing of frames into separate threads. "-p 2" moves color-denoising into a separate thread. So far, that's as parallel as y4mdenoise gets. I can make it more parallel by making it denoise separate slices of the same frame in different threads, but that's going to be a LOT of work, so I figure I'd write this first & get it out there. I'm testing it on a dual-processor Athlon MP 2800+ machine. Even running the massive "near-perfection with videotapes" pipeline I posted to this list a few days ago, I can't quite fill up the processors -- total idle time hovers around 20%. But it's better than it was. And it's nice to see the main thread (the one denoising intensity) running near 100% CPU usage. Anyway, enjoy. Steven Boswell ulatekh at yahoo dot com __ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com --- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almosthttp://www.thinkgeek.com/sfshirt ___ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users