Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
10th August 2009, 14:54 | #1 | Link |
Registered User
Join Date: Aug 2005
Posts: 21
|
XVID multi-core support?
Hello Forum!
I have a dual core AMD X2 CPU. I will buy a new CPU: a triple-core AMD Phenom CPU. Or, maybe, I will buy a quad-core X4 CPU. My question: the XVID codec can support the 3, or 4 CPU core, to improve the video transcoding speed? I convert MPEG2 VOB files to Xvid avi with AutoGK, or GordianKnot, or Virtualdub mpeg2 edition. Thanks |
10th August 2009, 21:53 | #2 | Link | |
Actually in reserve
Join Date: Oct 2004
Posts: 1,605
|
Quote:
I wish multicore owners (more than2) could answer. Last exchange on the subject began kind of flame war Multicore will scale accordingly to the number of cores for x264 (not as well for other H264 encoders it seems). Latest Xvid revs support multicore but results may vary, depends on the encoder: vfw or cli; depends on avisynth scripts if any, optimized for MT or not, etc ... All contributors agreed for some improvement; not on the degree of improvement. Let's hope multicore owners will share their experience (or you may try searching for related threads). Did
__________________
Having a problem with AutoGK? Read & use the FAQ & MORE FAQ first Want to exchange on AutoGK? try doom10.org In reserve (inactive) for an undefined period of time. |
|
11th August 2009, 00:06 | #3 | Link |
Registered User
Join Date: Mar 2005
Posts: 249
|
I don't really know if Xvid has multi core support but the last time I converted a DVD to Xvid using MeGUI at the highest settings (except Qpel and GMC) I got like speeds of like 150 FPS on a Phenom X4 955 3.2 Ghz quad core (progressive source, threads=6). If I use AviSynth processing like Yadif de-interlacing the speed drops considerably, but I still got like 50-60 FPS.
Really I wouldn't get a quad core per se if your aim is to encode Xvid, it will always be fast even on dual or triple core (unless you start to use complex de-interlacers and AviSynth filters I guess). Personally I don't really use Xvid anymore, it's old, not really developed anymore and very much inferior to x264. (Not trying to start a flame war just my opinion ) |
11th August 2009, 00:24 | #4 | Link |
Registered User
Join Date: May 2006
Posts: 335
|
I have a Q6600, upgraded from an E4400, and I can say that x264 encodes a little faster than XviD here, for SD content, and same speed for HD content (on dual core, XviD was a bit faster, but on quad, it's put to shame by x264). So I guess the multithreading on XviD is not half as efficient as the one on x264, sadly. It doesn't scale at all.
Will XviD get faster on a quad core, if you're upgrading from a dual core? Yes, it will. Absurdly faster, like, 100% improvement? Nope. But you can always multithread the AviSynth, if you like lots of filtering. |
11th August 2009, 08:15 | #5 | Link | |
Registered User
Join Date: Dec 2005
Posts: 1,460
|
Quote:
Since you have a quad core you can encode a short clip (~5000 frames, progressive source, no filtering) with multithreading and again with only one thread. Then compare the fps and you'll know how xvid scales. Alternatively one of the guys with a dual core can do the same and then extrapolate what the best case scenario on a quadcore would be. |
|
17th April 2012, 08:03 | #6 | Link |
Registered User
Join Date: Oct 2008
Posts: 42
|
Sadly, Xvid does not have multi-core support. On my 6-core Thuban cpu, x264 used the full load. But xvid encoding did not. I still want to encode 4x3 videos in Xvid, b/c standalone DVD players do not support H.264 playback, only Divx/Xvid .
|
17th April 2012, 14:56 | #8 | Link |
Registered User
Join Date: Mar 2011
Posts: 4,821
|
Edit: I originally ran these two encodes with the dual core running at 3.17GHz, but having a slight case of OCD I ran it again at the same clock speed as the quad core (3.2GHZ) and so I've edited the post. For the record the extra little bit of clock speed shaved 7 seconds off the dual core's total encoding time.
So as a rough guide...... AutoGK encode, 19 minute vob file, 640x480 output, no audio. XviD 1.2.1 (UTC 2008-12-04) Q9450 @ 3.2Ghz [17/04/2012 11:33:00 PM] Analyzing source. [17/04/2012 11:33:16 PM] Source has percentage of interlacing in motion areas: 13.68 [17/04/2012 11:33:16 PM] Source is considered to be interlaced. [17/04/2012 11:33:16 PM] Output will contain 28525 frames [17/04/2012 11:33:16 PM] Using VAQ in XviD [17/04/2012 11:33:16 PM] Overhead: 0 bytes (0.00 Mb) [17/04/2012 11:33:16 PM] Video size: 262,144,000 bytes (250.00 Mb) [17/04/2012 11:33:16 PM] Running compressibility test. [17/04/2012 11:33:43 PM] Duration was: 27 seconds [17/04/2012 11:33:43 PM] Speed was: 83.89 fps. [17/04/2012 11:33:43 PM] Compressibility percentage is: 70.26 [17/04/2012 11:33:43 PM] Chosen resolution is: 640x480 ( AR: 1.33 ) [17/04/2012 11:33:43 PM] Predicted comptest value is: 70.26% [17/04/2012 11:33:43 PM] Running first pass. [17/04/2012 11:37:23 PM] Duration was: 3 minutes 40 seconds [17/04/2012 11:37:23 PM] Speed was: 129.56 fps. [17/04/2012 11:37:23 PM] Expected quality of first pass size: 73.00% [17/04/2012 11:37:23 PM] Running second pass. [17/04/2012 11:42:10 PM] Duration was: 4 minutes 46 seconds [17/04/2012 11:42:10 PM] Speed was: 99.63 fps. [17/04/2012 11:42:10 PM] Job finished. Total time: 9 minutes 13 seconds E6750 @ 3.2Ghz [18/04/2012 12:16:40 AM] Analyzing source. [18/04/2012 12:16:56 AM] Source has percentage of interlacing in motion areas: 13.68 [18/04/2012 12:16:56 AM] Source is considered to be interlaced. [18/04/2012 12:16:56 AM] Output will contain 28525 frames [18/04/2012 12:16:56 AM] Using VAQ in XviD [18/04/2012 12:16:56 AM] Overhead: 0 bytes (0.00 Mb) [18/04/2012 12:16:56 AM] Video size: 262,144,000 bytes (250.00 Mb) [18/04/2012 12:16:56 AM] Running compressibility test. [18/04/2012 12:17:37 AM] Duration was: 40 seconds [18/04/2012 12:17:37 AM] Speed was: 56.78 fps. [18/04/2012 12:17:37 AM] Compressibility percentage is: 70.26 [18/04/2012 12:17:37 AM] Chosen resolution is: 640x480 ( AR: 1.33 ) [18/04/2012 12:17:37 AM] Predicted comptest value is: 70.26% [18/04/2012 12:17:37 AM] Running first pass. [18/04/2012 12:21:46 AM] Duration was: 4 minutes 9 seconds [18/04/2012 12:21:46 AM] Speed was: 114.49 fps. [18/04/2012 12:21:46 AM] Expected quality of first pass size: 73.00% [18/04/2012 12:21:46 AM] Running second pass. [18/04/2012 12:29:02 AM] Duration was: 7 minutes 16 seconds [18/04/2012 12:29:02 AM] Speed was: 65.37 fps. [18/04/2012 12:29:02 AM] Job finished. Total time: 12 minutes 26 seconds Total times also include the demuxing and indexing. Not a huge difference between the two, but I guess it could add up to about 20 minutes for a 2 hour movie (or less I guess, if you're comparing faster CPUs). At least for the version I'm using, it seems to be the second pass where Xvid takes more advantage of quad cores (CPU usage went up from around 50% to 60% to 75% to 80%). For some reason CPU usage for the E6750 was around 95% for the first pass, but seemed to drop about 5% for the second pass. Of course if you use a quad core, you can negate some of Xvid's ability to push 4 cores to 100% by running two encodes simultaneously. I do, if I've got a few encodes to run. PS While I haven't timed it, I'd be willing to bet the quad core encodes using x264 at twice the speed of the dual core, given it pretty much always pushes both CPUs to 100%. In fact I never bother encoding with x264 on the dual core. It's way too slow. Last edited by hello_hello; 17th April 2012 at 16:06. |
Thread Tools | Search this Thread |
Display Modes | |
|
|