Forum Archive Home -> ffmpegX general discussion -> Intel Quad - full tilt boogie
| Intel Quad - full tilt boogie | ||
| awsedr posted 2008 Jun 24 17:11 | ||
| I have a new Mac with an Intel Quad processor and noticed while converting a movie that Ffmpegx seems to use one processor at a time, rotating between them and back and forth and not using all 4 at the same time? MenuMeters is running and indicates such things as the number of processors, processor load, memory usage, hard drives etc. I was hoping for "full" tilt processor usage for maximum speed. This setup IS 4 times faster than my dual G5 2.0 though, and a lot cooler. Any ideas why? | ||
| rumplestiltskin posted 2008 Jun 24 17:29 | ||
| My 2.4GHz Core2Duo iMac is about twice as fast as my old Dual-CPU 2GHz G5 in functions like video conversions. These Intel CPUs really rock.
Some apps are not specifically programmed to take advantage of multiple CPUs/cores; it been a problem for Apple because they can't claim all of the computing high ground without apps that really know how to make use of multiple cores. Allegedly, Snow Leopard (10.6) may have a more simple set of APIs that would permit more developers to call the code for multiple cores (or multiple CPUs) without jumping through hoops. All the above being said...there are a few apps that have preferences permitting you to dedicate more cores to the app's functions. XLD (X Lossless Decoder - found here: http://tmkk.hp.infoseek.co.jp/xld/index_e.html ) is one such app that I use to convert audio from FLAC to Apple Lossless (and other transformations). You may assign multiple cores to the process. There are others and, once Apple makes it a no-brainer for the developers, many more will follow. | ||
| tomlee59 posted 2008 Jun 24 18:36 | ||
| Not to hijack this thread (too much), but I'm curious, Rumplestiltskin, how fast do your FLAC-to-ALAC conversions run on your setup? | ||
| jagabo posted 2008 Jun 24 20:08 | ||
| You may have to add "-threads auto" or "-threads 4" to the ffmpeg command line. | ||
| awsedr posted 2008 Jun 24 21:51 | ||
| I do notice in the ffmpegx progress "info" button that this is in the header:
"MEncoder dev-CVS-060307-04:23-4.0.1 (C) 2000-2006 MPlayer Team Mac OSX static build for ffmpegX FIXME:Hardenabling SSE and SSE2 without detection CPU: Intel Pentium III Katmai/Pentium III Xeon Tanner (Family: 6, Stepping: 7) CPUflags: Type: 6 MMX: 1 MMX2: 1 3DNow: 0 3DNow2: 0 SSE: 1 SSE2: 1 Compiled for x86 CPU with extensions: MMX MMX2 SSE SSE2 My processor is SSE3? Don't know if thats a big diff. | ||
| rumplestiltskin posted 2008 Jun 29 11:33 | ||
Sorry for the delay in replying. Using XLD and one "thread" (set in the XLD prefs), I clocked 59 seconds for 13 tracks totaling 45:45. When altering the prefs to two threads, the time dropped to 29 seconds. By the way, I was running a P2P app while this was going on. |
Login/Register to our forum to be able to post here.