Finally got around to testing out the differences in compression performance between my Matrox CompressHD PCI-e card and QMaster using 4 CPU cores.
Test was done using Compressor 3.51 on a MOV sequence captured at 1080i 60 (HDV) (Field upper) -- nothing was done to the sequence that is 1 min 2 secs long 206.4MB file size (not even de-interlaced).
Matrox CompressHD card software is 1.8.1 (firmware updated).
My goal was to get the 1920 x 1080 (actually 1440 x 1080) footage close to Apple TV 1280 x 720 progressive format (m4v).
The Results:
1. H.264 for Apple Devices 1280 x 720 (16:9 960 x 720) 5Mbps, progressive 44.1kHz Stereo (Medium quality 50) single pass without Qmaster:
CPU Usage - User 68-71% - System 5-7%
Time = 1 min 49 sec -- file size 39.9 MB
2. H.264 for Apple Devices 1280 x 720 (16:9 960 x 720) 5Mbps, progressive 44.1kHz Stereo (Medium quality 50) single pass with Qmaster 4 cores:
CPU Usage - User 87% - System 7%
Time = 2 min 01 sec -- file size 39.9 MB
3. H.264 (mov) Custom 1280 x 720 5Mpbs, progressive 29.97 fps - 44.1kHz Stereo High Quality Multi-pass using 4 cores Qmaster:
CPU Usage - User 88% - System 6%
Time = 4 min 02 sec -- file size 38.7 MB
4. Matrox MAX H.264 (mov) Custom 1280 x 720 5Mpbs, progressive 29.97 fps - 44.1kHz Stereo:
CPU Usage - User 28-30% - System 8%
Time = 1 min 06 sec -- file size 39.7 MB
What I find interesting is that the Apple TV quality was actually pretty good, the color was a little less vibrant that results from #3 and #4. What is really strange is that QMaster with 4 cores was actually slower than a without QMaster?? This leads me to believe the .m4v codec is able to detect and use more than 1 core regardless of QMaster -- in fact, adding QMaster just takes a little away from the efficiency of the .m4v codec.
It was nice to see the Matrox CompressHD only using 28-30%, allowing my system to be very responsive to doing other tasks.
Just a small sample of H.264 results, but with the type of work I do I get anywhere from about 2X to 4X faster compression times using the Matrox CompressHD.
This is probably old news for most of you, but it's new life for my aging 2007 MacPro
Rob.
Test was done using Compressor 3.51 on a MOV sequence captured at 1080i 60 (HDV) (Field upper) -- nothing was done to the sequence that is 1 min 2 secs long 206.4MB file size (not even de-interlaced).
Matrox CompressHD card software is 1.8.1 (firmware updated).
My goal was to get the 1920 x 1080 (actually 1440 x 1080) footage close to Apple TV 1280 x 720 progressive format (m4v).
The Results:
1. H.264 for Apple Devices 1280 x 720 (16:9 960 x 720) 5Mbps, progressive 44.1kHz Stereo (Medium quality 50) single pass without Qmaster:
CPU Usage - User 68-71% - System 5-7%
Time = 1 min 49 sec -- file size 39.9 MB
2. H.264 for Apple Devices 1280 x 720 (16:9 960 x 720) 5Mbps, progressive 44.1kHz Stereo (Medium quality 50) single pass with Qmaster 4 cores:
CPU Usage - User 87% - System 7%
Time = 2 min 01 sec -- file size 39.9 MB
3. H.264 (mov) Custom 1280 x 720 5Mpbs, progressive 29.97 fps - 44.1kHz Stereo High Quality Multi-pass using 4 cores Qmaster:
CPU Usage - User 88% - System 6%
Time = 4 min 02 sec -- file size 38.7 MB
4. Matrox MAX H.264 (mov) Custom 1280 x 720 5Mpbs, progressive 29.97 fps - 44.1kHz Stereo:
CPU Usage - User 28-30% - System 8%
Time = 1 min 06 sec -- file size 39.7 MB
What I find interesting is that the Apple TV quality was actually pretty good, the color was a little less vibrant that results from #3 and #4. What is really strange is that QMaster with 4 cores was actually slower than a without QMaster?? This leads me to believe the .m4v codec is able to detect and use more than 1 core regardless of QMaster -- in fact, adding QMaster just takes a little away from the efficiency of the .m4v codec.
It was nice to see the Matrox CompressHD only using 28-30%, allowing my system to be very responsive to doing other tasks.
Just a small sample of H.264 results, but with the type of work I do I get anywhere from about 2X to 4X faster compression times using the Matrox CompressHD.
This is probably old news for most of you, but it's new life for my aging 2007 MacPro
Rob.