Okay, time for some quick tests:
CPU: Core2Duo E8400 (3.0 GHz)
GPU: Radeon 4870
Test 1:
starting file: AVCHD format (1920x1080 .MTS) 2min 15s length
converted file: PSP High Quality (320x240 AVC .MP4)
encoding time with GPU encoding: 00:01:43
encoding time without GPU encoding: 00:01:55
conclusion: Only 10% decrease in encoding time. Maybe not a huge difference here because it's already h.264 to start with?
Test 2:
starting file: AVCHD format (1920x1080 .MTS) 2min 15s length
converted file: BD MPEG-2 (1920x1080 .MPG)
encoding time with GPU encoding: 00:01:47
encoding time without GPU encoding: 00:03:11
conclusion: Nice! Nearly cut the encoding time in half!
Test 3:
starting file: BD MPEG-2 (1920x1080 .MPG, the output file from Test2) 2min 15s length
converted file: PSP High Quality (320x240 AVC .MP4)
encoding time with GPU encoding: 00:01:11
encoding time without GPU encoding: 00:01:18
conclusion: Only 9% decrease in encoding time. Looks like my hypothesis from Test1 is incorrect.
Overall: According to these limited tests, for my CPU and GPU, the improvement in Mpeg2 encoding is dramatic, but the improvement in h.264 encoding is very disappointing. Also, as a slightly off-topic issue, I wish PD7 had more options for h.264 .MP4 container encoding (larger or customizable video sizes).
EDIT: Hmm, just noticed a weird consequence, too. The file size of the clip from Test2 with GPU encoding enabled is 276 Mb, while the file size of the clip with GPU encoding disabled is 410 Mb. Both clips play just fine, and appear to be of similar quality.
This message was edited 1 time. Last update was at Feb 05. 2009 01:13