View Single Post
Old 15th April 2009, 22:40   #7234  |  Link
DrNein
Registered User
 
Join Date: Sep 2002
Posts: 145
Quote:
Originally Posted by kikker View Post
Hi, just curious as to why there are two decoder selections each for VC-1 and AVC. For instance, my ATI card will only accelerate AVC video if ONLY the AVC (DXVA) is selected, but it will not accelerate the video if both AVC (DXVA) and AVC (ffmpeg) are selected.

The same thing applies to VC-1. Acceleration occurs when only VC-1 (DXVA) is selected.

My question is: Shouldn't the end user have a choice of either DXVA or ffmpeg but not both?

Another interesting point to mention is the fact that my lowly ATI 3450 definitely accelerates video with both AVC and VC-1 in MPC-HC. I can easily play 1080p BluRay M2TS files with a weak C2D processor at around 10-15% CPU usage. According to the FAQ at least, MPC-HC should only be able to accelerate video beginning with cards that are slightly better than the HD3450.
When both are selected, hardware (DXVA) should be used when possible (i.e. standards compliant video), and fallback to software (ffmpeg) otherwise.

HD2000 and higher have essentially the same hardware decoder performance. The CPU almost does not matter.

http://en.wikipedia.org/wiki/Unified_Video_Decoder


Quote:
Originally Posted by tetsuo55 View Post
3. Select the best renderer by default, and fall back 1 by 1.
3b. Order would be: EVR-CP, EVR, VMR9, VMR7, Overlay, Old renderer. (maybe mpc should remember which renderer connected and not try higher ones again after that?)
With XP, VMR7 has very good quality and the best features (and can use Overlay). VMR9 is good for subtitles but quality suffers. EVR-CP also has poor quality in certain instances. EVR has good quality but lacks features. So, if "best" was implemented then "fallback order" would have to be very different.
DrNein is offline