MakeAvis and CCE speed tests

[DDogg]

I have MakeAvis (without audio) working in 2.66.01.07, and 2.67.09 using force colorspace as YUY2 (Still fails in CCE 2.50) (see note below). With this script:

LoadPlugin("D:\twoweeks\mpeg2dec3.dll")
mpeg2source("D:\TwoWeeks\2weeks.D2V",idct=2) # 2 for AMD. Should be 5 for P4 I think
#converttoyuy2()
#monitorfilter()
#addaudio()

Note:
Surprisingly Force colorspace as YUY2 in MakeAvs was actully slightly faster than the direct avs script using ConvertToYUY2. I do understand they may well be some quality differences between the two conversion methods?, but I don't really see them at the moment.

Above avs with MakeAVIS using forced YUY2 vs AVS script below loaded directly into CCE:
2.66.01.07 - 1.60 vs 1.55 realtime
2.67.09 - 1.55 vs 1.50 realtime

CCE Direct AVS:
LoadPlugin("D:\twoweeks\mpeg2dec3.dll")
mpeg2source("D:\TwoWeeks\2weeks.D2V",idct=2)
converttoyuy2()
#monitorfilter()
addaudio()

CCE 2.50 was still fastest at 1.67 realtime using the direct avs script immediately above. CCE 2.50 would not accept the forced YUY2 signpost from MakeAvis and throws a decompression error. CCE 2.50 is a pain, but I still think this is worth investigating as 2.50, I am pretty sure, is still the most widely used CCE version.

For your replication -
CCE 2.50 6000 CBR no VAF no anti noise filter.
CCE other - Used RB's progressive template Registry setting available here in this thread:http://forum.doom9.org/showthread.php?s=&threadid=50287

Addaudio function saved as addaudio.avsi

#"Wmansir version"
function AddAudio(clip v1) {
v2 = Blankclip()
v1 = AudioDub(v1,v2)
return v1
}