Offical badger donald graft thread.

[meleth]

In order to get him to improve the speed of decomb when doing 2pass encodings we will have to badger him. So bump this thread if you do 2pass encodings and would like to see this :P

[Guest]

OK, but be prepared with technical arguments if needed. First thing: prove that the saved analysis time will not be wiped out by the log file reading and writing time.

[meleth]

Well ofcourse you can't be 100% sure until somethign has been actually tried.

I only assume that doing this would save time, you know much better than me how much cputime your analysis function/s uses.
I do remmember that sometime ago there was another filter(IVTC4 i think it was called) that used it's own cache for 2pass encoding due to the speed increase.

Ofcourse as was pointed out in another thread, Quality over Speed is way more important. So this is something that can wait, imo anyways.

[Guest]

You mean this one:

http://forum.doom9.org/showthread.php?s=&threadid=18342

I'll study it as time permits.

[meleth]

hehe yup, that's the one i was thinking off. Man didn't even remmember that conversation :P

[kayman]

i encode while i sleep so speed isnt an issue as longs as its done when i wake up , i couldnt care less if it takes 2 or 8 hours

great job neuron2, keep the good stuff coming!


kayman

[int 21h]

If I get some time tonight, I'll fire up VTune and let you know exactly how much time Decomb spends in detection.