iDCT Algorithm.

[Lord_GalAthon]

What's the difference between the settings? I found I SOMETIMES get a better encoding speed with 64-bit IEEE.

Encoding Gone in 60 seconds using a T-Bird 1200Mhz, 384MbRAm on a MSI K7T PRO 2A motherboard. Using 4 pass VBR. Output is two files of approximately 750Mb.

32-bit SSE MMX - .604 (avg.)
64-BIT IEEE - .860 (avg.)

I haven't tried the third option.

Same setup but encoding Shrek, output is two files of just over 700Mb.
Since the movie is shorter I suppose the datarate is higher even though the files are slightly shorter.

32-bit SSE MMX - .944 (avg.)
64-BIT IEEE - .790 (avg.)

Weird ah?

[markrb]

I am guessing that one reason for the better encode speeds is that the T-Bird does not have the SSE instruction set. AMD first added these to the XP line of cpu's.
It's a guess though.
These are the options in DVD2AVI and I always thought it was only the way that program handled the files.
However your results seem to show that there is an effect in CCE too.

Mark

[mrbass]

http://www.edensrising.com/dvd2svcd_advref.html
32-bit SSE MMX
64-bit Floating Point
IEEE-1180 Reference
iDCT Algorithm
iDCT algorithms possess the quality of MPEG-2 decoding.

First of all, these three iDCTs are all IEEE-1180 compliance which confirms the quality of iDCT algorithms. It's really hard to tell the difference perceptually.

Generally speaking, speed of 32-bit MMX/SSEMMX > 64-bit Floating Point > IEEE-1180 Reference iDCT. In view of this, 32-bit MMX/SSEMMX is recommended.

Mathematically, quality of IEEE-1180 Reference > 64-bit Floating Point > 32-bit MMX/SSEMMX iDCT Reference iDCT implies lossless transformation for those who pursue for the limit.

SSE MMX is a bit faster than MMX iDCT because it invokes PSHUFW instruction.
If your CPU supports (P3, P4, Athlon, Duron, some Celeron), DVD2AVI will choose the faster one automatically


If someone would please take the time to record the time each one took and just a general which one resulted in better quality that'd be cool. Here's a email I got from ChristianHJW http://www.powerdivx.com
hope he doesn't mind me quoting him here
Mr Bass <my email> schrieb am 04.01.02:
> Any other reason to use 64 BIT rather than 32-bit?

No, in fact i have to admit i didnt do a real test on this, but on my machine ( SMP dual PIII 800 ) the speed increase of using 32 bit is marginal ( +5% ) so i feel better using 64 bits ;-) ..... stupid it is on the other hand, because for my 1 CD rips i will use temp smoother afterwards to remove picture noise .... so why have a highest quality MPEG2 decoding when removing picture information afterwards .....

[markrb]

mrbass could you kindly translate a little into these term if possible.

1. What is suppose to be faster
2. What is suppose to give better quality.

I think, but am not sure:
1. is 32-bit MMX
2. is IEEE-1180

could you confirm I am reading your post right?

Thanks,
Mark

[mrbass]

a > b means A is greater than B
a < b means A is less than B
I learned that in 3rd grade...hehe

32-bit MMX/SSEMMX (fastest speed)
64-bit Floating Point (2nd fastest speed)
IEEE-1180 Reference iDCT. (3rd fastest speed)

IEEE-1180 Reference iDCT. (highest quality)
64-bit Floating Point (2nd in quality)
32-bit MMX/SSEMMX (3rd in quality)
Any other math questions?

[Lord_GalAthon]

I see. Then my results are inexplicable.

:p

[Solo]

Any idea if using a 64-bit CPU with the 64-bit Floating Point iDCT will make a difference ? I just tried it on my Athlon64 and it encoded a lot quicker than before ... maybe I accidently changed something else somewhere, but I doubt it. Any 64 bit CPU users out there ?