CCE 2.66 2x slower than 2.64 !!

[jpsdr]

Hello.

I've noticed that the 2.66 is almost two times slower than
the 2.64 !
I don't know if there is a problem in the configuration, or
if it's normal...

I've both win98SE and 2000 on the same HD.

The tests have been made with the same file, an
uncompressed
AVI file of 720x576 25Fps of 3 minutes (split in 2Go files).
The source file was on an HD on an IDE controler, the
result file was put on another HD, on the other IDE controler.

Config :
PIII 1GHz, 512Mo RAM
Chipset Intel i815ep, latest drivers installed and
Intel Application Accelerator 2.3 installed on both OS.
FAT32 system.

With CCE 2.64.01.10 on Win98SE : Speed around 0.8
With CCE 2.66.01.07 on Win2000 SP3 : Speed around 0.4

Is this normal ?
Does someone have made the same kind of tests ?
Does someone know if Win2000 slow the process, or the
HD transfert ?

[Ntsita]

I have tried time ago a comparison between the 2,64,01,10 and the 2,66,1,7 with the same OS (WInXP) and on the same machine and winner it is exited the last version (i remember thhere is a 0.1 difference in realtime). :o

[jpsdr]

Sorry, i'm not excellent in english, and i don't
understand what you mean by "it is exited the last version".


Wich what kind of file have you made the test ?
If i've used uncompressed files, it's for a good reason :
not having a codec wich may slow down the process.
What i mean, if 90% of the total time is used by the
codec of avi file, even if the MPEG encoder is ten time
faster than the other version, you'll not see more than
10% difference....

[Ntsita]

_______________________________________________________

...even if the MPEG encoder is ten time
faster than the other version, you'll not see more than
10% difference...
_______________________________________________________

Wait a minute.:confused:
If cce (during encoding) says 1.0x faster than realtime, it means that for encode 60' film lenght, it needs effectively 60 minuts;
but if says 10.0x faster than realtime, it means that need only 6 minuts for encoding the same 60' film lenght, indipendently by single time of reading, decoding, encoding and saving of the file.
Do you agree?:p
My test was made with this simple avisynth 2.06 script:
loadplugin("mpeg2dec3.dll")
mpeg2source("test.d2v",cpu=0)

and another with interlaced compressed avi (codec morgan v3).
My pleasure, when i made these tests, was the little more speed in this new version of cce at same quality parity.:D
(the cce's developers have two squared balls)
Then the winner is CCE 2.66
Sorry for my english (it's a babelfish translation from italian)

[jpsdr]

Reading, decoding, are include in the time processing.
In my opinion, it simple calculate how many time (t), it
make to encode N frames, and make a speed like N/t.
But, to encode N frames, include reading/decoding if
necessary !!
I may put it other way, the encoder says : I've finished to
encode the frame number N, i've beginnig since t seconds,
my speed is N/t.

Your test may have been biaised, by a lot of factors :
You seems to use a frame server, wich is not something
fast, i've seems to understand, and your file is MPEG-2
encoded, if i've understand.

Personnaly, i think your test is not valid, to much other
things in the process...

This still not explain my speed difference of x2 !!
Is-it windows 2000 wich slow down, or CCE 2.66...?