Analysis of Standardized MS Pro 6.5 results so far...

[Hulk]

I have calculated how many CPU cycles each of the CPU's below requires to run Doc's test at 1fps.

Example P4 1700
Time: 7:34 converted to 454 seconds
3903 frames in the test file
3903/454=8.6 frames per second
1700MHz/8.6fps=198MHz to render 1frame per second

The CPU's can now be compared on an equal basis, isolated from their total MHz rating.

Dual Athlon 1.2 158
PIII Tualatin 1.26 o/c 1.4 178
Dual PIII 933 183
Athlon XP 1800+ 185
Athlon C 1.4 186
P4 1.7 198
PIII 850 199

The dual athlon system is most efficient, followed by a single Tualatin PIII. That prefetch must be doing something. Not much difference between the Athlon XP and C, I wonder why the SSE instructions aren't having an impact?

Mark

[wkulecz]

I don't find the preview rendering time differences very significant for deciding P4 vs. Athelon since nothing I preview is even one tenth as complex as Doc's test. Rendering color correction filters would be a better decider for me, but unless someone has both P4 and Athelon it'll be impossible to get a fair comparison without exchanging files that are too large to be practical.

MPEG2 VBR5500 encoding of Doc's test is a worthwhile proxy for me. As I'm basically cheap, my next editing computer upgrade will likely be an Athelon unless I see compelling reason with software available when I buy to pay extra for the P4, or I get totally PO'd at Gates & Co. and buy a Mac (being basically cheap puts Apple at a big disadvantage, I use Mac's at work and don't see compelling performace for the price on things I do). Now I'm trying to decide if DDR is worth the extra cost over SDRAM, if I wait long enough the way prices are dropping it may be moot :-)

So what does your figure of merit say for mpeg2 encoding?

--wally.