on 2/3/03 9:55 AM, Alex Ray wrote:

> 350MHz is 5.7 times slower than 2GHz....

IF we are talking the same chip. The G4 isn't on parity speed wise with the
Pentium. So it's always apples and oranges. And a G3 is really prehistoric.
 
> Sounds like the mac did quite well. Remember, a G3 has no AltiVec Core.
> (Dunno if fuji s/ware would even use this tho!)

True. And the software may have no AltiVec calls! This is a big issue when
you look at these speed tests. Galbraight's review isn't any better (or
worse) in light of how really silly these reviews can be.

First, I'd use a software product I know the vendor has done their best to
make platform agnostic as far as raw speed (that means Photoshop). If you
look at the huge difference in say the Canon RAW software, it's pretty clear
that the people that wrote it don't really have a clue about optimizing
code. 

Then there are the issues with conducting the tests. About 6 or 7 years ago
I was asked by PEI magazine to test the fastest machines running Photoshop.
At the time, there was even an SGI version running and I had an SGI box,
which at the time was supposed to be SOOOOO fast. Now it wasn't really and
if I told you how long it took to set up the scratch disk typing UNIX, even
it was faster, I'd have been well behind in my work.... There is something
to be said for working with an OS that is user friendly.

What we found was that the suite of tests played a huge role. Some routines
were clearly optimized for a certain OS while others were not. NOT every PS
function is optimized for MMX or AltiVec (and Multi-processors). Then you
have the huge difference in running a routine once, then a 2nd time and then
a 3rd time. Boy, the times changed a lot.

Bottom line in my mind is that you can produce any sets of speed tests to
produce a "Winner" if you like OR if you don't have all the specific testing
parameters down, one box may "win" simply due to improper testing
procedures. Those procedures don't always operate the way real world users
operate (how many people run an Unsharp Mask on the same image 3 times and
average the speeds from each operation?).

Then we have the leap frogging of hardware every 3- 6 months. When Rob did
his review, the same week, Fortune (or maybe it was Money) magazine did a
review that put the Mac on top. Who's right?

The Mac verses PC wars are getting so tiring and yet, we get these reviews
like Galbraith's that do nothing constructive but aim people to web sites. I
remember running Photoshop 1.0.7 on a Mac Iici with 16mb of ram and having
to rotate an 15mb file 1.5 degrees. Took 25 minutes back in 1990. We've come
a long way thankfully. Yes time is money. How many users on a Mac are going
to toss all their hardware because someone was able to process files faster
on a PC (or vise versa).

What these reviews do provide is evidence that a lot of code written isn't
very well optimized for all users on all platforms.

Andrew Rodney 

===============================================================
GO TO http://www.prodig.org for ~ GUIDELINES ~ un/SUBSCRIBING ~ ITEMS for SALE

Reply via email to