The speed of a system is the sum of the speed of it's parts. If the hard drive
is to the specifications of the manufacture, and in a Mac you certainly don't
roll your own so to speak, then the analysis is valid. I don't think I'd be
very happy to buy a machine crippled by the manufacturer using substandard
components. He may have attributed the difference in speed to the processors
but that hardly changes the bottom line.
At 12:36 AM 1/25/2003 -0600, you wrote:
I thought some people might be interested in this item from the latest
Imaging Resource Newsletter. It turns out that Rob Galbraith's comparison of
Mac and PC processor speed was fundamentally flawed--it mainly measured
hard-drive speed, and the Mac he used had a slow hard drive installed.
An HTML version of this newsletter (without Deals) is on the Web at
http://www.imaging-resource.com/IRNEWS/
--Mike
>>>>>>>>>>>>By Chris Russ
(On Jan. 7, Rob Galbraith published his test of four computer systems
(http://www.robgalbraith.com/diginews/2003-01/2003_01_07_macpc.html) for
"RAW photo processing speed," concluding "the fastest dual processor Mac has
been soundly thumped by one of the fastest single processor PCs." For a
programmer's perspective on the issue, we asked Chris Russ, the wizard
behind Optipix's cross-platform performance, what he thought. -- Editor)
In Rob Galbraith's article "In Pro Digital Photography, Megahertz Matters" I
think Galbraith missed a few important variables that greatly affect how his
"comparison" turned out.
First of all, virtually all of the operations he was testing were disk
bound. What this means is that they were highly dependent upon the speed of
the _hard drives_ and the _interfaces_ to them. So the computer's CPU was
actually spending most of its time waiting for the disk to do its thing.
If you're going to use a bunch of tricked-out machines against a stock
Macintosh, at least put in an ATA-133 (or better) controller and a faster
hard-drive. In order to reduce costs, Apple has recently been using slower
drives and controllers. An ATA-133 card costs a paltry $100-$200 and can
dramatically speed up disk bound processes (like Photoshop!). Low seek-time
hard drives also make a difference.
Secondly, the other factor is that he was testing third party software, much
of which was not optimized for the G4 either because of difficulty in
programming, time to market or because optimization really wouldn't matter
much (especially for disk-bound algorithms).
A fairer test would have used images that fit in memory (which would have at
least eliminated the speed of the hard drive and the interface as
variables). At the same time, it would have been a good idea to reduce the
History to a small number of previous states.
So, you could probably use a RAM disk for file conversion, but that really
doesn't make sense, does it?
Let's look at the tests themselves:
Issue: Processing RAW Photos
Fujifilm RAW File Converter LE 1.0 -- This is disk-bound and third-party
software. There was no pressure on the manufacturer to make it faster.
Canon File Viewer Utility 1.1 -- Disk-bound and third-party. Look at how
much slower the 2.5-inch hard drive is in the TiBook.
Kodak Photo Desk 2.0 -- Disk-bound and third-party. Even PC laptops are
slower.
Bibble Labs MacBibble 3.0 beta/Bibble 3.04 -- Even Eric Hyman (the author)
says it wasn't G4-optimized.
Nikon Capture 3.5 -- Disk-bound and third-party.
Nikon NEF plug-in for Photoshop -- Disk-bound and third-party.
Issue: Single Image and Batch Processing in Photoshop 7.01
Multi-step resample -- Disk-bound!
Unsharp Mask -- _Actually isolates processor speed_ if the image is small
enough to fit in memory and the Mac was faster. Imagine that!
Batch process using Web site Action -- Disk-bound
Batch process using event Action -- Disk-bound
Process photos into a Web Photo Gallery -- Disk-bound
Issue: Multitasking with Photoshop 7.01 et al.
Every single one of these tests is disk-bound. They all involve converting
one file type to another (or resampling) and exacerbate the performance of
the disk.
Issue: Cataloging Photos in Extensis Portfolio 6
Same thing.
Issue: Transferring Photos from Card to Computer
Can you say "Writing _to the disk_?" Disk-bound.
CONCLUSION
Only one of these many tests actually tested the CPU speed of the Macintosh.
_Only one_. And in that test, the Mac was faster and the TiBook put in a
really respectable showing.
What Galbraith's article does demonstrate is that the speed of your hard
drive makes all the difference in the world.
<<<<<<<<<<<
Outside of a dog, a book is man's best friend.
Inside of a dog, it's too dark to read. --Groucho Marx
