MaximumPC 2003 12

 MAXIMUMPC DECEMBER 2003

THE TESTS AMD Athlon 64 FX-51 Intel P4EE Apple Dual G5

Mathematica 5.0 572 sec 639 sec 997 sec

SETI@Home 106 min 111 min 173 min

UT2003 332 fps 334 fps 80 fps

Jedi Outcast 130 fps 133 fps 72 fps

Quake III 445 fps 501 fps 404 fps

InDesign 2.0 Export to PDF 47 sec 46 sec 61 sec

Photoshop 7.0.1 All Filters 269 sec 266 sec 330 sec

Photoshop 7.0.1 MacAddict Filters 38 sec 41 sec 37 sec

Bibble/MacBibble 451 sec 354 sec 240 sec

QuickTime 6.3 Export to MOV 803 sec 744 sec 706 sec

Compressor “Fast” vs. Procoder “High-Speed” 225 sec 225 sec 263 sec

Compresssor “High-Quality” vs. Procoder 1.5 “Mastering” 1105 sec 1190 sec 724 sec

And Without Further Ado... The Results!

Best scores are bolded in red

the hard drive to CPU, but because the

MacAddict script uses a gigantic 50MB

file, there’s likely a little more pressure

put on the hard drive. http://www.adobe.com

BIBBLE/MACBIBBLE

When snaping a picture, some digital

cameras can dump the contents of their

image sensors to a noncompressed

image file (aka a “raw” file). Think of it

as digital negative. This helps preserve

the image’s actual visual properties, but

RAW files are a PITA to deal with. They

eventually need to be converted into a

more popular format, and conversion

takes time—up to 20 seconds per pic-

ture. For this test, we timed how long it

took to convert 85 files from Nikon’s raw

NEF format to TIFF using Bibble Lab’s

Bibble on the PC and MacBibble on the

Mac. These file conversions are almost

entirely CPU-dependent. Both Bibble

and MacBibble support multithreading,

so the P4EE’s Hyper-Threading and the

G5’s dual procs should come into play.

http://www.bibblelabs.com

QUICKTIME 6.3

On all three machines, we used the app

to convert a 2GB digital video file into

a QuickTime MOV file using default

settings. But let’s be clear: This test

was picked to humor MacAddict. The

QuickTime code, after all, was written

by Apple! Oh well. We don’t know any

PC users who would ever choose to

encode to QuickTime , but the app is

indeed cross-platform. Video encoding

speeds are typically CPU-dependent,

though hard drive write speeds can

also have an influence on performance.

http://www.apple.com

COMPRESSOR/PROCODER

The last test—DV-to-MPEG2 conver-

sion—is our most controversial, as our

testing procedure for the G5 was bor-

rowed directly from Apple (as per Apple’s

very strong request), and is by definition

a blatant apples-to-oranges comparison.

Nonetheless, it’s the same testing that

Apple quotes in its own benchmarking

reports, so we figured readers would be

interested in Maximum PC’s take.

Apple’s test involves two different

apps that do ostensibly the same type

of video conversion. On the G5, the test

uses the Mac-only app Compressor. This

app offers two encoding quality settings:

“fast” and “high-quality.” On the PC

side, Apple opts for Canopus’s ProCoder

1.0. This app provides some extra con-

trol parameters, but it too offers two

basic quality settings: “high-speed” and

“mastering.” Each app is tasked with

converting a 1.5GB DV file.

The first problem with this scheme

is obvious: Compressor and ProCoder

1.0 are entirely different apps! Apples to

oranges? Definitely. Still, there’s some

validity to the reasoning that says, “It

doesn’t matter how you convert the

video. All that matters is which com-

puter finished the job first.”

Fair enough, but this brings us to

the second problem: It’s very difficult

to discern whether the similarly named

settings in each app actually produce

the same visual quality results. For

what it’s worth, Canopus says that

Compressor’s “high-quality” setting

actually produces visual quality that

falls between ProCoder’s “high-speed”

and “mastering” settings. Canopus also

told us that using the “mastering” set-

ting puts complete emphasis on visual

quality while simultaneously ignoring

processing time, hence the warning that

appears when the “mastering” setting is

enabled (“Mastering is the highest-qual-

ity setting, but encoding can be 10-20

times slower than the other options”).

The upshot is that Apple may— may —be

asking PCs to do a harder workload.

Regardless, the issue is moot because

we decided to run the test essentially

as Apple defines it in its benchmark-

ing document (our only change was to

update ProCoder to a new 1.5 version).

In the benchmarking chart, you’ll also

find that we compared Compressor at

its “fast” setting with ProCoder at its

“high-speed” setting. As with any video-

encoding test, the Compressor/ProCoder

battles reveal CPU power, and how well

an OS copes with large file sizes.

(Continued on page 54)