Benchmarking

by Eric W. Schwartz
from the AmiTech Gazette, September 2013

Now that I have a new high-point machine in the Powermac G5 running MorphOS, I thought I would run a handful of tests comparing speed and the time it takes to do the same Amiga software task between systems. This is by no means a clinical scientific investigation, as there are several variables that were not weeded out. It's just a guy with a stopwatch app and some curiosity.

The Players:

First test was the one that means the least unless you run Windows -- boot time (defined here as the time from when the power switch is hit to when the desktop is fully up.)

1st: Powerbook 26.75 seconds
2nd: Powermac G5 38.4 seconds
3rd: Amiga 4000T 41.5 seconds
4th: Pegasos 43 seconds

Apparently the G5 has more hardware checks on boot. The Pegasos' time was inflated, as I have a wait delay on boot that I neglected to skip, which might have reduced the time up to ten more seconds, but the Amiga can't lose them all in this competition.

Test #2: Photogenics, maximum gaussian blur operation on 2560x1920x24 image

1st: Powermac G5 22.9 sec.
2nd: Powerbook 29.7 sec.
3rd: Pegasos 46.2 sec.
4th: Amiga 20 min. 7 sec.

Here the CPU power becomes most apparent, though the G5 is not quite the runaway leader as I might have expected.

Test #3: ImageFX, Oil Paint effect on 1280x 1280 image, brush size 5

1st: Powerbook 21.5 sec.
2nd: Powermac G5 22.6 sec.
3rd: Pegasos 37.9 sec.
4th: Amiga 3 min, 34.5 sec.

Stunning upset as the G4 Powerbook beats the G5. I ran it twice to make sure I didn't mess up on the stopwatch. I may have also make a settings mistake on the Amiga, but that only helped its time.

Test #4: Lightwave 3D 5, 2010 Xmas card render, 1280x960, full ray tracing

1st: Powerbook failed, ran for over an hour, power plug knocked loose, battery expended.
2nd: Amiga cancelled because I didn't want to wait around
3rd: Pegasos cancelled because I didn't want to wait for it.
4th: Powermac G5 failed, did not finish before MorphOS demo time limit was reached.

Well, that one was kind of a cluster-fudge.

Test #4b: Lightwave 3D 5, 2010 Xmas card render, 640x480, ray traced shadows only

1st: Powermac G5 25 min, 40 sec.
2nd: Powerbook 27 min, 21 sec.
3rd: Pegasos 44 min, 59 sec.
4th: Amiga 8 hours, 35 min, 52 sec.

Even with the reduced specs, the G5 barely squeaked through within its demo time. This was a much more demanding render than I first imagined.

Test #5: Lightwave 3D 5, Metrobot render, 1280x960, full ray tracing

1st: Powermac G5 3 min, 50 sec.
2nd: Powerbook 4 min, 10 sec.
3rd: Pegasos 6 min, 48 sec.
4th: Amiga 1 hour, 12 min, 9 sec.

This was a much easier render overall (less complex surfaces and transparency). Still, it shows that in most things where CPU speed is a primary factor, the G5 is usually the winner, with the G4-based Powerbook not far behind. Every bit of extra speed is useful for intensive tasks though, and the Macs outpace the Pegasos II, and not surprisingly, the Amiga, though not by factors that mere clock speeds might have you believe.