Beige G3 266 on X 10.2.3--How fast Is It?

We run the latest version of Mac OS X client and server from separate partitions on an original Beige G3 with 266 MHz G3 CPU. It has 256 MB RAM and an ATI Radeon graphics card in place of the original video but is otherwise stock with a 6 GB hard drive. We use it to test new software before it is deployed to the clients on our network and as a server. It runs Mac OS X very well but we wondered how it stacked up speedwise against our fastest desktops, four upgraded G4 Cubes.

One G4 Cube has been upgraded with an 800 MHz PowerLogix G4 CPU, 1 GB RAM, a GeForce3 graphics card and and 80 GB 7200 RPM Seagate Barracuda IV hard drive. The second and third have been upgraded with a faster 1 GHz PowerLogix G4 CPU, 1 GB RAM and a lesser GeForce2 MX graphics card. The latter also still have the original slower 20 GB 5400 RPM Maxtor hard drive.

We used Xbench 1.0, a donationware benchmarking utility available for download. This benchmarking utility has become stable and fairly consistent in its final, non-beta 1.0 release. It it easy for anyone to obtain in order to easily compare the results from their own computer. Here are the results as averages of 3 runs:

Test Scores for Xbench 1.0

Test	Beige G3 266 MHz	Cube 800 MHz	Cube 1 GHz (a)	Cube 1 GHz (b)
.
CPU	29	90	112	116
Threading	20	50	62	63
Memory	22	84	80	90
Quartz Graphics	30	95	104	98
OpenGL Graphics	23	91	115	120
User Interface	29	84	100	108
Disk	18	88	54	67

While the 266 MHz Beige G3 runs applications on Mac OS X 10.2.3 at a slow but acceptable speed, it tests out at about 30 per cent or less of the speed of the Cubes.

It is also clear from these tests that graphics, both Quartz and OpenGL, are still highly dependent on the speed of the CPU despite Quartz Extreme. The 800 MHz Cube has a MUCH faster GeForce3 graphics card but scores up to 30 per cent less that the Cubes with a 1 GHz CPU and a much slower GeForece2 MX graphics card.

The two identically equipped 1 GHz Cubes score differently, mostly less than 5 per cent. However, the RAM and hard drive tests differ by 12 and 24 per cent.

The bottom line from Xbench? There is no substitute for CPU MHz when it comes to running graphics on OS X. The faster the CPU the better, regardless of the graphics card. A faster hard drive also is helpful.