Single Player Quake 2 Scores
All tests were run in three consecutive iterations and averaged. Raw scores are in the following table, rated in fps (frames per second).
|
demo1 |
demo2 |
| 640 x 480 |
| MacOS X Server |
40.4 |
38.9 |
| MacOS 8.6 |
34.6 |
33.8 |
| Virtual PC |
8.6 |
7.9 |
| 800 x 600 |
| MacOS X Server |
40.1 |
39.0 |
| MacOS 8.6 |
34.5 |
33.6 |
| Virtual PC |
8.4 |
7.9 |
 (click for full image)
In the graph above, its easy to see what difference native vs. emulated code makes. Despite the fact that all graphics are being rendered by Voodoo 2 hardware, Virtual PC's emulated Pentium core just cannot keep up with the serious floating point demands of geometry calculation in Quake 2 to feed the data fast enough to the 3D hardware. With sub 10fps scores across the board, I would rate Quake 2 as barely playable, single player, out of the box in Virtual PC with Voodoo 2.
MacOS 8.6 vs. MacOS X Server however is a different story. Both delivered respectable over 30fps scores. It is interesting to note that MacOS X Server was 16% faster on average than MacOS 8.6. This could be attributed to any number of things, but I'd estimate at least one factor is MiniGL. MiniGL in MacOS X Server probably has less overhead, or slowdown, than even optimized Mesa drivers. At the very least, these results are a good indicator of how well MacOS X games will run. MacOS X (client) is based on MacOS X Server and will have an entirely PowerPC native microkernel core, and all drivers will be written with a new toolkit that was recently released to developers (Omni's Glide drivers tested here are in the old Mach/OPENSTEP style). In other words, don't be surprised if even better frame rates are achieved with MacOS X (client) early next year for native 'Cocoa' games.
 (click for full image)
Looking at the results for 800x600, its interesting to note that there isn't a significant difference in these scores and those benchmarked at 640x480. This is an indicator that the Voodoo 2 card is not being fed data fast enough to expose its lower performance at higher resolutions. For those of you considering a Voodoo 3 2000 card (which runs around $100), unless you have a G3 faster than 300Mhz you might consider saving some money and buying a Voodoo 2 (which can be found for $50 or less), if what you are mostly interested in is 3D game performance and compatability on your current machine. And if you currently own a Voodoo 2 on a machine with less than a 300Mhz G3, don't expect a Voodoo 3 to bring you higher frame rates (but it will bring you a larger desktop if you hook up a second monitor to the Voodoo 3, which is not a passthrough card and can be used in addition to built-in video circuitry or other video cards).
previous page | next page
|