As you would expect, with a 300mhz G3 applications flew. I had just finished a review of the MacPicasso MP540/3DO video card where I had used Infini-D 4.0 for some tests (with the stock PTP 180mhz 604E card). Running the same application at 300/300 speeds of the XLR8 card made it feel like the Nexus GA video card was on steriods. Moving shaded models in real time was much smoother, effortless in fact. I've always said no video card is a substitute for a fast CPU, and this was proof positive. (MacBench video scores were boosted by more than 200 points as well).

With a CPU this fast, all operations of your Mac seem faster, from opening windows in the Finder, to scrolling images and games. It's hard to go back to a slower CPU after experiencing this kind of speed (and I've got the bills to prove it).

I'm sure some of you will ask why I didn't run my normal full range of tests (as done with other G3 cards) but I'm very late with this review (over a month) and the two tests I've run were of sufficient duration to show some differences. They were also very CPU intensive. I know that will not make everyone happy, but with half my day spent answering mail I can't do better at the current time.

I used the same Infini-D 4.0 tutorial file rendering test as I've done in past reviews (chapter 7 completed). Rendering quality was set to Ray Trace, medium anti-aliasing, shadows on, patch detail low. The table below shows the results of the XLR8 card at various speeds and a some other cards scores for comparison.

Note: The other CPU cards listed were tested in a 8500 with Infini-D version 4.0, whereas the XLR8 card was tested in a PowerTower Pro with version 4.01. Infini-D ran off a Seagate Barracuda (7200 rpm) drive on the 8500 and a IBM 7200 rpm drive on the PowerTower Pro (the IBM drive tested 10% slower than the Barracuda in Macbench). Infini-D was allocated 40MB of RAM on both systems (which used the same OS version and extension set), but the version update makes the comparison of times suspect. The Newer Tech cards were tested with the control panel set to 60ns (fastest) memory timing and all tests in the 8500 were done with the motherboard cache removed.

I can only guess that the combination of different Infini-D version (4.01 vs 4.0), and/or different machine are responsible for the better performance seen by cards in the 8500. When I get time I'll update the 8500 Infini-D to 4.01 and rerun the tests with PowerForce G3 to see if that affects completion times. The Bryce test results made a lot more sense.

I also ran tests in Bryce 2.0, using the samples scenes in the KAI folder called "Magical Knight I" and "Alexandria II". Again I show the times other CPU cards took for those same scenes in the 8500. Bryce 2 was left at the default memory allocation of 16980k. My version of Bryce on both the 8500 and the PowerTower Pro are identical, and the comparison times seem more logical. I've also updated the results for tests with a 9600 400/100/50 (overclocked 350MHz 604e) and Apple G3 system running 333/222/66.

Since applications performance was excellent, I rated this category a 9.

The next page describes the software cache control supplied with the card. Or you may use the links below to jump to a specific page.