The Source for Mac Performance News and Reviews

Review: XLR8's CarrierZIFTM CPU Card The "Upgradeable Upgrade" Published: 6/11/99

MacBench 5.0 Test Results

Intro | Benchmarks  | Appl. Tests | Software Controls  | Docs/Installation | Specs/Design | Summary

Benchmark Tests

Benchmark tests were run with MacBench 5.0 as it is the accepted Mac standard. I also included BYTEMARK tests although I am not a fan of that benchmark as it is not an indicator of other important system performance areas like disk, memory and graphics card. Throughout this review I used the convention of showing CPU, Cache and Bus speeds (in MHz) in the format of CPU Speed/Cache speed/Bus speed, so 440/220/55 would mean a CPU speed of 440 MHz, a backside cache speed of 220 MHz and a Bus speed of 55MHz. Due to the current G3 CPU limit of 8X Bus/CPU speed ratio the maximum CPU speed possible with the CarrierZIF's 60MHz max bus speed is 480MHz (60MHz bus times 8X ratio = 480MHz CPU speed). This summer a 10X bus ratio G3 CPU is said to be released which will raise this limit to 10 times the bus speed. And remember, your Mac may not be able to run 60MHz bus speeds reliably, most can't (exceptions are often the PowerCenter Pro and Mach 5 Apple systems). So remember max CPU speed with be 8 times your reliable bus speed, at least until the new 10X G3 CPUs are available. MacBench 5.0 was run on all 3 systems used for the review - a 9600/350, Genesis (based on a 9500 motherboard) and PowerCenter Pro. Remember that Macbench Graphics scores will vary depending on graphics mode and the installed video card. Since all these Macs have graphics cards of widely differing performance capability, this is not a reflection on the CPU card under review. All tests were done at 1024x768, thousands colors, 75Hz (the PowerCenter Pro has only 2MB VRAM onboard so the Publishing test could not be run on that system since it can't support the required millions colors, 1152x870 mode). The 1000 normalized score in Macbench 5 is from an Apple Beige G3/300 running millions colors, 1152x870, so consider this when evaluating the scores at lower resolutions and color depths. I did not include disk scores since again the systems varied widely in disk performance and in most cases a CPU upgrade has little benefit on Macbench disk scores (Disk Cache settings in the Memory Control Panel boost scores as well up to a point (about 6MB or so) - see the Disk Cache Study from last year for details). Be aware that each card and system has some tolerance variation, so these speeds may not be attainable with every card/system/hardware combination. The specific CPU chip on each card, your motherboard components/installed hardware and RAM mix can affect maximum reliable speeds. Some Genesis systems contain a modified 9500 motherboard that has added cache termination for lower noise. The 512k soldered-in cache of the my Genesis motherboard was not disabled during the tests and was not a limiting factor, and bus speeds as high as 55MHz were reliable. (Note: Initial tests in my PowerTower Pro 180 at 50MHz bus speeds required removing the L2 Cache dimm. I did not deinterleave RAM.) MacBench 5.0 Performance Notes: Some explanation of the MacBench graph. 9600/400: This is the 9600/350's stock Mach 5 CPU card overclocked to 400MHz. As noted in my Mach 5 page, these CPU cards had a 1MB inline L2 cache running at 100MHz and a 50MHz system bus speeds. These CPU cards are not usable in anything but the original Apple Mach5/Kansas motherboard systems, the 8600/250 and 300 and the 9600/300 and 350's. This is noted in the FAQ as well but is often overlooked until an owner of another Mac model can't boot with a 'bargain' Apple Mach5 CPU card.. Although this card has been reliable at 400MHz, I do not recommend others overclock their cards due to the delicate soldering required. Graphics Scores: The Ultimate Rez graphics card used in the Genesis scored better than the Thunder 3D in the 9600, partly due to the generally lower scores consistently seen in the Kansas motherboard based 9600s (300 and 350MHz systems) and the fact the IX3D is faster in 2D than the Glint chip. The PowerCenter Pro onboard video is very old and limited, with only 2MB (not expandable) VRAM on the riser in my model. All tests used 1024x768 graphics mode due to the limits of PowerCenter Pro onboard video. Why no Disk Scores? Since the Genesis uses a Seagate Wide Ultra SCSI 10,000 RPM Cheetah hard disk, Conley Softraid 2.1 driver connected to an ATTO PCI SCSI controller and the 9600 uses the IBM/Apple 4GB OEM disk this would hardly be a valid comparison. The 9600's disk is fairly slow as shown in my PCI SCSI vs Onboard SCSI article and was also 80% full. ATTO Tools' benchmark shows this disk is only capable of about 5MB transfer rates. The PowerCenter Pro's 2930 UltraSCSI card and (narrow SCSI) drive are a bit faster but the IBM OEM drive is nothing to write home about. CPU upgrades generally do not improve disk scores as in most cases the disk itself it the bottleneck even at stock CPU speed. This is shown in my older 604E CPU card reviews when I included disk scores. BYTEMARK Tests: I ran the BYTEMARK DR/3 test on the XLR8 CarrierZIF while installed in the Genesis and 9600/350. I include it here for curiosity only, as I consider BYTEMARK the least indicative of all benchmarks as far as actual OS/end user performance since it is a pure benchmark and is highly affected by compiler optimizations (some tests on the PC with VisualC++ compiled versions show the PII scoring higher than the G3 for instance). Scores from a previous XLR8 G3 266/177 CPU card review at speeds of 266/177, 300/300 and 322/214 are shown for comparison. All scores listed are from G3 cards with 1MB of backside cache. Scores are rounded off to two decimal places. Speculative Processing was enabled for all tests (since it was not an option with previous control panels). BYTEMARK DR/3 SCORES Card/Speed: Integer Score: FPU Score: XLR8 CarrierZIF 440/220/55 (Genesis/9500) 14.27 9.62 XLR8 CarrierZIF 412/206/55 (9600/350) 13.70 9.04 XLR8 G3/400 CPU Card 400/200/50 (Genesis) 12.97 8.75 Blue &White G3/400 400/200/100 12.90 8.71 G3 322/214 8.92 6.70 G3 300/300 8.37 6.65 G3 266/177 7.36 5.75

For comparisons to other CPU Card upgrades and systems see my site list of CPU Card Reviews.

Since the CarrierZIF/XLR8 400Z combo posted the best scores of any CPU slot card I've tested at the time of this review (6/11/99), I rated Benchmark performance a 10. The next page has comments and some results from real world applications. Or you may use the links below to jump to a specific page.

Index of XLR8 CarrierZIF Review Pages Intro | Benchmarks  | Appl. Tests | Software Controls  | Docs/Installation | Specs/Design | Summary - or - Back to WWW.XLR8YOURMAC.COM

Copyright © Mike, 1999. No part of this sites content or images are to be reproduced or distributed in any form without written permission. All brand or product names mentioned here are properties of their respective companies. Users of the web site must read and are bound by the terms and conditions of use.