Link to The #1 Reviews and Perf. News Site!
Click Here to see my complete list of CPU Card reviews.

A note about "real world" testing:
There is always the issue of repeatability in application test runs between products. Even using the same machine and cpu card, run to run variations as much as a few % of the score are common. Compound that variable with how repeatable you are at starting and stopping the stopwatch, along with the thermal issues of extended runs on the hard disk performance (drives do thermal calibration, esp. an issue in non-AV drives) as well.

After Effects, Bryce2 and Infini-D 4.0 do their own recording of render times, so no human error is present there. However in most of the tests human error is a consideration, especially when test results are less than one second apart. I go to great lengths to ensure consistent timing, but I'm only human.

All these factors can add variation to the run times. I personally do as much as possible to ensure consistent test results such as:

  • Rebooting between each test/application
  • Performing the identical series of steps for each test.
  • Running the tests in the same order for each product.
  • Use standard image files for timings (samples usually) when possible.
  • Machine under test is not used for any recording of data
    (only the application under test is running)

Keep these things in mind when comparing the application test run times - what is most important is the % difference between products. Even when scores are close, a consistently lower time for one product is a clear indication it is the better performer. See my CPU Upgrades and Systems pages for links to reviews and more information on the G3 upgrades and systems listed here.

Application Testing

Update: To see how an earlier rev G4 CPU upgrade compares to the same speed G3 CPU in an older Mac, see my First Look at the XLR8 G4 Upgrade article.

Application tests are somewhat of a mixed bag on G3 upgrades, considering the claims made regarding the PowerPC 750 chip in comparison to the 604E. As you'll see from the tests below, the 750 does have an advantage in literally all the tests - but in some cases such as heavily floating point (FPU) performance dependent applications, it is a relatively minor one. Operations such as searches (Find File) were dramatically faster than the 604E cards, and database applications should really see a performance boost with the PPC750 based cards, as does web browsing, a pleasant surprise. I've just bought FileMaker Pro 4.0, and will be reporting on this in a future update. [Note - there are many more real world tests with G3 systems and upgrades in my CPU Reviews.]

FPU speeds of the PPC750 seem to be only 10-15% faster in single precision than a similar speed 604E. The chip itself is said to be half as fast at double precision FPU operations, due to one less execution unit in the chip for double-precision floating point, but the backside cache's speed more than makes up for this in most applications it seems. However if you're upgrading from a 120, 150 or even 200mhz 604E, especially with a smaller L2 cache, you'll notice substancial improvements in performance. Keep in mind the PowerLogix Pro 233 scores were taken running at 263mhz, with a 1 meg L2 cache installed - not your Average Mac 604E to be sure. However I'm not a magazine, or online subsidary that depends on advertising so I wanted to show how the 750's compare to the best and fastest of the lower cost 604E cards, not just a stock 120 or 200mhz system. I'll let you make up your own mind - I just provide the facts.

These cards perform very well with no L2 cache, saving you an upgrade there - and they feel very responsive in all areas of use. If you can afford the delta in price - I say go for the 750 chip every time.

To provide some concrete application test results, I used After Effects 3.1, Bryce2, Photoshop 4, Infini-D 4.0, Freehand 7.0 and Virtual PC to perform typical real world tasks (except for VPC) to compare the speed of the MaxPowr Pro card with the best of the 604E cards, the PowerLogix Pro 233 card (set to 52.67mhz bus/263.35mhz cpu speed) and the more expensive PowerLogix Powerforce 250 cards (running at 45/292/292), as well as the stock 8500/120 (w/1meg L2 cache). I didn't have the patience to run the 8500/120 yet on the Infini-D test, as it would take several hours I estimate.

The After Effects rendering test involved a short movie with a explosion/fire filter effect. The Infini-D 4.0 test was to render a 150 frame animation, with ray tracing, shadows and medium anti-aliasing - using the Chapter 7 Tutorial file. This PowerForce G3 250 card tests were run with and without the PowerLogix 1meg RapidCache installed on the motherboard, as noted (it's the same cache used with both PL Pro 233 & 8500/120). The Powerforce results are all with a full speed backside cache setting (1:1), as are the Newer 275 card, and the Pro+ card when indicated by 266/266 (cpu/cache speed). The results are shown below:

After Effects 3.1 Rendering test:

  • PF G3/466 ZIF @533/267/66: 2:41 [in beige G3]
  • PF G3/466 ZIF @466/311/66: 3:03 [in beige G3]
  • B&W G3/400 @400/200/100: 3:33
  • Apple G3 w/PF 400 ZIF @433/288/66 (chipped): 3:18
  • Apple G3 w/PF 400 ZIF @400/266/66: 3:33
  • XLR8 400/200/1MB: 3:34
  • 9600/400 (chipped) Mach 5: 3:38 (no LibMoto)
  • Apple G3 333/222/66 (chipped): 4:14 (no LibMoto)
  • PowerForce G3 292mhz/no L3 = 4:49 (8500/45mhz bus speed)
  • PwrCenter w/PF G3 65/292mhz no L3 = 4:51 (65mhz bus speed)
  • PowerForce 285mhz w/1meg L3 = 4:58
  • MaxPowr 275 275/275/45 w/no L3 = 5:05 (60ns memory setting)
  • MaxPowr Pro+ 266/266/45 w/no L3 = 5:14 (60ns memory setting)
  • MaxPowr Pro 250/166/45 w/no L3 = 5:37 (60ns memory setting)
  • PowerLogix Pro 233 w/1meg L2 (604E@263mhz) = 5:34
  • MaxPowr Pro 250/166/45 w/no L3 = 5:38 (70ns memory setting)
  • 8600/250 (Mach 5): 5:54
  • Vimage G3/233/117: 6:08
  • Stock 8500/120 w/1meg L2 = 12:18

Infini-D 4.0 Rendering test:
Note: v4.01 seems much slower than 4.0 was, so results between the two versions can't be compared

Bryce 2 Rendering Tests

Test scenes are from the Sample Scenes folder, Kai subfolder (I used these std scenes to allow other Bryce2 owners to compare their results with mine). Default 16980k memory allocated to Bryce2.

Time to render the "Magical Night I" scene

  • PF G3/466 ZIF @533/267/66: 2:06 [in beige G3]
  • PF G3/466 ZIF @466/311/66: 2:24 [in beige G3]
  • B&W G3/400 @400/200/100: 2:49
  • Apple G3 w/PF 400 ZIF @433/288/66 (chipped): 2:32
  • Apple G3 w/PF 400 ZIF @400/266/66: 2:44
  • XLR8 400/200/1MB: 2:47
  • Apple G3 333/222/66 (chipped): 3:23 (no LibMoto)
  • 9600/400 (chipped) Mach 5: 3:30 (no LibMoto)
  • XLR8 266/177 @322/214: 3:29
  • MacTell 266 @300/300: 3:34
  • Railgun 250 @312/208: 3:37
  • XLR8 266/177 @300/300: 3:40
  • 8600/350 (chipped) Mach 5: 3:41 (no LibMoto)
  • PwrCenter w/PF G3 65/292mhz no L3 = 3:45 (65mhz bus speed)
  • PowerForce G3 292mhz/no L3 = 3:46 (8500/45mhz bus speed)
  • Railgun 250 @292/194: 3:46
  • PForce 220/110 @291/116: 3:52
  • PowerForce G3 285mhz w/1meg L3 = 3:52
  • MaxPowr 275 45/275/275 w/no L3 = 3:58 (60ns memory setting)
  • MacTell 266 @266/266: 4:04
  • MaxPowr Pro+ 45/266/266 w/no L3 = 4:06 (60ns memory setting)
  • XLR8 266/177 @266/177: 4:12
  • Railgun 250 @250/166: 4:24
  • MaxPowr Pro 45/250/166 w/no L3 = 4:27 (70ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 4:28 (60ns memory setting)
  • PForce 220/110 @240/120: 4:37
  • PowerLogix Pro 233 (604E@263mhz) = 4:46
  • Vimage G3/233/117 = 4:51
  • Stock PTP180: 6:37
  • Stock 8500/120 w/1meg L2 = 11:36

Time to render the "Alexandria II" scene

  • PF G3/466 ZIF @533/267/66: 2:51 [in beige G3]
  • PF G3/466 ZIF @466/311/66: 3:14 [in beige G3]
  • B&W G3/400 @400/200/100: 3:51
  • Apple G3 w/PF 400 ZIF @433/288/66 (chipped): 3:28
  • Apple G3 w/PF 400 ZIF @400/266/66: 3:43
  • XLR8 400/200/1MB: 3:45
  • Apple G3 333/222/66 (chipped): 4:35 (no LibMoto)
  • 9600/400 (chipped) Mach 5: 4:39 (no LibMoto)
  • Railgun 250 @312/208: 4:41
  • XLR8 266/177 @322/214: 4:46
  • MacTell 266 @300/300: 4:52
  • 8600/350 (chipped) Mach 5: 5:00 (no LibMoto)
  • XLR8 266/177 @300/300: 5:03
  • Railgun 250 @292/194: 5:08
  • PowerForce G3 292mhz/no L3 = 5:10 (8500/45mhz bus speed)
  • PwrCenter w/PF G3 65/292mhz no L3 = 5:12 (65mhz bus speed)
  • PForce 220/110 @291/116: 5:14
  • PowerForce G3 285mhz w/1meg L3 = 5:19
  • MaxPowr 275 45/275/275 w/no L3 = 5:29 (60ns memory setting)
  • MacTell 266 @266/266: 5:32
  • MaxPowr Pro+ 45/266/266 w/no L3 = 5:39 (60ns memory setting)
  • XLR8 266/177 @266/177: 5:47
  • Railgun 250 @250/166: 5:59
  • MaxPowr Pro 45/250/166 w/no L3 = 6:08 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 6:08 (70ns memory setting)
  • PForce 220/110 @240/120: 6:16
  • PowerLogix Pro 233 (604E@263mhz) = 6:33
  • Vimage G3/233/117 = 6:34
  • Stock PTP180: 9:04
  • Stock 8500/120 w/1meg L2 = 15:32
Photoshop 4.0 Tests

** NOTE: MaxPowr Pro tests run after the PhotoShop Swap Disk was defragmented, therefore PowerForce tests will be repeated soon - Photoshop pages to the swap disk on most tests with larger images so this could have affected the test times **
(For Photoshop 5 tests of faster Macs (and a dual PII 450 NT/PIII Win98) see my Photoshop Performance page. The latest CPU Card reviews have additional PSBench results as well.)

Time to launch Photoshop 4.0
(50 megs ram allocated to Photoshop)

  • MaxPowr 275 45/275/275 w/no L3 = 7.90 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 8.31 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 8.38 (70ns memory setting)
  • MaxPowr Pro+ 45/266/266 w/no L3 = 7.90 (60ns memory setting)
  • PwrCenter w/PF G3 65/292mhz no L3= 7.52 (65mhz bus speed)
  • PowerForce G3 292mhz/no L3 = 6.37 (8500/45mhz bus speed)
  • PowerForce G3 285mhz w/1meg L3 = 7.35
  • PowerLogix Pro 233 w/1meg L2 (604E@263mhz) = 7.85
  • Stock 8500/120 w/1meg L2 = 13.04

Time to load/display 16.1MB (file size) image

  • MaxPowr 275 45/275/275 w/no L3 = 9.62 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 10.00 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 10.41 (70ns memory setting)
  • MaxPowr Pro+ 45/266/266 w/no L3 = 9.90 (60ns memory setting)
  • PwrCenter w/PF G3 65/292mhz no L3 = 12.36 (65mhz bus speed)
    (Using same drives as 8500 below), but Pwrcenter SCSI may be slower)
  • PowerForce G3 292mhz/no L3 = 9.51 (8500/45mhz bus speed)
  • PowerForce G3 285mhz w/1meg L3 = 11.18
  • PowerLogix Pro 233 (604E@263mhz) = 11.50
  • Stock 8500/120 w/1meg L2 = 14.82

Rotate 32MB image 45 degrees
(50 megs allocated to Photoshop)

  • MaxPowr 275 45/275/275 w/no L3 = 45.87 (60ns memory setting) **
  • MaxPowr Pro 45/250/166 w/no L3 = 48.22 (60ns memory setting)**
  • MaxPowr Pro 45/250/166 w/no L3 = 49.38 (70ns memory setting)**
  • MaxPowr Pro+ 45/266/266 w/no L3 = 46.88 (60ns memory setting)**
  • PwrCenter w/PF G3 65/292mhz no L3 = 53.71 (65mhz bus speed)
    (PowerCenter video slower, and has only 64MB ram)
  • PowerForce G3 292mhz/no L3 = 46.47 (8500/45mhz bus speed) **
  • PowerForce G3 285mhz w/1meg L3 = 49.57
  • PowerLogix Pro 233 (604E@263mhz) = 57.31
  • Stock 8500/120 w/1meg L2 = 1:30
** tests used optimized swap file disk.

Rotate 32MB image 45 degrees (PS allocated 80MB ram)
To test if more Ram added to PS would increase the performance of the 60ns Pro Timing setting. PS still hits the disk swap file during this test, but the 60ns memory timing did show consistently better performance with the larger memory allocation.

  • MaxPowr 275 45/275/275 w/no L3 = 39.87 (60ns memory setting) (Ran 3 times)
  • MaxPowr Pro 45/250/166 w/no L3 = 37.65 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 41.32 (70ns memory setting)
  • MaxPowr Pro+ 45/266/266 w/no L3 = 37.75 (60ns memory setting)
  • PwrCenter w/PF G3 65/292mhz no L3 = Not Tested (65mhz bus speed)
  • PowerForce G3 292mhz/no L3 = 36.49 (8500/45mhz bus speed)
  • PowerForce G3 285mhz w/1meg L3 = TBA
  • PowerLogix Pro 233 (604E@263mhz) = TBA
  • Stock 8500/120 w/1meg L2 = TBA

Resample (up 4X) 2MB image

  • MaxPowr 275 45/275/275 w/no L3 = 5.28 (!) (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 8.03 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 7.89 (70ns memory setting)
  • MaxPowr Pro+ 45/266/266 w/no L3 = 7.20 (60ns memory setting)
  • PwrCenter w/PF G3 65/292mhz no L3 = 7.57 (65mhz bus speed)
  • PowerForce G3 292mhz/no L3 = 6.57 (8500/45mhz bus speed)
  • PowerForce G3 285mhz w/1meg L3 = 7.16
  • PowerLogix Pro 233 (604E@263mhz) = 10.34
  • Stock 8500/120 w/1meg L2 = 15.06

Gaussian Blur 8MB image

  • MaxPowr 275 45/275/275 w/no L3 = 1.85 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 3.75 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 4.13 (70ns memory setting)
  • MaxPowr Pro+ 45/266/266 w/no L3 = 3.41 (60ns memory setting)
  • PwrCenter w/PF G3 65/292mhz no L3 = 1.66 (65mhz bus speed)
  • PowerForce G3 292mhz w/no L3 = 1.71 (8500/45mhz bus speed)
  • PowerForce G3 285mhz w/1meg L3 = 3.71
  • PowerLogix Pro 233 (604E@263mhz) = 4.04
  • Stock 8500/120 w/1meg L2 = 6.13

Premiere 4.2 Make Movie Tests:

Make movie test - Sample Proj. Quicktime 320x240

  • Apple G3 333/222/66 (chipped): 0:37.50 (CPU and Cache speed boosted)
  • 9600/400 (chipped) Mach 5: 1:01
  • MaxPowr 275 45/275/275 w/no L3 = 45.02 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 51.60 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 52.84 (70ns memory setting)
  • MaxPowr Pro+ 45/266/266 w/no L3 = 45.47 (60ns memory setting)
  • PwrCenter w/PF G3 65/292mhz no L3 = 44.35 (65mhz bus speed)
  • PowerForce G3 292mhz/noL3 = 44.72 (8500/45mhz bus speed)
  • PowerForce G3 285mhz = 44.75
  • PowerLogix Pro 233 (604E@263mhz) = 57.90
  • Stock 8500/120 w/1meg L2 = 1:45

Virtual PC 1.0 "Wintune 97" Test Results: Virtual PC shows slightly better performance in some areas (Video primarily) with the faster 60ns memory control panel setting.

MaxPowr 275/275 (45/275/275/60ns) VPC results:
Using 1.31 control panel - set to 1:1 cache, 60ns memory timing.

  • 60ns Timings - 1.0 Cache ratio
  • Load - 26.81 sec
  • Reported Clock Rate: 355 Mhz
  • Dhystone: 160 Mips
  • Whetstone: 54 MFLOPS
  • Video Speed: 7.4 MP/sec
  • Cached Disk: 6.6 MB/sec
  • UnCached Disk: 1.6 MB/sec
  • RAM Read Avg: 260 MB/sec
  • RAM Write Avg: 106 MB/sec
  • RAM Copy Avg: 210 MB/sec

To compare this performance to the MaxPowr Pro card click here, for PowerForce 250 VPC results click here

Note: Wintune warns that the CPU L1 cache is turned off with any PPC750 based, backside cache card, but not with the 604E cards. (Further investigation leads me to believe this is due to the delta between the Read and Write times, which WinTune attributes to a disabled L1 cache. WinTune could be correct, or just pinpointing a VPC emulation issue). VPC Full screen mode was used in all MaxPowr Pro tests, Win95 video set to 640x480, 256 colors as usual.

VPC Results Summary: In general the Pro outperforms the 263mhz 604E by a good margin on the Video, Disk and Ram copy performance and surpassing the real P90 on the memory tests. It also does a respectable job compared to the PowerForce card on most tests. Odd that the Dhrystone scores are slightly lower for the PPC750 than the 604E, which is the opposite of what I expected. Slight differences (+-3% or so) in some tests can be attributed to run-to-run variations, which I've seen before. I'm eager to report these results to Connectix.

Note that I have not applied the update patch from Connectix due to the problems I've read regarding the patch. Note that overall Ram performance was far superior to the P90 systems - from 3X to 10X faster with the with the PPC750 cards.

Freehand 7.0 Tests:

AfterBurner compress "Associated Press.." document

  • MaxPowr 275 45/275/275 w/no L3 = 3.19 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 4.12 (60ns memory setting)**
  • MaxPowr Pro 45/250/166 w/no L3 = 3.82 (70ns memory setting)**
  • MaxPowr Pro+ 45/266/266 w/no L3 = 3.22 (60ns memory setting)
  • PwrCenter w/PF G3 65/292mhz no L3 = 4.16 sec (65mhz bus speed)
  • PowerForce G3 292mhz/no L3 = 3.05 sec (8500/45mhz bus speed)
  • PowerForce 250 @285mhz w/1meg L3 = 3.07 sec
  • PL Pro 233 604E @263mhz w/1meg L2 = 4.88 sec
  • Stock 8500/120 w/1meg L2 = 7.74 sec
** Test run 3 times (machine rebooted each time), but
each time the 70ns speed delivered the better performance

AfterBurner Compress "Hispanic.." document

  • MaxPowr 275 45/275/275 w/no L3 = 10.13 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 11.72 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 11.72 (70ns memory setting)
  • MaxPowr Pro+ 45/266/266 w/no L3 = 10.88 (60ns memory setting)
  • PwrCenter w/PF G3 65/292mhz no L3 = 11.25 (65mhz bus speed)
  • PowerForce G3 292mhz/no L3 = 9.75 sec (8500/45mhz bus speed)
  • PowerForce 250 @285mhz w/1meg L3 = 10.07 sec
  • PL Pro 233 604E @263mhz w/1meg L2 = 14.73 sec
  • Stock 8500/120 w/1meg L2 = 26.68 sec

Load & Display "Hispanic.." document

  • MaxPowr 275 45/275/275 w/no L3 = 6.31 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 7.56 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 8.19 (70ns memory setting)
  • MaxPowr Pro+ 45/266/266 w/no L3 = 6.90 (60ns memory setting)
  • PwrCenter w/PF G3 65/292mhz no L3 = 6.20 sec (65mhz bus speed)
  • PowerForce G3 292mhz/noL3 = 6.60 sec (8500/45mhz bus speed)
  • PowerForce 250 @285mhz w/1meg L3 = 6.97 sec
  • PL Pro 233 604E @263mhz w/1meg L2 = 9.81 sec
  • Stock 8500/120 w/1meg L2 = 20.20 sec

Launch Freehand 7.0 (96K disk cache setting)

  • MaxPowr 275 45/275/275 w/no L3 = 15.40 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 15.93 (60ns memory setting)
  • MaxPowr Pro 45/250/166 w/no L3 = 15.91 (70ns memory setting)
  • MaxPowr Pro+ 45/266/266 w/no L3 = 15.44 (60ns memory setting)
  • PwrCenter w/PF G3 65/292mhz no L3 = 13.96 sec ((65mhz bus speed)
  • PowerForce G3 292mhz/no L3 = 12.88 sec (8500/45mhz bus speed)
  • PowerForce 250 @285mhz w/1meg L3 = 15.37 sec
  • PL Pro 233 604E @263mhz w/1meg L2 = 17.23 sec
  • Stock 8500/120 w/1meg L2 = 20.35

Note: Disk cache was 96K during this test, as I forgot to reset it to my normal 512k size after resetting the Cuda chip during previous CPU card swaps. So I tested all cards with 96k disk cache on this test.
(the following is repeated from the PowerForce G3 review)

Application Observations: In general, most 3D applications make heavy use of Floating Point instructions, which don't show a large increase in performance with the PPC750 (vs. a similar speed 604E), since it's optimized for integer operations. Databases and most other business applications primarily use integer math, which would show a much larger advantage in performance with the PPC 750 chip.

Web Browsing shows a large increase in performance - except for slow server connections of course.

Going back to the 120mhz cpu card is painful after using the PPC750 cards. How quickly we get spoiled by speed.

Feedback
Send us your questions or feedback
Don't like forms? Then send us a quick Email instead!
= Top of Page =

= Back to WWW.XLR8YOURMAC.COM =


Copyright © , 1997/1998.

No part of this sites content is to be reproduced in any form without written permission. All brand or product names mentioned here are properties of their respective companies.

Users of this web site must read and are bound by the terms and conditions of use.