Remember the most important results are on the next page of the review which covers real world applications and game performance. This is where the "rubber meets the road" so to speak. Benchmarks have their place, but actual applications performance is what really matters.

Be aware that Macbench Graphics scores will vary depending on graphics mode and the installed video card or graphics chip. In Macbench 5.0 a 1000 score is the baseline based on performance with an Apple Beige G3/300 with RageII chip (onboard video) running millions colors, 1152x870, so consider this when evaluating any scores at lower resolutions and color depths.

MacBench 5.0 Tests: Details of the systems shown in the graph of results.

• The original PowerBook G3/250 system
  (OS 9.04, VM Off, 20GB IBM Drive, 83MHz bus speed, 125MHz cache speed)

• The same system with the Newer Tech MAXpowr G3/466 upgrade installed
  (66MHz bus speed, 233MHz cache speed)

• A PowerBook G3/500 Firewire
  (OS 9.0, VM Off, original 12GB drive, 100MHz bus speed, 200Mhz cache speed)

The graph below compares my Wallstreet G3/250 before and after the upgrade, as well as results of the PowerBook G3/500 Firewire model previously reviewed.

I can only guess that the PB G3/466 MacBench CPU score was slightly higher than the PowerBook G3/500 due to the Wallstreet having less extensions (no USB or Firewire extensions, which the PB Firewire has for its built-in ports), or perhaps the fact the PB Firewire used OS 9.0, vs 9.04.

Notes: Some explanation of the MacBench scores and test system components.

• Disk Scores: The Wallstreet PB hard drive performance is higher than the PB G3/500 Firewire's 12GB drive due to the fact I upgraded to a very fast IBM 20GB hard drive recently (the procedure is shown in this article, as well as comparisons of drive performance before and after the upgrade).
  
  
• Graphics Scores: My Wallstreet PB G3/250 has the ATI RageLT graphics chip, which is slower than later (Wallstreet 2/Mainstreet and Lombard models) that have the RageProLT chip, and much slower (especially in 3D/games) than the current PowerBook Firewire models' Rage128 Mobility chip which also has twice the video RAM of my Wallstreet - 8MB vs 4MB. (Althought the Lomboard models have 8MB of VRAM, the Rage128 Mobility is a better performing chip. See the PB G3/500 full review for 3D/game performance.)

Fractal FPU Benchmark:

I also used a freeware Altivec Fractal demo program (available here) to compare FPU performance with the original G3/250 and the G3/466 Upgrade. (MegaFlops/sec = Millions of Floating Point Operations per Second)

• PB G3/250: 159.9 MegaFlops/sec
• With G3/466: 293.0 MegaFlops/sec

As expected, the G3/466 nearly doubled performance in this test.

Memory Bandwidth Tests:

I used Newer Tech's freeware Gauge Pro to test Memory Bandwidth before and after the CPU upgrade. Due to the lower bus speed, memory bandwidth was actually about 3MB/sec less with the G3/466 (233MHz cache speed/66MHz bus speed) vs. the original G3/250 (125MHz cache speed/83MHz bus speed) - 74.7MB/sec vs just over 71MB/sec. For comparison, a PB G3/500 (200MHz cache speed, 100MHz bus) reported 89.3MB/sec. In most applications this small difference in memory bandwidth would not have any effect, even if CPU speeds were identical. CPU speed is the primary performance factor as you'll see on the next page of applications performance.

CPU Temperatures:

I also used GaugePro's CPU junction (internal) temperature reporting feature to compare readings before and after the upgrade. (The CPU temperature was checked immediately after the Infini-D 30+ minute rendering test.)

• PB G3/250: 75C (max)
• With G3/466: 59C (max)

The lower temperatures with the G3/466 is due to the fact that CPU runs at a lower core voltage, which offsets the higher clock speed. For reference, I've seen PowerBook G3/500 models hit near 60C temperatures (indicated) after extended use. (The new Powerbooks have a fan that will turn on once CPU temperatures exceed about 55C in my experience.)

My Wallstreet PB G3/250 has run 75C reported temperatures since I've owned it (summer 1998) with no ill effects other than the bottom getting very hot. The original CPU is rated for either 85C or 105C. (I can't check this due to the CPU markings being covered by the disc shaped heatsink.) Desktop G3 CPUs are usually rated for 65C max operating temperatures I believe.

Battery Runtime:

To compare battery run-time with the MAXpowr G3/466 CPU upgrade vs. the original G3/250 CPU, I created a RAMdisk with a copy of Quake1. I then shut down, disconnected the AC adapter and rebooted with a fully charged battery and let Quake1 run its looping demos. The purpose of the test was not to see how long I could extend battery life, but just to see if the faster CPU had any negative effect on battery runtime in as fair a test as possible. Since the 466MHz CPU would be running more loops of the demos, running from a RAMdisk prevented its more frequent accesses to the hard drive from affecting the times.

The same battery (original one) was used for both tests, with the display set to half brightness, sound to mid volume level. As I noted in the news, the tests with the G3/250 CPU ran until the battery completely drained with no warning (system just shut off after 1 hour, 57 minutes). Apparently the game code did not allow the Finder's low battery warning to be displayed. This resulted in a corrupted disk. For the tests with the G3/466, I didn't want to repeat this mistake, and had the control strip visible. After 1 hour and 55 minutes, the control strip showed 10 minutes of battery time remaining. I considered this basically a draw. Despite the higher clock speed, the lower core voltage of the G3/466 seems to have no negative effect on battery life compared to the original G3/250 CPU.

The next page of the review covers performance in real-world applications.