PowerBook G4/400

I got my 400 MHz TiBook at the end of January 2001. As I was preparing to replace the stock 10 GB Toshiba hard drive with a newer, faster 20 GB IBM drive, I realized I'd never done a benchmark page. Well, here it is.

The TiBook has 512 MB of RAM, is running Mac OS 9.2.2, and has not been optimized in the least. Except for changing the size of the disk cache for some tests, these benchmarks were run with my usual complement of fonts, control panels, and extensions.

Hard disk benchmarks were also run on an external Western Digital 80 GB 7200 rpm hard drive with an 8 MB onboard cache in a FireWire enclosure using the Oxford 911 bridge - since this is the drive I used to store my files while moving from the old drive to the new one.

The drives:

• Toshiba MK 1016GAP/HDD2152, 4200 rpm, 1 MB buffer
• IBM Travelstar 40GNX, 5400 rpm, 8 MB buffer
• Western Digital WD800JBRTL, 7200 rpm, 8 MB buffer

Remember that benchmarks are arbitrary. They measure certain types of performance that may or may not reflect the way you work.

Speedometer 4.02

The system was tested on 5-6 October 2002 under Mac OS 9.2.2 in my standard configuration. Display run at default 1152 x 768 at millions of colors. Results are relative to a Quadra 605, which rates 1.0. Numbers rounded off to one or two decimal places.

Benchmark numbers compare performance at with three different hard drives. All tests were run with the cache at the 8 MB default setting.

drive      cache   CPU  graphics  disk    math
Toshiba 10G  8MB  30.61   n/a     3.35  1124.6
IBM 20G      8MB  29.88   n/a     4.54  1123.7
WD 80G       8MB   ---    n/a     4.80    --- 

The CPU and math settings are virtually identical. Both the external Western Digital drive/FireWire combo and IBM Travelstar drive handily outperformed the stock Toshiba drive. In the case of the IBM drive, the 5400 rpm speed and larger cache managed to boost overall drive performance by 35%.

MacBench 5

The system was tested on 5-6 October 2002 under Mac OS 9.2.2 in my standard configuration. Display run at default 1152 x 768 at millions of colors. The disk cache was set to 256 KB or 8 MB as indicated. Results are relative to a Power Mac G3/300, which rates 1000.

drive       cache  CPU     math    disk graphics
Toshiba 10G  256K  1282    1519     896     n/a 
Toshiba 10G   8MB  1277    1509    1274     n/a 
IBM 20 G     256K  ----    ----    1676     n/a 
IBM 20 G      8MB  1279    1518    2193     n/a 
WD 80 G      256K  ----    ----    1676     n/a 
WD 80 G       8MB  ----    ----    2445     n/a 

These figures show that there is huge difference between running with a 256K disk cache and a much larger 8 MB one. Everything about the TiBook felt slow with the smaller cache - booting, opening files, you name it. If you're not using as big a disk cache as possible, you're not getting all the performance your Mac is capable of.

Still, even with the big cache the Toshiba drive didn't hold a candle to the performance of the two other drives -- even when they were set to the smaller 256K cache. Overall, the Western Digital drive in the FireWire box provided almost twice the performance of the Toshiba drive, and the IBM Travelstar was over 70% faster.

SpeedRun & OS 9

There's a new benchmark in town. SpeedRun benchmarks Macs running the classic Mac OS and is also available in an OS X version. Results cannot be directly compare between the two operating systems.

The system was tested on 5-6 October 2002 under Mac OS 9.2.2 in my standard configuration. Display run at default 1152 x 768 at thousands of colors, as required by SpeedRun. The disk cache was set to 256 KB or 8 MB as indicated.

drive       cache graphics  disk  CPU   RAM 
Toshiba 10G  256K    548    425   249   1337
Toshiba 10G   8MB    690    679   266   1358
IBM 20G      256K    ---    562   ---    ---
IBM 20G       8MB    695    748   263   1246
Reduced       8MB    525    586   198   1025
WD 80G        8MB    ---    687   ---    ---

Just for kicks I ran SpeedRun at the reduced CPU speed of 300 MHz, which is an option in the Energy Saver control panel. It makes a significant difference.

Again, the difference between a 256K disk cache and 8 MB is monstrous. If you've got the memory, you should use as large a disk cache as the Mac OS allows. With this benchmark, the IBM Travelstar doesn't show off nearly as well, offering only a 10% higher score on the disk benchmark.

That's one reason for running several benchmarks. Each provides a different picture of how much certain changes affect the overall system, and drive performance varies depending on how your applications are using the drive.

SpeedRun & OS X

This is the first Mac I've benchmarked under OS X.

The system was tested on 5-6 October 2002 under Mac OS  X 10.1.5 in my standard configuration. Display run at default 1152 x 768 at thousands of colors, as required by SpeedRun. The disk cache cannot be set in Mac OS X. Disk numbers are the average of three tests.

drive       graphics  disk  CPU  RAM
Toshiba 10G    210    257   137  261
IBM 20G        210    302   138  263
WD 80G         218    269   138  263

The IBM Travelstar acquitted itself very nicely, coming in at a 17.5% higher score than the stock Toshiba drive. What's really surprising is how little difference there was between the fast 7200 rpm Western Digital drive in a FireWire enclosure and the original Toshiba hard drive.

The computer was tested again on 14 February 2003 under Mac OS X 10.2.4 after optimizing the drive with Norton Speed Disk.

drive       graphics  disk  CPU  RAM
IBM 20G        182    273   145  260

The CPU score is up about 5%, the RAM score is virtually unchanged, but both the disk and graphics results are worse than under 10.1.5.

Let1kWindowsBloom

Let 1000 Windows Bloom has become a standard graphics benchmark for Mac OS X. Run on the 400 MHz TiBook under Mac OS X 10.2.4 on 14 February 2003, it turned in times of 57 seconds in both 16-bit and 24-bit video modes.

Xbench

Another new, OS X-specific benchmark program is Xbench, which runs a fairly comprehensive set of benchmarks. Following results were obtained on 14 February 2003 under 10.2.4.

Test            16-bit  24-bit
CPU              38.5    37.6 
Thread           29.8    29.7 
Memory           65.6    63.4 
Quartz           49.2    52.6 
OpenGL           50.0    45.8 
User Interface   50.0    47.0 
Disk             58.8    59.0 
Overall          45.9    45.1 

Based on these tests, the TiBook averages 2% faster overall in 16-bit mode, with the biggest difference (almost 10%) in the OpenGL test. Interestingly, Quartz is more efficient in 24-bit mode, and the TiBook benchmarks roughly 5% better with the higher bit-depth. Except for programs that take advantage of OpenGL, there is no compelling reason for switching to 16-bit video.

Conclusion

The stock Toshiba drive was no slug, but I had nearly run out of space. With under 1 GB free, it was only a matter of time before I ran out of room. DealMac recently featured a bargain price of $101.50 shipped from Googlegear, and I'll probably sell my older Toshiba drive to one of my sons to replace the 4 GB drive in his WallStreet, so my net cost will be minimal.

Overall, the IBM is a much perkier drive. The computer boots faster in both OS 9 and X, applications launch faster, and BBEdit zips through global search-and-replaces faster than ever. With nearly half the drive free, I don't think I'll outgrow this until it's time to replace the whole computer - and that's something I hope to postpone until January 2004 if at all possible.