Umax SuperMac J700

The UMAX SuperMac J700 is the first "Power Mac" I benchmarked. My J700 has a 180 MHz 604e CPU, 104MB of memory, and a Quantum Fireball 2110 hard drive. It shipped standard with an ixMicro Twin Turbo 128 video card, which I have just replaced with an ixMicro Ultimate Rez 3D card.

The drive was not optimized before benchmarking.

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

Speedometer 3.06

The system was tested on 26 November 1998 using Mac OS 8.1 with all inessential extensions off. Computer attached to a 17" color monitor and tested in 8-bit video mode at 640 x 480 resolution. Results are relative to a Mac SE or Classic, which rates 1.0. Numbers rounded off to two decimal places.

The first set of numbers compares performance at different cache settings. These results are without graphic acceleration.

cache   CPU  graphics  disk   math 
 96KB   87.6    15.8   6.19   377.6
128KB   83.4    16.6   6.35   377.9
256KB   83.4    16.8   6.40   378.2

The cache setting should have little influence on non-disk tests, which these numbers bear out. With this particular setup, cache size makes only a small difference in performance.

Since Speedometer 3 is completely written in 68K code (it predates the PowerPC), I also tested it with Speed Doubler 8, which claims to have better 68K emulation than Apple provides.

test     CPU  graphics  disk   math 
w/o SD   83.4    16.8   6.40   378.2
with    159.3    16.8   6.46   490.6

This bears out the claim by Connectix that Speed Doubler has better emulation than Apple - the CPU score almost doubles with Speed Doubler 8 and the math score improves by about 30%.

Finally, I ran comparisons with two different video cards, both with and without video acceleration. (base = no acceleration, Twin = Twin Turbo, TT/SD = Twin Turbo plus Speed Douber, and Ult. = Ultimate Rez 3D)

test     CPU  graphics  disk   math 
base    83.4    16.8   6.40   378.2
Twin    87.6    55.7   6.12   377.0
TT/SD  159.3    70.3   6.22   494.5
Ult.    83.4    58.1   6.07   377.6

Except for disk results, the combination of an accelerated video card and Speed Doubler 8 is a winning combination when running 68K programs.

Speedometer 4.02

The system was tested on 26 November 1998 under Mac OS 8.1 with all inessential extensions off. Computer attached to a 17" color monitor and tested in 8-bit video mode at 640 x 480 resolution. Results are relative to a Quadra 605, which rates 1.0. Numbers rounded off to one or two decimal places.

The first set of numbers compares performance at different cache settings. (Because Power Macs don't support 1-, 2-, or 4-bit video, Speedometer 4 was unable to test the graphics.)

cache   CPU  graphics  disk    math
 96KB  11.96   n/a     2.40   532.7
128KB  11.99   n/a     2.39   533.1
256KB  11.93   n/a     2.49   533.2

The cache setting should have little influence on non-disk tests, which these numbers bear out. As above, with this particular setup, cache size makes no significant difference, except that the disk score is a bit higher with a 256 KB cache.

Testing with Speed Doubler verified that Speedometer 4 has PowerPC code: Speed Doubler 8 didn't make any difference on benchmark results.

MacBench 3

The system was tested on 26 November 1998 under Mac OS 8.1 with all inessential extensions off. Computer attached to a 17" color monitor and tested in 8-bit video mode at 640 x 480 resolution. The disk cache was set to 256 KB for all tests. Results are relative to a Power Mac 6100/60, which rates 10. Numbers rounded off to two decimal places.

test       CPU    math    disk  graphics
VM off    44.75   42.47   13.86    20.95
VM 105MB  43.91   42.17   13.62    18.20
VM 128MB  41.74   42.52   13.46    18.48
VM 208MB  41.77   41.94   13.53    18.62

There are several claims about virtual memory. One is that setting VM to 1 MB more than physical RAM (in this case, 104MB) provides the best balance of speed and efficient memory use. Others claim the best results come when VM is set to a multiple of 32 MB. Still others, that making VM double installed RAM is best.

These figures show that there is a small (2-7%) reduction in CPU performance with virtual memory enabled and no significant change in math performance. Although some claim disk performance improves when VM is used, these tests do not bear out that claim.

The great suprise is the 12-15% reduction in graphics performance when VM is running.

All this should be weighed against the great advantage of VM in a Power Mac: it reduces the amount of RAM most programs need to operate.

Finally, I ran comparisons with two different video cards, both with and without video acceleration. (base = no acceleration, Twin = Twin Turbo, and Ult. = Ultimate Rez 3D)

test     CPU    math    disk  graphics
base    44.75   42.47   13.86    20.95
Twin    44.39   42.22   13.24   136.04
Ult.     n/a     n/a     n/a    167.31

The graphic acceleration didn't really change CPU, math, and disk scores, so I didn't repeat them for the second video card. Overall, the Twin Turbo benchmarks 6.5 times faster and the Ultimate Rez 8 times faster than running either card with acceleration disabled.