Overclocking and Benchmarks for the 6500

See the excellent 6400 Zone for info about the similar 6500!

overclocking
benchmarks
wannabe G3 benchmarks
Sonnet G3L2 benchmarks
return to the home page

overclocking

overclocking the 6500 can be accomplished rather easily by moving resistors around on the motherboard.
information courtesy of http://www.bekkoame.or.jp/~t-imai/pm55e1.html. Also visit Mike Breeden's excellent overclocking links at xlr8yourmac

repeated here are the relevant resistor settings (red) to overclock a 250 MHz machine to a 300 MHz machine:


CPUClock Config  R132   R133   R131   R130   R326   R330   R329   R328
250MHz    x5      S       i     S      S       x      S      x      x
300MHz    x6      S       S     i      S       x      x      S      x

where S means in-place, x means missing, and I is irrelevant.

to overclock the 250 MHz motherboard to 300 MHz, move resistor 131 to 133 and resistor 330 to 329. That is all.
to move the resistors, it is required to pop the heat-sink off the 603ev. This is not difficult. Be sure to apply fresh heat-sink silicone compound before reassembly. You might consider installing a larger fan in the process.
Final Note: the work described here will invalidate any warranty you have on the 6500. Proceed only if this is not an issue to you.

benchmarks

Below are some general benchmarks I have made for the 6500/250 with and without a 1-Meg cache, and with and without overclocking to 300MHz.

Macbench 4.0 scores (with percentage gain from same-clock baseline):

MacOS 8.1.
The only 3rd party xtens: Speed Doubler 8.1 and CDRomToolKit 3.0.2. (my faves)
Fresh boot before each set of benchmarks run.
96Meg physical RAM (32 original).
Apple VM on, set to physical+1Meg.
Disk cache = 7680K.
Monitor resolution 1024x768, thousands of colors. Factory video.
Apple 15" Multisync AV (not blinking with odd colors... yet!)


                  250MHz/0-Meg      250MHz/1-Meg
    Processor:        259             433 (67%)
    FPU:              280             332 (19%)
    Disk:             270             336 (24%)
    Pub disk:         210             256 (22%)
    Grafix:           216             316 (46%)
    Lo-rez grafix:    200             289 (45%)
    CD:               364             453 (24%)

                  300MHz/0-Meg    300MHz/1-Meg
    Processor:        271             463 (71%)
    FPU:              296             383 (29%)
    Disk:             266             341 (28%)
    Pub disk:         205             233 (13%)
    Grafix:           225             335 (49%)
    Lo-rez grafix:    208             302 (45%)
    CD:               369             471 (28%)

Notes:
- The addition of the cache yields reassuringly similar results for 250 and 300 MHz. Comparing 300 to 250 MHz under the same cache conditions shows 300 MHz to uniformly yield (on average) significantly less on processor and FPU than the 20% one might guess from just comparing raw clock speeds, which is somewhat naive but which serves as a rule of thumb. There are some anomolous results in the disk scores. All benchmarks should all be taken with a grain of salt, as some of the comparisons don't make perfect sense...
- The drive was not defragged before any benchmark, but the computer was also relatively unused over the test period, so I believe the relative fragmentation effect, if any, was small and constant.
- Although the gains are minimal, the box remains at 300 MHz since I imagine myself a speed phreak. There has been no degradation in reliability, with one exception: Carmageddon will regularly crash the machine with the stock fan installed. A heftier fan has eliminated the crashes. For those who want to argue the point, I have confirmed this by swapping the fans in and out. Otherwise this machine is rock solid, one of the best Apple has ever made, IMHO, although it was overpriced at its inception.
- The shrewd observer will wonder why the 250 MHz benchmarks are without the stock 256k cache. 1) I wanted to see the maximum range. 2) The 6500/250, bought as an open box buy, was missing the cache! Perhaps sold on the black market by a Circuit City employee?!?!?
- I am unaware of any difference between the 6500/250 and 6500/300 motherboard except the changes necessary to overclock. If there are other differences, it might explain why *this* 300 did little better than the 250 in general for the same cache.

wannabe G3 benchmarks

This is a brief comparison of my overclocked max-cached 6500 to one published review of a G3 300MHz upgrade that I am aware of. Below are my results and comments...
- The review's benchmarks:
```
MacBench 5.0 processor score.......596
Launch Communicator 4.04...........09 sec
Launch Communicator 4.04...........04 sec       (second time)
Launch 7.7Meg TIFF Adobe PS 4......15 sec
Launch 7.7Meg TIFF Adobe PS 4......09 sec       (second time)
```
- My 6500 benchmarks:
```
MacBench 5.0 processor score.......422
Launch Navigator 4.08..............05 sec
Launch Navigator 4.08..............03 sec       (second time)
Launch 7.2Meg TIFF Adobe PS 5.02...23 sec
Launch 7.2Meg TIFF Adobe PS 5.02...20 sec       (second time)
```
Obviously there are differences. I used Navigator (standalone) whereas the reviewer used Communicator. I also used Adobe PS 5.02 as opposed to 4, and my TIFF file size was slightly smaller. Some of these differences may be significant. In addition, by the time I did these last benchmarks, the 6500 had been upgraded to 8.5.1 and Speed Doubler 8.1.2, so comparisons to the Macbench 4.0 numbers may be inaccurate, even if the differences in Macbench 4.0 and 5.0 are taken into account.

Sonnet G3L2 benchmarks

I have installed a Sonnet 500MHz G3L2 card in a 6500, along with a Voodoo3 2000 card.
MacBench 5.0 scores:
The similar launch times are clearly more a measure of disk speed than anything else. The machine is very much faster and much more enjoyable with the 500 upgrade. I strongly suggest tha the Sonnet 400/200 card be avoided. For more information, e-mail me or see my G3L2 upgrade review on xlr8yourmac. I haven't yet tried the G3L2 card with LinuxPPC!

rapierbit