The Performance Edge: A Closer Look at XLR8's
Cache Performance Boosting Software.
A couple of weeks back we received a press release from XLR8 announcing their new MACh Speed G3 Control software. The G3 Control software, which comes bundled with XLR8's G3 upgrade cards, allows you to change the speed of your processor's backside cache. MACh Speed G3 Control works with other accelerator cards as well as native G3 machines. For our previous Performance Edge article we ran MacBench 5.0 on our "revision B" iMac, using XLR8's software to bump the backside cache from 116MHz to 155MHz. The results showed a 20% improvement in the processor score. We recently ran more extensive benchmarks on our G3 PowerBook/266 and unfortunately the results were not as dazzling.
Installation: Was quick and simple. The installer placed an extension and control panel which allows you to change the backside cache speed. A couple of other items are placed on your drive as well, a program called "Power Control" which you can use to run tests on your processor and RAM a la TechTool. Power Control also provides you with information on your current hardware configuration. The last item installed is "PowerFrax" a fractal generator with built in timer. The idea presumably being that you can run PowerFrax with different cache settings to test for performance gains. I didn't note a significant difference with any of the settings including with the cache disabled!
On first restart, the MACh Speed software runs a quick test to determine the fastest speed the cache can safely use. You are then presented with a dialogue asking if you want to use this default speed or try a faster speed which also might work. Throwing caution to the wind I selected the fastest choice available and promptly crashed my PowerBook! I rebooted and humbly selected the "default" button in the MACh Speed control panel. Below you will find results from MacBench as well as several "real world" tests that I ran on our PowerBook.
Question: We are puzzled by the lack of performance gains
in our PowerBook especially after the impressive results in our original
tests on the iMac. Even more perplexing are the real world results with
the backside cache disabled. We have sent off our results to XLR8 and will
post their response, if any, here. Readers on our discussion
board have noted similar
findings to the ones above. We would like to hear your thoughts as well.
Have you used this cache utility under real-world conditions? What are your
your input with the MacSpeedZone community.