Overclockin.com heeft een review gepost van de Swiftech MC2000 peltier cooler. Het ding ziet er behoorlijk heftig uit. Een PIII die met luchtkoeling 580MHz deed bleek echter niet stabiel op 600MHz gedraaid te kunnen worden:
So far, so good. Nothing that you wouldn't have expected given a processor that was already stable with air cooling at 580MHz. Now, what will happen once it is pushed to the next level? I set the system to use a 133MHz FSB with the 128MB DIMM from AZZO installed and fired the system up. The POST screen flashes up. All the drives are found. Next the Win98 splash screen comes up and the system freezes. Not what I expected. Just to verify that things were ok with the system, I popped in the 64MB DIMM of PC133 memory from Crucial and tried again. Same thing. Gets to the splash screen and freezes. Let's bump up the voltage and give it another try. 2.1V, same thing. 2.2V, same thing. 2.3V, same thing. Hmmm. What is going on here? This processor will run just fine at 580MHz @ 2.0V and won't even get past the splash screen at 600MHz? That just doesn't make sense. After a few conversations with Gabe over at Swiftech, he graciously sent me another MC2000 (the Rev. 2 version) just to make sure that there wasn't a problem with the unit I had. He didn't think there was any way that was the case, but wanted to make sure that I was able to test things out properly.
When I got in the Rev. 2 version of the MC2000 cooler, I proceeded to do the same tests with the same results. At 600MHz, it would get past the POST and freeze on the splash screen. Well, that seems to rule out a cooler that wasn't functioning properly. Since the Peltiers were definitely cooling the processor down significantly below room temperature (and below freezing if the processor isn't doing anything), what could be the problem?
After reading a few sites and posts about the cache chips on a PIII being a limiting factor (even though this processor has Toshiba 4ns cache chips rated at 250MHz), I decided to try it with the cache disabled to see if that was indeed the problem. With the cache disabled, the system booted up at 600MHz and ran Windows with no problems.