Of course, we had to try overclocking on this board. Before I share the results let me explain what we were doing for cooling. First off it's all passive (heatsink/fan based) for now. Maybe some watercooling or peltier action later, but keep in mind this system was on loan and it would make an expensive paperweight.
The Athlon 500 we got for testing was an OEM unit so came with no fan. EYO supplied the GlobalWin VEK-32 for use on it, which is kinda like 2 CPM25603-32's stuck side by side. (If you're not familiar with that heatsink, check out our socket370 cooler comparo from a couple of months ago). This uses 2 of the same fans from the CPM-32's for a total air-moving capacity of 52cpm - no slouch.
However, we had some trouble fitting the heatsink to the board because of the proximity of the ATX power connector to the Slot-A. You can see in the following picture that the leftmost fan actually rests on top of the connector - there's no way the power cable will go in.
Eventually, I pulled one fan off, and mounted a similar 60mm fan - but from the CPM25603-16. This fan is slightly narrower and only pushes 20cfm (6 less than the -32's original fan). Leaving one corner with no screw in it allowed the assembly to squeeze into place with no stress on the processor slot:
..and now I was set for some overclocking! 110MHz FSB, or 550MHz from our 500 unit, was a no brainer, simply change the FSB setting in the BIOS. It ran stable for a couple of hours of Driver (this stress testing is terrible work, I can tell you) with SETI in the background. 120 refused to POST. 115 would lock up halfway through loading Win98. 111 would load Win98 but crash shortly after loading Driver. Opening the case and ducting cold air onto the CPU fans with another fan prolonged it slightly, but the results were similar. I tried boosting the core voltage using the jumpers on the board (vcore adjust in the bios woulda been nice, guys) .. from the default 1.6v to 1.7v made no difference, at 1.8v it made no difference except the board beeped continuously during the boot process.. I decided not to try any higher (the board allows up to 2.0v) with this cooling. Even setting the mysterious "VIO" jumper to the highest setting, which is supposed to boost the voltage to the chipset, ram, PCI etc had no effect. Here's some pics at 550:
So where does that leave us? I definitely isolated it to the CPU or the motherboard by slowing the ram down, turning off UDMA etc.. no improvement. Some people will say "The Athlon 500 can do 650 easily!" but I would say not ALL can do 650 guaranteed.. however, if we're debating whether it's the CPU or the board that is stopping at 110MHz, a few things stand out. One is the razor-sharp line of working and not working. At 110MHz it works at default voltage. At 111MHz it won't even work with +.2v in the core. You would expect the CPU to go a little higher with more volts in the core. Secondly, it's an odd coincidence that, despite the BIOS offering adjustments way up to 150MHz+ for FSB, the dipswitches to set it on the board only allow a maximum of..? 110MHz. Hmm. I also wonder about the location of the North Bridge chip, right next to the Slot-A and therefore right in the way of the hot air escaping the CPU heatsink. Or, perhaps, the board itself (possibly the chipset specifically) really just isn't up to the task.
During the testing, the reported temperature never went above 28C, and was usually 26C. I'm not sure what "normal" is for an Athlon, and of course I can't measure the on-die temperature, but it doesn't seem too extreme to me. Certainly the casing of the CPU, at the back of the core, was not overly hot to the touch. I don't think heat was a major factor.
This testing revealed another feature of this board. When staring at the blank non-POSTing screen, I had the usual thoughts about having to clear the CMOS via the jumper on the board, and mentally kicked the board for not having a quick-restart key such as many others do - you know, turn it on while holding down INS, or DEL, or HOME, and the system will boot up with the default speed set. Well, I turned the Athlon system back on for one more try at 120MHz - it POSTED, and in cute Twinglish flashed on the screen "The CPU overclock had been happened - please press DEL to enter setup.." ..it was back at 500 again. Somehow this board can detect a failure to POST due to over-enthusiastic overclocked settings having been selected, and the next time it's booted resets itself to the default setting automatically - neato.
Summary:
Well.. in some ways it's tough to make a decision on this board.
It's hard to proclaim it an overclocking-friendly board - certainly
it is more so than any other Athlon board to date, but a 10% increase
in speed is not going to blow anyone's socks off - but keep in
mind it's 10% more than any other board is offering! This is dissapointing,
but I think, for the moment, major Athlon overclocking remains
in the hands of the soldering-iron wielders. So, if we set aside
major FSB-based overclocking for now and also set aside the fact
that it's an Athlon board.. it is packed with features to rival
the greatest BX-based boards. Hardware monitoring, UDMA/66 support
and other features all earn the board respect. The fact that it
is not officially supported by the manufacturer is dissapointing,
because it may be tricky to get BIOS updates etc in the future.
When you DO factor in that it is an Athlon board, though, and
take a look at the other competitors in that field.. it's a clear
winner. The K7M is definitely the Athlon board of choice for now.
Overclockers Australia would like to thank EYO Technologies (http://www.eyo.com.au) for lending us the review board and CPU.
Updates:
You may have noticed the motherboard-closeup photos on page 1 have the power connector in a different place to the board we reviewed. That's because that kind of straight-on motherboard photo is really hard to do, so I borrowed the photo from www.bxboards.com - read their review of the K7M here. I guess that's related to why theirs needed the voltage regulator heatsink but ours didn't - obviously some work been done on power supply in the later revisions of the board.
Ken Peter sent word that TAKA (in Japan) make an Athlon cooler that looks like it might clear the power connector on this revision of the board.
Some very eagle-eyed readers on AcesHardware spotted my mistake - the picture that used to be on this page, showing the North Bridge under the green heatsink, was of a VIA part and not the expected AMD unit. Well, the theories being batted back and forth on the discussion group there were fascinating, but it was a simple case of "too many reviews, too many photos in the digital camera, too little sleep" .. sorry to anyone who got freaked out by that. You guessed right, I have a Soyo SY-6VBA here for review, which uses the VIA Apollo Pro chipset, hence the pic. :)
