articles
FORUMS!
links
contact

PC Database

Folding Team

SETI@HOME Team
RC5 Team
Genome Team

OCAU chat

Ok, enough of the breathless paparazzi-talk. Looks are fine, but performance is what matters. Few things will take away the pain of a crawling display or jerky gameplay, especially not the thought "Yeah, but the motherboard looks sweeeeeet!" - First up, let's examine the differences between the Millennium Edition and the Pro which we reviewed earlier. It is essentially the same motherboard, so read the other review if you're looking for tables of FSB settings etc. The cosmetic changes are the most obvious - these are unlikely to offer any real benefit. The platinum heatsink may be more thermally efficient, but I don't think it's going to be a major difference to the Gold model, or even the more conventional green sink found on the plain Pro board.

The 2 ISA slots from the Pro are gone, having been replaced with another PCI slot for a total of 6. Interestingly, the IRDA header has moved from near the PCI slots to near the floppy connector. The board is slightly wider to accomodate the designer's names and the platinum plate. Bios v1.06 for the Pro II added Coppermine support. Curiously, 1.06 is the highest available on the website, but this unit shipped with v1.08. As mentioned earlier, there are larger capacitors near the processor slot for added stability.

So what about performance? In my tests, the 2 boards were identical performance-wise at 366MHz (66Mhz FSB), 400 (66), 550 (100) and 600 (100). Sandra's CPU, Memory and Multimedia benchmarks showed only very slight differences, enough to be dismissed as experimental error.

Ok, so at the same MHz they offer the same performance. What about those big capacitors on the Pro II, though? Do they let you get to higher FSB's? Now, this is a tricky thing to test. Both boards have a maximum FSB setting of 153MHz - that is seriously moving. Even putting a Celeron 300A onto the board at that speed would be trying to run it at a ludicrous 688MHz - I haven't heard of anyone running a C300A at that kind of speed, even the nutters with cascaded peltiers and freon cooling. Fortunately, I happen to be the very lucky owner of a non-clock-locked P2-400. With this chip I was able to set the multiplier to 2x. Even at 153MHz FSB it's only trying for 306MHz, well within the core's capability. Of course, running at that high an FSB on a BX-based board introduces a whole host of other problems. Given that the goal here was to prove the stability of the MOTHERBOARD, I slowed everything else to the point where it wouldn't die. An AGP card was simply out of the question - even at 2/3 AGP/FSB setting, it would be a blistering 102MHz and frankly I value my Viper Ultra too much. So my only PCI video card, a Matrox Millennium (odd coincidence?) 4MB was dusted off and brought out into the light. I used a single 64MB stick of Kingmax PC133 RAM set to 3-3-3 - it needs CAS3 even at 133MHz - in bank 1. UDMA was turned off for the IDE drives. I also found that turning "Spread Spectrum" on increased the stability of the boards. Actually, ON is the default from BIOS r1.06 onwards on the Millennium Edition but I kept turning it off out of habit at first.

So what speeds could I reach with each board? The results are quite surprising. I admit, I was hoping to find the Pro model would fail at some ludicrously high speed and the Millennium Edition would get a chance to shine above it. However that simply wasn't the case. Both boards ran rock-stable at their maximum setting of 153MHz FSB. Let me say that again. Both boards ran ROCK-STABLE at 153MHz. Remember the BX chipset was designed to run at 100MHz, kids. This is not just interesting trivia, this is outstanding. Check the screenshot:

there is no bigger picture, my ISP would kill me, thousands of people downloading a 150kb jpg.. cope with it :)

Note SETI and the Unreal flyby happily chugging along. Ignore the internal speed of the CPU and focus on the external speed, the speed of the motherboard. I don't know why WCPUID is trying to tell us it's a Celeron with no L2 cache when it is identified by the BIOS as a P2 with cache enabled - this is no doubt an artifact of the high FSB. Strange things happen on the edge.. (UPDATE:Daevon let me know: That's not an high fsb issue that the cache is turned off. ALL of the Pentium 2 Cpus AUTOMATICALLY disable L2 Cache at 2.0X and 2.5X multipliers. In fact, you can try even the 66X2 setting: you'll see that the cache is still disabled! - OK, you learn something new every day. :) )

What this essentially means is this - if you're building an extreme rig or even just cranking your chip to the conventionally-cooled limit, you can completely eliminate the motherboard as a source of instability. If your machine is failing at high overclocks, it simply must be some other component. Now, I admit, I only ran it for about 15 minutes at a time on each machine - the non-heatsinked MGA chip on the video card was becoming very warm due to the overclocked PCI bus. I had no problems during those runs. If you were seriously building a machine to run at anywhere near these speeds and wanting to run the other components in a similarly tweaked state, you would have to choose those components very carefully.

Something I forgot to mention in the previous review but is present on both boards is the ability to write the CMOS values into an EEPROM. This is great, you save a current "safe" set of settings, then try for something more extreme. If it fails to boot, you hold down HOME while turning the PC back on (so it reboots with defaults loaded), load the full set of values from the EEPROM and you're ready to start tweaking again. No more wading through BIOS screens re-setting everything between tests.

In conclusion - an excellent motherboard. For it to be anything less, following in the footsteps of such giants, would be a genuine surprise. There are some niggles - they're covered in the earlier review of the Pro board, mostly centering on the (comparative) lack of FSB settings and limitations of the BX chipset (not the fault of the board). I guess the price premium over the standard Pro will scare some people off, too. I'm sure people will buy it for the looks and the exclusive, almost collectable nature. Given there's only 2500 to be made I'm sure AOpen won't find many left gathering dust on shelves. In my testing the Pro II was not noticeably more stable than the standard Pro board - perhaps some longer-term testing will allow it to shine there. However this is only due to the excellence of the board upon which this model is based, with both demonstrating rock-solid stability right up to their high FSB limit. Regardless, it wins absolutely hands down in terms of sheer coolness - it really has to be seen to be appreciated. Obviously, this board means a lot to the designers and rightly so. The family of motherboards it culminates has meant a lot to the overclocking and hardware community and this is a truly outstanding tribute to that.