MSI MOA Benelux qualifier 2011

Tradeshow & OC events by leeghoofd @ 2011-03-30

It's that time of the year again. Overclockers all over the world gearing up to get them precious tickets for the World Finals in Taipei. But before taking the KLM 747 to Taiwan, there's a long road ahead. For our Benelux clockers an online event was created at the website. As usual Superpi 32M was one of the chosen tests. The other one was the new 3DMark011 benchmark. Weapons of choice for this online qualifier would be either a very fast SandyBridge CPU or an insanely fast ( read +6.4Ghz) Gulftown.  Since only single GPU cards were allowed GTX 580 was an obvious choice. Two teams from Madshrimps, Massman & Pt1t and Gamer & Leeghoofd submitted a score to grab at least one of the three spots for the Benelux qualifier at the MSI HQ in the Netherlands. Last wednesday we got the official mail from MSI that both the [M] teams were selected. Two belgian teams vs one Dutch Team consisting of Viss and Expat_Griz. How did the belgian waffles do in last saturdays event, time to read on...

  • prev
  • next

Round 4 and 5

Round 4 Superpi 1M :

Time to insulate the board and GPU. All teams decided to opt for LN2 cooling too on the GTX480. It's still a world of difference between each overclocker how they insulate their gear. Me and PJ usually using a lot of tissue paper and avoiding as much kneaded erazor gum on the motherboard. Our thoughts are to keep the PWMs under normal operating temperatures. With the erazor gum all over the place the cold will spread to rapidly and might limit the efficiency of the PWMs.



GPU pot was checked for good contact, before it was assembled. Loads of erazor gum though on the GTX 480 PCB. We only had the small Kingpin pots at our disposal, PJ and thomas used the FAT version, being able to maintain loads far easier. Pascal ( Teemto) did a good job on the insulation. Note that he almost never worked with LN2. Well done mate !




First thing to do is to find the cold boot bug and/or cold bug. Our system was rockstable till -100°C. From there on we would have freezes or shutdowns. Initially we topped at only 5.4Ghz for 1M, the board refused to post above, no matter what temperature or voltages we put in. Till I recalled that my review board had similar issues with certain CPU multipliers. Upping the Bclock got us a pretty stable 5600Mhz CPU. Trying to push higher would result in either a BSOD or a freeze. Maybe lower temps would have helped, but the CPU was a no go over -100°C. Mind you we were already at 1.875Vcore, indicating this CPU was not a match for Massmans ahum handpicked one. We overheared 5.9 and 6.1Ghz 1M results.  Bart (Viss) was in the same position as us, topping around 5.5-5.6Ghz... This round was pretty fast over lol... No tweaking could make up for 600Mhz in CPU difference... After a few runs we got a nice 7sec 444score. (the latter we improved to 7sec 340 at the same overal CPU clockspeed but with a bit higher uncore and ram speeds)  The focuss then went to 3Dmark 011, as we lost already some precious time just getting the CPU at 5.6Ghz.

The winning 1M score with the CPU at almost 6.1Ghz. Massman told us this was the best B1 ES CPU they ever worked with.


Round 5 :  3DMark011 Extreme preset. A benchmark more relaying on raw GPU power than CPU. So most teams opted to back down on the CPU frequency, but trying to push hard on the GTX480. Till now my GTX480 adventure hadn't been that good with either crappy cards ( hardly being able to do 900Mhz on air ) or not even being able to do 1000mhz on LN2 ( even afer being masterly modded by mista Massman )

Me and Pascal did one stock run (before the insulating) just to get a feel for the benchmark. We scored 1300ish points. So we crancked up the GPU clocks to 1000 and upped the GPU voltage to +250. Temps were far subzero. The outcome was a score lower than stock. Something was terribly wrong. We started to overlook the Nvidia driver & the operating sytem settings and found a few extra tweaks. A rerun at lower clocks, yet still we didn't manage to beat our air result. Time for a stock run then. Keeping temps just around the 0°C level. Wow we beat our previous score by over 600 points. Crancked up the speeds, keeping temps again around 0 level. The score improved again. Lowered the temps drastically and we got the infamous slow mode again. What it all boils down to was that we were unaware of the existence of switches to enable the cold slow bios and several other OC features. This information was told to us a mere seconds from the finish :p
Yet we managed to get a 2358 score with the card at 0 temperature level and at 925Mhz GPU and 2100Mhz ram clocks. Far away from the 1125Mhz GPU clocks and 2300Mhz ram speeds from the other belgian team. Next time (ofcourse if you know which hardware you will use) a read through a manual would be nice...






The best score was by the Belgian soldiers (again :p )







  • prev
  • next