Introduction

Shocker! There has been a European overclocking competition and no Belgian team was present! Although we'd like to attribute our non-presence to many different causes, there's actually a very simple explanation: the Belgian team, formed by Pt1t and Massman, lost in the Benelux live event. So, instead of the Belgian team, there was a Dutch team present in Munich: Gendo and Wava, two relative newcomers in the overclocking scene. As they said themselves before the event, the expectations weren't set that high, so, because of the low expectations the team could start the competition as underdog and try to surprise the overclocking world by beating some of the top teams of Europe.

Before we start throwing pictures and detailed information about the event, let's talk about the prizes first. Basically, the first three teams of the event received a cash reward ($2000-1000-500), but more importantly, the five best European teams received a ticket to participate in the World finals in Beijing end of August. Indeed, the top five, which is much more than solely the European champion, which was the case at the Gigabyte overclocking competition. And that wasn't the only thing different from the Gigabyte competition.

First of all, instead of just one single day of overclocking, the 17 teams had two days to get the most of their system in three different benchmarks: Superpi 32M, Aquamark3 and 3DMark03. At first sight, maybe a weird choice knowing that each benchmark is quite old, but since everyone is competing with the same hardware, that doesn't really matter. In fact, it's probably one of the best choices of benchmarks I've seen on an overclocking event. Why?

When we talk about overclocking events organised by a major hardware manufacturer, we need to take into account that this mainly is a marketing tool. The money spent on such events (the whole picture) is in fact quite a lot and it would be naive to think it's solely to please a bunch of extreme overclockers. Secondly, it's also important for the manufacturers to reach a larger audience than only the overclocking community. Judging by the number of participants on Hwbot.org, which is the world's only continues overclocking competition, that would be an audience of 20.000 people; 30.000 if we take into account that hwbot isn't that popular in Asia. The amount of hardware enthusiasts and, in general, people who are interested in IT is much larger, thus more important. And that's where the selection of benchmarks is important: using older benchmarks may be less interesting for those who don't know the benchmarks and/or the difficulty level of it. Only the true benchmarking community can appreciate, for instance, a 32M result of 6 minutes 50 seconds with an i7 920.

An extra reason why the three benchmarks are so interesting is because each benchmark requires a different skill. For Superpi 32M, it's important to work hard to get the memory subsystem working as performant as possible (notice: I don't use the word 'fast') and also to have the operating system optimised to run low 32M times. To give you an idea: without optimising the memory subsytem, you can lose up to 20 seconds and without optimising the operating system possibly even more! 3DMark03, on the other hand, is a very GPU-bound benchmark, even more than 3DMark05 and 06. So, if you want to score well in this benchmark, you need to figure out how to configure the drivers (yes, more than just setting to 'performance') and be able to figure out the limitations of your graphics card: core, shader and memory clock frequencies are influenced by temperature, voltage and by each other. Finding the right balance is, in other words, the key to success. Aquamark3 might have been the most interesting benchmark, although the cpu frequency of many teams was a bit too low in the end. Aquamark3 is mostly CPU/memory bound, which means that even with a bad videocard, you can still score quite high. Also, with a bad cpu, you can overclock the memory subsystem and still score quite good.

On the following page, we'll go more in detail on the specific benchmarks, what problems the overclockers ran into and how they solved it.