| ||
| by MAC | October 2, 2008 | ||
| Competition I: Battle for the Best P45 OC Competition I: Battle for the Best P45 OC
The first round consisted of memory bandwidth benchmarking with the immensely popular Lavalys Everest program. Teams were given 25 minutes to post the highest possible result, and given the handpicked Corsair DDR3 modules (Micron D9GTS ICs) , we were all expecting some impressive numbers.   Colin Brix (Technical Marketing Manager) and Daphnee Kao, the two incredibly energetic hosts for the event, announced the official start of the competition, 25 minutes and counting...   Team Taiwan was one of the first to upload a result, although it was a pretty weak 9,146MB/s. They quickly and consistently improved up to 9,480...10,800...10,421 but they peaked at 10,500 and were inactive for the last 10 minutes, having apparently experienced some 'technical difficulties'.   Here we see Lok from Team Hong Kong, which totally dominated Round 1. Not only did they annihilate everyone right off the bat with a comparatively huge 11,220MB/s, but they posted a new higher score almost every 3-4 minutes. For the curious among you, their processor was running at 5.8Ghz(!) while their DDR3 modules were set to an impressive DDR3-2050 7-7-7 at around 2.1V. Because of these high system clocks, they were definite front-runners for Round 2: SuperPI 1M.   Both Swedish teams had some pretty serious difficulties during the Everest round. In fact, it took over 20 minutes before either team posted a result. This was partly due to the fact that Team Sweden 2 (Elmor & SF3D) actually swapped out their processor, and only had their system back up 2-3 minutes before the end of Round 1. However, as you will see in Round 2, their replacement chip was a gem.   Having reclaimed the top spot from the impressive Team Turkey, Lok could sit back and wait for the seconds to run down. Here are the end results for Round 1:  Here is the full list of uploaded Everest results: Click for full size… 
Round 2 was focused on SuperPI 1M, which is easily the most popular and competitive benchmark in the enthusiast realm. Latency and memory speeds are important in SuperPI, but processing power is King, so whoever could clock their Intel Core 2 Extreme QX9650 the highest had a good chance of winning.     Early in Round 2, Team Russia (DeDal & Neoforce) posted two strong results (8.329 & 8.281) in quick succession, and it looked like they had the momentum to win the round. On the completely other end of the 'Success Scale', Team Australia was still encountering severe issues with their system: erratic motherboard behaviour, USB issues (hence the ghetto beige PS/2 keyboard), and they even managed to completely corrupt the Windows installation. To add insult to injury, their processor/motherboard combo could occassionally boot into the bios at 5.2Ghz, but would often barely post at 4.9Ghz. Therefore, they wisely chose to get a new QX9650 and EP45T-Extreme.   With Pt1t from Team France (Boblemagnifique & Pt1t) behind the desk (or anywhere else for that matter) anything is possible, and the Belgian/French duo managed to surpass Team Russia with 18 minutes left in the round.   Sometime in the middle of Round 2, Team USA 1 (Ross & Maxi) finally had a working system, and even uploaded a few competitive results (8.360-8.375). The reason they missed all of the first round and much of the second was due to an issue that affected many teams. Naturally, everyone removed the stock Radeon HD 4870 cooler to install their own LN2 pots, and obviously unplugged the fan(s) as well. Well apparently, the ATI drivers or Catalyst Control Center (CCC) or VGA BIOS was detecting the lack of fan RPM and rebooting the system. None (or few) of the competitors had previously experienced this issue with Radeon HD 4870's, so it is perfectly understandable that it took so long to narrow down the source of the problem. By the way, you may notice that their table still looked suprisingly clean...a little too clean...well that's because they wisely invaded a neighbouring table.   Team Italy was suffering many of the same problems as Team Australia (they were neighbours ironically enough), and they were only able to post one uncompetitive result (9.062), which was still better than none in Round 1. With 6 minutes left, Team Singapore (NightRaven & T_M) managed to tie Team France's 8.266 mark. Did the rules even allow for a tie in an individual round? Or would it even matter?   Well it didn't matter, because 15 seconds after Singapore's tying result, Indonesia surpassed everyone with 8.250 seconds. Their secret? The prerequisite Red Bull and delicious finger foods.   However, less than two minutes after Indonesia's success, Team Sweden 2 took the lead with 8.172 seconds and held it to the end. The new processor they received at the end of Round 1 could hit a massive 5.75Ghz at ~2.05V in SuperPI, so they really did not even have to bother with specific software tweaks. Both Russia and Singapore battled to the end for second and third place, and finished with 8.188 and 8.187, respectively. In the end, Round 2 finished like this:  Here is the full list of uploaded SuperPI results: Click for full size…  | ||
| |
| Latest Reviews in Articles | |||||||||
|
