Time for the second day of benching at Intel. We got up at about nine after benching till 2 am. We (crotale and me) talked to Sampsa on the night before the event and we decided that we should bench together to perhaps see if we can reach the top 3DMark scores of today. Since last night was perhaps one of the most pain-free experiences of benching we’ve ever experienced, I have to say that I’m feeling very optimistic. The hardware was working exceptionally well with literally no complications. We pushed it really,really hard and it just kept responding.
Our Danish fellows ordered this container with 200 l liquid nitrogen
Mine and crotale’s gear
We decided that even though we have 49 hours of benching ahead of us, there was no time to spend on air cooling, so we mounted crotale’s Dragonpot (made by Kingpin) and just poured (very) small amounts of liquid nitrogen to keep it just below zero degrees Celsius. After completing a default run with the Radeon HD 2900 XT cards running at 850/950 MHz we focused on bumping up the frequency of the processor.
Robert is very photogenic
And man did it scale! At default voltage we managed to reach higher frequencies than any quad-core has gone before at this voltage. We were simply amazed. We raised it by 0.3 V and all of sudden we’re talking speeds not possible with a quad-core before. Another bump by 0.2 V and all of a sudden we’re at frequencies capable of 3DMark world record results and close to world record CPU scores. And then we have a another 0.4-0.5 V to go…
All this time we’re using Intel’s latest high-end chipset and ATI’s Radeon HD 2900 XT CrossFire. Sampsa’s 2900 cards were flashed with overclocked BIOSes to get access to a higher voltage (1.3V) and then we just used the software method to clock the cards to about 880-885 and 999 MHz with the stock cooler. At this moment, we have results competing for the top spots at the ORB, with just air cooling on the graphics cards, but today we’re going in with the heavy artillery,
The system was based on the ASUS Maximus Formula (that’s DDR2) and a pair of really nice pair of Kingston HyperX modules running at 3-3-3-3 and 420-445 MHz. The hardware has been working more or less flawlessly, if there were problems it was because we needed more voltage. The processor? You’ll just have to guess.
Our fellow benchers from Denmark and Norway haven’t had the same luck, but after working out most of the bugs from their systems and reinstalling Windows last night, they’re catching up to us. To be frank the perhaps most impressive feature so far is our fellow Swede KnightCharger which managed to do very well with his air-cooled system. He’s using voltages that we didn’t think was quite healthy for a processor of this kind, but he’s reaching record frequencies using air cooling, and considering he’s using an all air-cooled system he’s probably very close to the air 3Dmark 06 world record.
Sampsa heating up the pot for the last time …
