Initially, the Belgian team had high hopes for this setup since the highest BCLK frequency on air cooling was 222MHz, which leads to a theoretical 5.1GHz cpu. That is, however, if the CPU does scale with both higher voltage and lower temperatures

Here you see a picture of the initial setup they ran running some SuperPi 8M:



Time to get down and dirty as the Pi competition started. It's mega-cooling time, baby !



Both Pt1t and Massman were very focused during the overclocking session to get things done right. The latter didn't eat nor drink during the competition, only in the final stages of the competition; he needed some fluids to get going again. Talk about dedication!

During the LN2 benching they encountered a weird issue. The CPU lost its ability to boot straight at high BCLK settings when using the extra cold obtained by pouring more LN2 into the container. Some words on this matter by Mr Massman :

In the pretesting stage, we cooled down the cpu to +20°C to test what the cpu was capable of on plain air cooling. The BCLK frequency went up pretty high, to 222MHz to be precise, this using multiplier 12x to eliminate the CPU frequency bottleneck. When we decreased the temperatures to -50°C and below, we noticed that the CPU was not able to perform at the same level we'd hit under air cooling conditions.

Actually, we were not able to run 220BCLK with a vcore set higher than 1.4v, which makes it impossible to run the CPU at high frequencies. After about ten minutes of testing, we found out that there was an issue with the voltage and bclk frequency. To use over 1.4v, we had to decrease the bclk booting frequency below 200MHz and use setfsb to increase the bclk frequency in windows. By using this 'trick', we managed to run 210+ MHz with 1.6v ... no issues whatsoever.

Very quickly, rumors of a golden CPU spread through the room. As most teams were topping at 4.6-4.8GHz, the BT Clocking team was able to get 4.9+ GHz out of their CPU, so calling it a golden CPU was definitely rectified in this situation. Was it the cascade or the processor which was the biggest factor in this good overclock ? We don't know exactly, but we do know that with this extra 100MHz, they set the pace for the other competitors. Some really serious tweaking by the other teams was required to make up for the lack in CPU speed. It was a nice neck-to-neck race between the Belgian team and Awardfabrik. They were getting closer to BT clocking's score, but they already knew they had to settle for 2nd place. Massman and Pt1t were running sub 1 min 37, yet couldn't validate the score as the pc always froze when opening CPU-Z. German Awardfabrik bested the Belgians 1 min 37 several times, but also had validation issues. The Swiss team had big motherboard issues and were required to swap it a few times. This limiting their time to obtain a decent score.

In the end the, final scores for the 8M Pi run, put team.BE on the second spot: