Intel Core 2 Extreme QX9650 - first look

Posted at: 3:49am 29th October 2007 by James Gorbold

Meet the world's first 45nm Penryn CPU ...

Next up was CineBench R10; a ray tracing benchmark based on the professional 3D modelling and animation package Cinema 4D. As in our Media Benchmarks, the QX9650 proved 6 per cent faster than the QX6850. Just like ducks up lining up in a row, the QX9650 also proved six per cent faster than the QX6850 in SuperPi. This is pretty impressive given that none of the test applications support SSE4, so this performance difference is solely due to the architectural tweaks and enlarged Level 2 cache.

However, the Penryn architecture is more than just a couple of performance tweaks; it’s also the first X86 compatible CPU to be made on a 45nm manufacturing process. To see what sort of a difference this makes we loaded up all four cores using two instances of the small FFT test in Orthos, while running Coretemp to measure the CPU temperature.

Using the same Arctic Cooler Freezer 7 Pro HSF and a freshly applied layer of TIM we noted an enormous temperature difference between the two CPUs; the QX6850 running at a toasty 69°C while the QX9650 was a much cooler 56°C. We also measured the power draw of the PC at the mains with both CPUs running flat out; the PC with the QX6850 drew 280W while PC with the QX9650 drew 232W. These are very impressive power figures, even if you’re not interested in saving the environment, as it means you won’t need as powerful a PSU to power a QX9650 as its 65nm predecessor.

This being Custom PC, we also knew it would be a crime against humanity not to overclock the QX9650. From our earlier review of the QX6850 we already knew that this CPU would max out at around 3.76GHz with standard air-cooling.

However, given how much cooler the QX9650 ran than the QX6850, we decided to up the ante somewhat and aim for 4GHz. As the QX9650 didn’t even blink an eyelid running at 4GHz (we simply upped its vcore to 1.525V and the multiplier from nine to 12) we decided to push even further ahead.

By dropping the multiplier back down to nine, but increasing the FSB from 333MHz to 466MHz and the vcore to 1.65V, the QX9650 was happy to run at 4.19GHz, but no higher. At these settings the video encoding score jumped from 1,671 all the way up to 2,357 (a 41 per cent increase) while SuperPi was completed in 36 per cent less time and CineBench R10 ran 32 per cent faster.


CONCLUSION

If the Penryn architecture only yielded an average performance increase of 6 per cent over the Core architecture it would still be a good step forward for Intel. However, the Penryn architecture has a lot to offer than increased performance. This is because the 45nm manufacturing process also brings a number of huge improvements to the table.

The 45nm QX9650 runs cooler, so won’t require as powerful or noisy cooler as the 65nm QX6850, and also consumes a lot less power. This means that using our Media Benchmarks as the metric, the QX9650’s performance per watt is 28 per cent higher than the QX6850 – great news for environmentalists and those who don’t like paying huge electricity bills. Also, thanks to the improved energy efficiency of the 45nm manufacturing process, the 45nm QX9650 is a lot more overclockable than the 65nm QX6850.