Articles: CPU

Bookmark and Share

Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 ]


Another aspect of the new dual-core AMD processors that we couldn’t leave out today is overclocking, of course. The thing is that transition to new 45nm cores stimulated new wave of enthusiasts’ interest to AMD solutions. New Phenom II CPUs started overclocking very well, especially compared with their predecessors. And even though we know that Deneb based processors can go maximum as high as 3.7-3.8GHz with air-cooling methods, we tried to experiment with the Phenom II X2 550 and Athlon II X2 250 processors that we received in our lab. We used a relatively old but reliable Scythe Mugen cooler for our overclocking experiments.

The first one to take the stand was Phenom II X2 550. As we have already mentioned, it belongs to the Black Edition family, so it can be overclocked by simply changing its clock frequency multiplier, which is unlocked by the manufacturer.

Frankly speaking, we didn’t expect this CPU to do too much better than the previously reviewed Phenom II X3 and Phenom II X4 solutions. However, we were very surprised with the obtained results. The thing is that with the Vcore increased by only 0.15V above the nominal (up to 1.475V), we could get it to work stably at 3.98GHz. We checked the system stability in this mode under heavy workload created by LinX Linpack code.

This is a truly unexpected result that contradicts everything we have seen before during our overclocking experiments with AMD processors on Deneb and Heka cores. However, our joy didn’t last long, unfortunately. Our tests showed that even though this system passed heavy processor tests, it was unstable in 3D applications including games.

Therefore, we had to lower the frequency quite a bit. Phenom II X2 550 could stay indisputably stable only at 3.8GHz.

As you can see from the screenshot, we increased the CPU Vcore to 1.475V. The second processor voltage, CPU NB, remained unchanged, because increasing it didn’t allow us to set the frequency of the integrated North Bridge anywhere above 2.0GHz. At 2.2GHz our test CPU started having some issues with the memory subsystem. As a result, Phenom II X2 550 processor ended up acting just like its elder fellows, despite the extremely promising start. No doubt that overclocking results were predetermined by the use of the same semiconductor die as in Phenom II X3 and Phenom II X4 CPUs.

Athlon II X2 250, however, is a different story. This processor is built on a unique semiconductor die that isn’t used in any other CPU models yet. And since it is smaller in size and has lower TDP, we can expect it to do much better during overclocking.

However, we didn’t get any significantly different results when we increased the core voltage by 0.175V (to 1.5V). The CPU worked stably at 3.9GHz, which is as good as it gets.

Note that since Athlon II X2 250 doesn’t belong to the Black Edition family, we had to increase the clock generator frequency to 260MHz to achieve these results. This is where the absence of any L3 cache came in very handy: Athlon II X2 250 didn’t mind speeding up its integrated North Bridge and we didn’t have to lower the corresponding multiplier. As a result, its frequency increased to 2.6GHz, which was no big deal for it as long as we slightly increased its voltage (by 0.1V).

So, Athlon II X2 250 proved to be a little more overclocking-friendly than its elder brother, Phenom II X2 550, even though it is not one of those Black Edition units. Of course, it is way too early to make any final conclusions about the new processors’ overclocking potential judging only by the results obtained on the very first samples. But at this point it looks like Regor core boasts slightly better frequency potential than Deneb and its modifications called Heka and Callisto.

Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 ]


Comments currently: 2
Discussion started: 06/02/09 05:28:46 AM
Latest comment: 07/21/09 10:54:34 AM

View comments

Add your Comment