However, even with this modest Vcore increase we could significantly improve our overclocking results. The CPU worked stably at up to 280MHz FSB frequencies.

So, the Pentium D 920 test sample overclocked by about 40% above the nominal and ran stably at 3.92MHz. Of course, this is not the world’s record, but it is quite a good result, anyway. At least this value is a way higher than the nominal clock frequency of the top Pentium D processors, which is very pleasing already.
After our experiments with the Pentium D 920, it was the turn of the Pentium D 930 fellow. Without any Vcore changes this processor could work stably at 258MHz FSB, i.e. at 3.87GHz clock speed. However, this is just a preliminary result. As we have expected, the increase in the processor Vcore pushed the overclocking results even higher. Since the nominal core voltage of this processor is slightly higher than that of the 920 model we have just overclocked, we took our chances in raising the Vcore a little bit more, up to 1.475V. After that our Pentium D 930 CPU could boast absolutely stable operation at up to 267MHz DSB frequency.

As a result, the maximum frequency we managed to reach during Our Pentium D 930 overclocking experiment was 4.0GHz. In absolute values, it is more than during Pentium D 920 overclocking, but in relative values it makes only 33%.
So, it looks like the Presler processors boast much higher overclocking potential than their counterparts on the Smithfield core, and can easily reach frequencies up to 4.0GHz thus outperforming even the top-of-the-line Pentium D CPUs working at their nominal speed. However, will this overclocking be enough for Pentium D 920 and Pentium D 930 to outperform the top dual-core AMD CPUs? This is the question we will try to answer later in this article.
As for the working conditions for the overclocked Pentium D processors, we have to stress that they are quite acceptable. The CPU temperature when they were topped with Zalman CNPS9500 LED cooler was close to 70o C, and the system proved stable in all widely spread benchmarks that were completed without thermal throttling activated.



