Let me remind you that D1 stepping processors feature “multiple VID”, which means that processors with the same S-Spec may have different core voltages. The Vcore is set by Intel on the packaging phase and depends on the quality of each given die. Of course, CPUs with lower voltages have higher overclocking potential. I had no choice, actually, and it happened that our Pentium 4 3.0 had “1.55V max” marking. It is the worst option as this is the highest possible voltage (there are Pentium 4 3.0 with 1.475V Vcore!). But let’s not get in despair.
The testbed we used for overclocking was based on the ASUS P4P800 Deluxe mainboard (the i865PE Springdale chipset). This mainboard (or chipset, if you like) supports 800MHz bus that is necessary for work (and overclocking!) of Pentium 4 3.0GHz. Besides, I had to spend some time finding the memory fast enough to match the CPU, so that they would work synchronously to achieve highest performance. The memory I used was PC3700 from OCZ. We didn’t adjust the cooling system and used the standard cooler. Also the testbed featured an ATI RADEON 9700 PRO based graphics card.
First, I set the FSB to 200MHz and started the system up in the regular mode. It worked! :) So, let’s rock and overclock. The multiplier of the CPU was fixed (what else could you expect from Intel?), so I sat down to raising the FSB frequency.
Step 1. Without raising the voltages, I set 220MHz FSB. The system woke up and worked without any problems. Well, that was nice. Let’ go to the next step…
Step 2. 230MHz FSB. Nothing changed. The system seemed to take no notice about what was going on. I congratulated myself with being rightly optimistic about the D1 core stepping: it did prove great. I achieved 230MHz frequency (x15 = 3.45GHz) without much effort and without even touching the Vcore! Having checked other websites for the overclocking results for earlier processors, I found that there had been few cases when anyone reached such a frequency. Moreover, they succeeded only with the help of extreme cooling or a huge voltage boost… But that’s not the end!
Step 3. Can the beast notch 240MHz FSB? I set the frequency in BIOS and the system didn’t load Windows XP up… But we have our trumps yet. I raised the voltage to 1.725V and Windows XP flew into the screen and worked without a squeak.
Summing up: I got a perfectly stable system that worked at 240MHz FSB, that is, with an effective (Quad Pumped) frequency of 960MHz! A simple calculation shows that the final CPU frequency equaled 3600MHz, which is not only an excellent result, but also a kind of record, considering the use of air-cooling only. And the core voltage was only increased to 1.725V.
So, now we know that the D1 core stepping offers a lot to an overclocker. But as you remember, we undertook our experiments with a top-end model. Of course, Intel usually picks up the best dies for its top-end CPUs, so we assume that the value CPUs won’t show us the same results. Or maybe they will? Let’s not make any more guesses, but just check it out.