CPU Overclocking

We put together the following open testbed for our experiments:

• abit IP35P mainboard, ver. 1.00, BIOS 1.1;
• CPUs:
  • Intel Core 2 Duo E8400 (3.0GHz, 333MHz FSB, 6MB, Wolfdale, rev. C0);
  • Intel Core 2 Quad Q9300 (2.5GHz, 333MHz FSB, 6MB, Yorkfield, rev. M1);
• 2x1024MB Corsair Dominator TWIN2X2048-9136C5D;
• NVIDIA GeForce 8800 GTS 320MB graphics card;
• Seagate Barracuda 7200.10 HDD, ST3320620AS, 7200RPM, 16MB, SATA 320GB;
• Zalman CNPS9700 LED CPU cooler;
• Antec NeoPower HE 550 PSU (550W).

We used Intel Core 2 Duo E8400 processor during the first part of our test session. This CPU works at 3.0GHz nominal frequency and can overclock over 4GHz. As for the maximum FSB frequency, it proved operational and stable at maximum 540MHz (540x7.5). So, how well will abit IP35P do here?

To find out what our mainboard is capable of we lowered the processor clock frequency multiplier to x6, set the memory frequency to its minimal value using 1:1 divider, increased the voltages. Here are the obtained results:

• At 530MHz FSB the board wouldn’t even boot;
• At 520MHz FSB the board took off but wouldn’t boot the operating system;
• At 510MHz FSB the board started, booted the OS, but failed the Prime95 stability test.

The saddest thing here is that there was no stability even at 500MHz FSB. Looks like the maximum operational FSB frequency for abit IP35P lies somewhere around 490MHz, because it passed all the tests successfully at 483MHz FSB. The processor clock frequency multiplier was set to x8.5, i.e. the CPU was overclocked to its maximum possible frequency of 4.1GHz.

This is a good result for a mainstream mainboard on Intel P35 Express chipset, but not for a super overclocker mainboard like abit IP35P was promised to be. Maybe the board will do better when overclocking a quad-core CPU? Unfortunately, our expectations didn’t come true. Having started to overclock our Intel Core 2 Quad Q9300 processors at 470MHz FSB, we had to go down to 440MHz, but still failed to achieve stability…

This is where we have to recall the story of our Intel Core 2 Quad Q9300 test CPU. We got it not so long ago and quickly overclocked on abit IP35 Pro mainboard. The result was truly impressive: the CPU conquered 475MHz FSB with the default x7.5 clock multiplier. We couldn’t get it run stably at 480MHz FSB right away so we continued trying on other new mainboards that arrived in our lab. However, none of them could even get close to the result achieved on abit IP35 Pro, not to mention hitting higher speeds. So, the question that seems quite logical in this case is: did we actually reach 475MHz FSB or it was all illusion? And if it was real, then maybe processor went bad with the time, “degraded” as they say these days and cannot demonstrate the same remarkable speed any more?

To dot all i’s we replaced our abit IP35P mainboard with an abit IP35 Pro. We wanted to take a screenshot with available memory dividers anyway, so why not recheck the Core 2 Quad Q9300 overclocking results, just to make sure? We set the FSB to 475MHz and the remaining parameters to the following values:

• CPU Core Voltage – 1.425V;
• DDR2 Voltage – 2.1V;
• CPU VTT 1.2V Voltage – 1.47V;
• MCH 1.25V Voltage – 1.56V;

So what do we see? abit IP35 Pro mainboard once again proved its excellent overclocking potential having repeated the same successful Intel Core 2 Quad Q9300 overclocking.

As you understand, we didn’t set any record by overclocking our processor to 3.56GHz. What amazed us, was the ability of our system to work stably at high FSB speed of 475MHz. This is a definite merit of abit IP35 Pro mainboard.

Having confirmed that our Intel Core 2 Quad Q9300 processor didn’t lose its overclocking potential, we had to admit that we could only blame not very impressive overclocking results on our abit IP35P mainboard. No wonder, actually. If the board couldn’t really overclock dual-core processors that well, why should it be better at overclocking of more complex quad-core CPUs? But this is where another question pops up: where do the rumors of abit IP35P’s unprecedented overclocking potential come from? They could have emerged as a result of incorrect translation from marketing to normal human language.

The answer to this riddle was a sticker on the front of the IP35P mainboard package that suffered on the way to our lab, as we have already said. Actually, you can see this logo not only on the box front but also on the back and even sides, in the press-release and on the company web-site. abit pays so much attention to it that they even put DDR2 1066 OC support in bold font in the list of the mainboard technical specifications. Yes, all these great sounding phrases like "Extremely OverSpeed" and "OverSpeed Edition" refer to the support of those few additional dividers available on abit IP35P that allow the memory to work at 1066MHz frequency.

We have been waiting for these dividers for a long time, this is great! However, other mainboards out there have been using these dividers for a long time now. Moreover, some mainboards like the recently reviewed Gigabyte GA-EP35-DS4, for example, set the memory at 1066MHz right away without any agitation. So, in this case we are not talking about any break-through or significant achievement: abit IP35P simply caught up with the competitors’ mainboards at this point.

It is pretty ironic that these dividers have nothing to do with CPU overclocking. Yes, if your DDR2 memory modules can work at 1066MHz frequency, then these dividers will allow you to get them to work at this speed. However, you can only do it if your CPU runs at nominal frequency. Dividers set the memory frequency too high, so it will rise even more during overclocking, so you will not be able to actually use them in this case.

So, looks like many newsmarkers spread the marketing word that new abit IP35P mainboard boasts excellent overclocking potential without actually checking it out. According to our test results, the mainboard’s overclocking potential is pretty average, even a little lower than that.