Date:
Wednesday, March 31, 2004
Author:
Kyle Bennett
Editor:
Steve Lynch
Google +1

Overclocking the Prescott 2.4A

The Pentium 4 2.4A CPU comes through with shining results in our overclocking tests. No fooling!

continued...

Overclocking the Prescott 2.4A

Test Systems

We used several different test setups for this article. Changes in motherboard, cooling and ram were used. Our experiences across the dynamics were almost identical showing us that our results were very specific to the CPU and not limitations of our equipment.

MSI PT880 Neo-FISR (VIA PT880), ABIT IC7-MAX3, (Intel 875P), ASUS P4P800-E (Intel 865PE) - Pentium 4 2.4A SL7E8, 1GB (2x512MB) Kingston PC3500 HyperX DDR400 & Corsair XMS3200LL (2,2,3,5), ATI 9800XT (ATI Catalyst 3.8 drivers), 40GB Maxtor ATA133 HDD, Windows XP w/SP1 and DX9B.

For our testing we used a 1:1 FSB:Memory ratio.

Stock Overclock

By "stock OC" we mean several different things. Since this is an Intel retail box processor, it comes with a heatsink and fan unit (HSF) specified for the CPU. The cooling units bundled with these Prescott CPUs are far from the "cheaply" made cooling solutions we have seen packaged with Intel CPUs in the past.

Article Image Article Image

This new HSF seen above comes with pressed in copper core that directly contacts the heatspreader, and also has a multi-speed fan attached with built-in temperature monitoring. This way the fan can spin up as the air moving through it becomes hotter. This is the HSF that we used for our “stock OC,” just like it came out of the box.

Also for our stock OC, we did not increase the voltage to the CPU. It was left at default voltage. For our 2.4A Prescott, that is about 1.4 volts. Depending on the mainboard and BIOS being used, the voltage could deviate from this number a bit, but should not do so by more than one tenth or so. Also be aware that you will see some fairly big voltage fluctuations while this CPU is running if you use a tool such as CPU-Z. Intel has informed us that this tool does not measure the voltage correctly, although they never did use the tool themselves in order to verify that statement. (You can read our article here about Prescott overclocking being dangerous to your CPU in which this fluctuation was first mentioned.)

So all in all, our stock OC numbers should represent what just about any enthusiast should be able to accomplish with the right motherboard and the right temperature conditions. Of course be aware that our results do not guarantee that you will be able to reproduce them yourself. But do know this. We talk to many others in our community that have shared similar results and we would not be publishing this information if we felt it was misleading to our readers. With that said, here is our accomplishment under stock conditions.

Article Image

What you see above is a 900MHz stock overclock. Now for the record, despite our efforts, we could not get this particular CPU over the 3.3GHz mark without voltage tweaking. So at this clock rate you are seeing a CPU that is very much on the edge and we cannot call it 100% stable. But in this particular CPU's case, it is hard to call it "not stable" and please let me explain why.

Our normal prerequisite for calling a Pentium 4 CPU stable is that it will run two instances of Prime95 overnight (6 hours or so) while looping the 3DMark2001 benchmark. The overclock shown above would not do that. Usually one instance of Prime95 would error out, but the OS would stay stable not crashing. That said, we also had 3DMark2001 crash to the desktop as well overnight while Prime95 did just fine.

Overall, we could game on the system "till the cows came home" with no issues at all. Under real-world conditions the system showed to be stable enough for everyday gaming usage. Bumping the voltage up to 1.5v did give the system some added stability under our torture testing, but it still was not perfect.