Overclocking and Undervolting
As I said at the beginning , I got a 1.6GHz 400MHz FSB processor with the goal of overclocking it to a 533MHz FSB, thereby increasing CPU speed to 2.13GHz. This was easily accomplished with stock voltages. I then proceeded to get Centrino Hardware Control (probably should be named differently; the program really doesn’t require a Centrino system, only a Pentium-M or Celeron-M CPU) to tweak voltages.
I only managed to lower my 2.13GHz voltage down to 1.324V from 1.340V. Barely anything, but then again I’ll probably be on AC power if I’m running full speed anyways. What I was mostly aiming at was to get my 6X multiplier (800MHz) voltage down substantially from the stock 0.988V. Thankfully, that went much better and I managed to get the voltage to 0.732V while maintaining full stability. I adjusted the inbetween multipliers’ voltages as well, getting them down anywhere from 0.100V to 0.200V lower than stock.
I took a nice thorough look at the cooling setup of this laptop. It doesn’t run hot at all, but being the curious person I am, I decided to take a look anyways. There’s usually room for improvement in these laptops anyways.
The CPU is cooled by your regular old heatpipe with a copper heatsink attached to the end. A fan is placed to suck air from the bottom and exhaust it out the back, passing through the heatsink. This is pretty efficient and since these Pentium Ms aren’t exactly huge heat producers, the setup is fine. Contact is good between the heatsink and the processor.
The only other fan present in this laptop is one integrated into a little heatsink/air duct that cools the chipset and graphics chip combination. The heatsink is hollow and made of aluminum. The fan itself is situated over the chipset and blows air through the duct and over the video chip. First off the lack of any fins on this heatsink doesn’t bode too well for its cooling ability. The second flaw is much more serious however.
The two chips interface with the heatsink through a crappy little pad that feels kind of like soft playdough. It’s got this somewhat plasticy (sp?) feel and I don’t think they’re meant to melt at all to create a better connection. When I took them off, they came off whole and showed no hint that they would actually become anything other than a piece of whatever jammed between the heatsink and the chips. The video RAM doesn’t make contact with the heatsink, but it’s only running at 500MHz stock so it probably doesn’t get too hot anyways. The RAM is rated at 3.3ns which means that it is rated to run at 600MHz, so that could mean there’s some overclocking potential there.
I put on some Arctic Silver in place of the pads but I soon discovered that the GPU wasn’t contacting the heatsink at all. Usually when you remove a heatsink, you’ll see a pattern of thermal paste on the bottom of it. In this case, none of the Arctic Silver touched the heatsink. So I had to sand down the screw-in points so that it would make better contact. Unfortunately, you can only sand them down so much. Since the chipset is also on the heatsink and makes better contact than the graphics chip, you can’t lower the heatsink too much or you’ll crush the northbridge. So I found a happy medium where the northbridge was making excellent contact and the GPU was making fair contact. It’s about the best I could do without getting a copper plate and inserting it between the graphics chip and heatsink. I may eventually do that when I have better resources than I do now.
My tip to you about cooling then? Don’t mess with it. It’s far from perfect but it works and you’ll probably run into more problems if you tinker with it.