Improving hard disk performance in laptops
Having resolved the problem with the laptops’ hard drives clicking constantly, I faced another annoyance, still related to the hard drives. The laptops were a bit sluggish when they were heavily reading or writing to the hard drives, which was a little below my expectations from the excellent Western Digital Scorpio drives. And they were making a loud “clunk” noise every once in a while which, even if did not have any impact on the system’s performance, was an annoying and fearful sound.
It’s a known fact that hard drives get hot during use. Very hot. And their platters spin fast. Very fast. The temperature and centripetal acceleration in fast drives makes the chemical layers of the platters slightly dilate. Considering the very high density of data tracks and sectors on every square inch, even the slightest dillatation can result in heads not becoming properly alligned with the spinning tracks below. As a result, all hard drives periodically perform a head allignment operation to be able to locate the correct position of any data sector. You can hear this ocasionally, it sounds similar to the noise of the drive at computer start-up. So I assumed the occasional “clunk” noise was caused by self-calibration. But it didn’t seem right to occur as often as a few minutes in some occasions.
So, I rolled up my sleeves and began investigating.
Each device connected to a computer, inside or out, requires a software driver to provide the operating system with the specific instructions for controlling the hardware. Most versions of Windows are shipped with loads of “generic” drivers created by Microsoft. Windows XP is famous for its humongous collection of drivers — almost anything can be plugged in and will work without extra software.
The trick is in the name: “generic” drivers. Although some drivers embedded were provided by hardware manufacturers, many are created by Microsoft and provide minimal or limited functionality, compared to the full-featured drivers and software prepared by the hardware manufacturers. This is exactly why my hard drives were not running properly: Microsoft’s drivers for the laptop motherboards and hard drives were working — I was able to use the computer — but not at its full potential.
So here’s what I did. I knew the laptops had Intel processors, so they must have had Intel chipsets as well. Off to find the drivers for them on Intel’s support website… Searching around for a couple of minutes revealed a tiny utility for identifying chipset model; that’s how I found out that the laptops have Intel 830 chips in them, controlling everything on the motherboards. So I found and downloaded the drivers for this particular chipset. The downloads section also listed Intel’s Application Accelerator, which was created to enhance performance in laptops. I downloaded and installed it as well.
The result is amazing. By spending half an hour finding the proper drivers from the component manufacturer, the system’s start-up time has reduced with almost one half. Applications start much faster, the drives are quieter, they no longer make the “clunk” noises. The Intel Application Accelerator allows me to make choices over performance versus power consumption.
In what I’m concerned, at least Windows XP’s reputation of “all drivers included” no longer stands. It still is recommended to get drivers from hardware manufacturers and use them. For instance, Dell’s supplied drivers for the Synaptics touchpad are really, really old. Windows’ built-in drivers recognize it as a mouse but doesn’t allow scrolling or other smart options. The most recent driver from Synaptics’ website is much, much better. Or, Canon’s drivers for the digital camera allows me to preview photos in camera and download them sorted in separate folders by shooting date, as opposed to Windows’ built-in support for digital cameras which only allows basic download of image files.
November 6th, 2005 at 1:53 am
“The temperature and centripetal acceleration in fast drives makes the chemical layers of the platters slightly dilate.”
Come on, if the platters would be made of pure-grade crap plastics it would be true; but they’re ceramyc or at least (in older models) of a very strong metal alloy; if the platters would expand like the CDs just because of the rotation speed…
In other words, thanks for reminding me to give a try to the new nVidia drivers for IDE adapters; up until now they were improving only the intensity of my headaches, but I never lose hope.
November 6th, 2005 at 8:59 am
Anything, anything dilates to some extent from heat and rotation. It has been discovered that even if the hard platters don’t change significantly, the magnetic layers on their sides do, and they even “flow” slightly on the surface of the platters. That’s why hard drives, unlike floppy disks, have invisible “service tracks” to help the recalibration process. Recalibration is more important in 3.5″ drives, where a molecule of the magnetic layer at the edge of the platter is subject to significantly higher forces (linear velocity, for instance) than 2.5″ platters — the extra inch makes a huge difference. That’s one reason 10,000 rpm or higher drives are only using 2.5″ or 1.8″ platters. But the extreme forces still exist, and a movement of microns makes a difference when data density is a few tens of gigabytes per square inch. It’s not me saying this, it’s the hard drive gurus — argue with them
Drivers have another story; make sure you check which version is recommended or best supports your particular hardware model. For instance, I have a motherboard with VIA KT333 northbridge, therefore I need the VIA Hyperion drivers for it. Well, the manufacturer recommends an older version, but I pushed it successfully to the latest Hyperion in the 4 series, 4.56, with tiny improvements on the IDE side. But when I tried version 5, it trashed my system. Newer versions no longer have improvements for older versions, but they can screw up more. The same happened with my nVidia Detonator/Forceware series for my GeForce 3 TI200 graphics card; anything above 53.03 makes my screen go blank for a second every time the CPU load increases, even by something as unimportant as scrolling a browser window. 53.04 does this, so do 56.64, 56.72 and up to even 77.72. One would imagine they should all perform the same, since support for my old card should have remained unchanged all this time. But the curious problem with later drivers proves otherwise.
November 6th, 2005 at 8:46 pm
Eh… No (more) comment(s) on HDDs, you made me rethink the matter (regarding the magnetic layer’s radial transition).
On the other side now I weep over giving away a perfect Asus motherboard and buying a replacement from Epox, theoretically w/ the same features but a more complete BIOS. BIOS that was never updated, and they don’t even list my model on their support site. Of course, the initial version has some „quirks” which I wasn’t able to overcome. Down the drain gone my respect for Epox.