Better Laptop Performance: What's the Best Upgrade?

Whether you're using a PC or a Mac (whether Intel, Power PC, or even 68K), there are four components that affect the speed of your laptop - and in some cases they also affect battery life inversely proportionate to speed.

Want a faster processor? It will suck your battery dry faster than the same chip at a slower clock speed.

On a desktop computer, these things are usually played out in terms of speed against price, with power being relatively unimportant. For laptops, however, manufacturers must choose components far more carefully.

The four components are hard drive, memory, processor, and system chipset (bus speed). There are other areas where performance or battery life can be optimized, such as your choice of wireless card on PC laptops, but the biggies are the ones I mentioned.

How much difference a given component makes to laptops performance compared to how big a hit your battery life takes is the big struggle in designing the ultimate high-performance, long-running portable. Lets look at each one.

Bus Speed

Bus speed is set by the your computer's chipset, and it's the one thing that buyers have almost no control over except in the used market, where bus speed is chosen by the age of a computer.

For Apple, look at a Lombard G3 PowerBook from 1999 side-by-side with its replacement, the Pismo model of the year 2000. Bus speed increased from 66 MHz to 100 MHz while most other components remained fairly consistent. A top-of-the-line 400 MHz Lombard and a bottom-of-the-line 400 MHz Pismo were fitted with almost identical processors, the Pismo performed considerably better even with the same clock speed on the processor. Any battery life advantage the Lombard may hold was insignificant, meaning that with everything being equal(which it seldom is), bus speed is an area where you can see real performance improvement without much penalty in battery life.

This doesn't always work out. When Intel moved from the Dothan to the Sonoma chipset last year, bus speeds rose from 400 MHz to 533 MHz, but performance gains were almost nonexistent while battery life went way down. I had an IBM ThinkPad T42p at the time, and compared to the T43p with the faster bus speed, the T42p returned almost identical benchmarks in both test programs and frame rates on real games. The T43p, however, ran for 30 minutes less on its 20% higher capacity batteries.

Was it the chipset that sucked more power? I'm really not convinced, but the fact remains that over on the IBM ThinkPad forums there were a lot of complaints about the T43 series, with many users ordering new T42s until the supply ran out a few months ago.

CPU

The next key component is the processor, and again we have a less-than-clear picture of performance versus efficiency. Intel's Core Duo processor is extremely powerful - more powerful than the Pentium M it replaced - and by most accounts thus far Core Duo returns spectacular battery life.

Looking at ThinkPads again, I currently use an X41 with a 1.6 GHz Pentium M of the final Sonoma series. Its replacement, the 1.66 GHz ThinkPad X60s, has the Core Duo. Both computers use the so-called "low voltage" versions of their processors, but the new X60s, despite a slightly higher clock speed and considerably more power, runs significantly longer on its battery.

Stranger still, the X60s has a larger and faster hard drive that should use more power, not to mention a faster bus speed. The Core Duo does run a bit hotter, but it's combination of power and efficiency are far better than its predecessor.

Memory

RAM is a no-brainer. More RAM increases computer performance and at the same time increases battery life. The RAM chip itself requires power to store data, but by preventing the need to swap that data to and from the hard drive, which consumes far more power than RAM does, having more RAM actually saves battery power - to a point.

With Windows XP or Mac OS X and moderate application use, 1-2 GB of RAM provides a fine balance.

Imagine that you could stuff 8 GB of RAM into your portable, however. In that case, you would be supplying power to keep data in those chips, but since you were only using 1-2 GB of RAM anyway, you would no longer be preventing hard drive swaps, and thus your batteries would run down sooner than on a 2 GB machine.

Still, RAM chips use very little power, so more is usually better.

Hard Drives

The final component, and the area where laptop users can give themselves real performance boosts, is the hard drive. Most laptops still come with 4200 rpm hard drives, while Apple has moved their entire line to 5400 rpm (and even 7200 rpm) drives.

Again we have a compromise. The faster a given drive spins, the faster its head can read and write data. This is a real performance boost you will see every time you boot your computer, launch an application, or open and save large files.

Conversely, spinning a platter at 5400 rpm uses more power than spinning it at 4200 rpm. It's a balance, made more complex by caching schemes, the areal density of the individual drive platters, and the controller chips on both the computer and the drive itself.

I like Hitachi TravelStar drives, not because they are better than their Toshiba, Fujitsu, Samsung, or Seagate competition, but because I've owned many over the years and they've always served me reliably, and, most important, quietly. I'm sure the competition is just as good, including a Samsung drive in one of my PowerBooks that gave excellent and silent service.

The key is to compare drives from a single manufacturer in the same capacity for performance and battery life when looking at the effects of rpm, and I've made that comparison with Hitachi drives.

Benchmarks

My daughter has a 4-year-old IBM ThinkPad X22, a small laptop with an 800 MHz Pentium IIIm processor and a 133 MHz system bus. In short, it's no speed demon - but not sluggish either. I took a disk image in Norton Ghost and flashed it onto the three 80 GB hard drives I have: 4200 rpm, 5400 rpm, and 7200 rpm Travelstars that all have the same 8 MB cache.

Booting Windows XP takes 72 seconds with the 4200 rpm drive from pressing the power button until the hourglass goes away on a full Windows desktop. With the 5400 rpm drive that figure drops to 65 second, with another drop to 54 second with the 7200 rpm drive.

Most people won't care about 18 seconds when starting their computer, but consider that booting up a computer is only part disk access, there are also memory and system tests going on as well as dozens of other hardware functions.

A better test is a disk copy. I prepared a folder filled with documents and files of various sizes ranging from 10 KB to 10 MB, with the total size of the folder at 400 MB, and copied it from one folder to another on the same drive. The 7200 rpm drive was a full 2 minutes faster than 4200 rpm drive, with the 5400 rpm drive in the middle, though closer to the 4200 rpm drive.

Battery life was similarly affected, though this is an example where the higher performance drive is a good tradeoff for its battery life penalty, depending on how you use your computer. The ThinkPad X22's battery is fairly large, though a few years old. When new it ran for over four hours, but these days its good for about 2.5 hours in casual use.

I tested runtime with the DVD Lawrence of Arabia, knowing the movie lasts longer than the battery. With the 4200 rpm drive that is normally in this computer, I was able to run the DVD for 1 hour and 44 minutes. With the 5400 rpm drive, it ran for 1 hour and 40 minutes, while with the 7200 rpm drive it ran for 1 hour and 33 minutes. That's only 11 minutes difference between the fastest to the slowest drive - and a good argument for running the faster drive unless every minute of battery life counts.

Even then, you might find yourself productive instead of watching the hourglass or spinning beach ball longer than the battery increase on the slower drive, making a fast drive the best improvement you can make.

More important than rotational speed, however, is areal density. I ran the same tests again with the 20 GB drive that came with the ThinkPad, a 4200 rpm drive with a 2 MB cache. Booting Windows took a full 98 seconds, 26 seconds longer than with the 80 GB 4200 rpm drive and its 8 MB cache. Some of that was cache, but most was the greater aerial density - how densely packed the data is in the drive.

Imagine you're trying to pick up cones on the road and have a machine that just grabs them as you drive by. Rotational speed is how fast you drive by picking up cones, whereas aerial density is how close the cones are to each other. At 10 MPH you will pick up cones twice as quickly if they are 4 meters apart than if they are 8 meters apart, just as you will pick up twice as many cones at 20 MPH as at 10 MPH.

Comparing the 20 GB and 80 GB 4200 rpm drives, the 20 GB drive has a pair of 10 GB platters whereas the 80 GB drive has a pair of 40 GB platters. The platters are the exact same size, only they hold 4x as much data, meaning the "cones" are four times as close together.

The Best Upgrade

So what's my point? Simple: Put as much RAM into your laptop as it or your budget will allow, although there's little reason for most users to go past 2 GB with Windows XP or Mac OS X, and strongly consider upgrading to a modern, fast hard drive.

Even if you have a fast 5400 rpm drive in your machine, if that drive is only a 20 GB or 40 GB model, you'll see a real performance increase moving to an 80 GB or larger drive even if the rotational speed is lower. Grab yourself a large drive with a fast rotational speed, and you'll be rewarded with greatly increased performance.

Modern drives are more power-efficient as well. In my DVD test, the old 20 GB drive was only able to run for an hour and 28 minutes. That's five minutes less than the fast 7200 rpm drive.