A quick question from a reader:

I’m running Vista 32-bit and would like to know which programs are the only ones absolutely required for startup. Thanks.

Turning off unnecessary startup programs is a great way to speed up your computer. If you do it wrong, though, some things might not work right.

The short answer is that no startup programs is required. Sure, Windows requires lots of things to start properly, but everything that’s absolutely mandatory starts as a service. Unless you’re familiar with which services you need, you shouldn’t disable any of them.

You might, however, want some of them to start. Generally, those little startup apps don’t slow your computer down a noticeable amount, and they occasionally make your computer work a bit better. Looking at my own computer, I have left the following programs to start automatically:

• Microsoft IntelliPoint and Microsoft IntelliType Pro. These give my keyboard and mouse extra functionality. If I did disable them, my keyboard and mouse would still work, though.
• RAID Event Monitor. This watches my computer’s hard disk for problems.
• NWiz. This is NVidia’s graphics card software. My computer would run fine without it, but I haven’t bothered to disable it because I’ve never seen it cause a problem.
• Google Update. This keeps my Google software (like Chrome) up-to-date. Wouldn’t it be nice if one piece of software could get all my updates?
• Adobe CS4 Service Manager. This just makes sure Photoshop keeps working.
• Virtual CloneDrive. This lets me mount ISO files like a physical CD or DVD.
• GrooveMonitor Utility. Helps me with Microsoft Groove (something I need for my work).
• Java(TM) Platform SE 6 U15. This running in the background just lets some website start up a bit faster.
• SMax4PNP Application. A tool used by my soundcard.
• Adobe Acrobat and Adobe Reader and Acrobat Manager. This isn’t required, but when I open a PDF, this lets it open a bit faster.
• Microsoft Office OneNote. I use OneNote, and having this startup automatically allows me to start it quicker.
• Ultramon. A tool for managing multiple monitors.

And I’ve disabled some apps:

• Dropbox. A tool for storing files on the Internet.
• AOL Service Libraries. Good ole AOL instant messenger, which I no longer use.
• InstallShield Update Services. This downloads updates to the update software itself. It’s incredibly annoying.
• QuickTime. This would allow QuickTime videos to play a bit faster, but I rarely watch QuickTime videos, so who cares?
• Steam. A service for downloading games online. I used it once, and I don’t need it always running in the background.

Over at the Official Google Webmaster Blog, Matt warns that Google might (someday) factor in page load times when returning search results. In other words, if your website is slow, it might fall off the search rankings.

He pointed me to a site I hadn’t seen before–webpagetest.org. It’s fantastic. I’ve used many different website optimization tools (including Web Page Analyzer, Google Webmaster Tools, and Page Speed), but WebPageTest provides several different sources throughout the world, and different bandwidths, with different browser versions, and a fantastic user interface.

I used it to examine the performance of two of my sites–www.vistaclues.com and www.northrup.org. Northrup.org was pretty good already because I’ve spent a lot of time analyzing it, but I did discover that simultaneous connections with older browsers was limiting the page load time a bit (despite the fact that I already use parallelization), so I juggled the locations of images around a bit.

VistaClues, however, was fairly awful. You can see the test results here. Specifically, compare my first test result with my last (with the same parameters).

Overall, I increased performance by about 250% and decreased bandwidth usage by about 30%:

• I eliminated a few unnecessary objects (including a reference to an object that didn’t exist because the theme developer had left it out).
• I enabled gZip compression. I had incorrectly setup the WP Super Cache plugin for WordPress and disabled gZip.
• I implemented parallelization by adding the sub-domains images1.vistaclues.com through images5.vistaclues.com and referencing different images and objects using these hostnames–even though they point back to www.vistaclues.com. I also activated the Parallelize WordPress plug-in.
• I removed the stupid Gravatar (Avatar) pictures in the comments. Almost nobody used them and they REALLY caused a lot of unnecessary requests.
• I enabled caching for images, CSS files, JavaScript files, and other static files by editing Apache’s .htaccess file. No, I’m not becoming a Linux guy. ;)
• I minified some scripts and plug-ins by activating the WP Minify plug-in.
• I disabled some WordPress plug-ins that I was no longer using to decrease the page generation time.

With all that said, 6.4 seconds still seems like a long time to wait for the first page to load. Other pages do load faster, and the page generally appears within a couple of seconds. So, don’t take the page load time literally, just use it as a point of reference.

Add a comment if you have other performance suggestions or different performance tools.

That’s a screenshot from my favorite free tool for testing hard disk performance–HD Tune (there’s a pro version you can buy, too). That’s from my Seagate Barracuda 7200 RPM 1.5 TB SATA drive backup drive.

The Blue Line–Transfer Rate

First, the blue line is the transfer rate. That’s the speed which the hard disk sends data to the computer when a big file is lined up nicely–defragmented, in other words. Notice that the line starts high (at 126 MB/sec) and drops (to about 57 MB/sec). That’s because the test moves from the outside of the drive to the inside. Disk drives are round, like a record, and so the outside spins faster (in inches/sec) from the inside.

Windows starts storing data on the outside of your disk, and works its way towards the center. That way, your files will be as fast as possible. The fact that the performance drops by 65% from the outside to the inside of the disk helps illustrate that disk performance is WAY better for files stored on the outside of the disk. It also shows one big reason why, as your disk fills up, it also slows down.

The Yellow Dots–Access Time

The yellow dots show random access time. That’s the time it takes for the disk to jump from one part of the drive to the next. In other words, that’s how long it takes to jump from the end of one file to the beginning of the next, or to jump between two segments of a fragmented file. The average access time for this disk is 13.5 ms–but lower would be better.

USB Flash Drive Performance

The previous graph shows typical performance from a hard disk, but USB flash drives behave very differently. They don’t have spinning platters–data is read by sending electrical signals. If you think they’ll be faster, you’re right. If you think they’ll be slower, you’re right.

As you can see, the transfer rate is WAY slower. My Seagate had an average transfer rate of 100 MB/sec, but my USB flash drive was at 16 MB/sec–about 6X slower. The access time is 0.8 ms, though–about 17X faster. So, USB flash drives are faster at random access and slower at sequential reads.

ReadyBoost uses this to improve the performance of Windows by caching files from the hard disk to a flash drive. If it would be faster to read it from the flash drive (like, if it’s a small segment of data that would need to be randomly accessed), it reads it from the flash drive. Otherwise, it takes advantage of the hard disks high sequential read performance.

Many new mobile computers ship with flash drives instead of conventional disk drives. In some ways, performance is better, and in other ways, performance is worse. The transfer rates of flash drives will increase over time, however, and Windows 7 includes some major improvements that speed up flash drives.

How to Get the Best Performance

I do a great deal of photo and video editing with my Canon EOS 5D Mark II–that’s 21 megapixel pictures and 1080P video–and a really fast disk makes things much more efficient. When I bought a new computer, I set it up with a very fast disk subsystem. Not the fastest in the world, mind, you, but a good bang for the buck. I chose two EXPENSIVE 15,000 RPM disks and put them in a RAID 0 array, which reads and writes to both disks at the same time. Here’s the performance I got:

Overall, that’s about twice as fast as my 1.5 TB drive (though I might have expected better…)

On Laptop Performance

So far, I’ve been discussing the disk performance of my desktop computer (a Dell Precision T3400). The main reason I bought a desktop was to get better disk performance–laptops tend to be really bad, owing to the use of small disks. Remember my lecture about how disks perform best at the outside edge of the platter? Well, smaller disks have smaller outside edges, so performance drops. Laptop disks also tend to be optimized to reduce power usage, so they spin at a slower RPM. Here’s the disk that shipped with my Dell D820:

Abysmal, right? The average transfer rate is 3X slower than the cheap Seagate drive I began this discussion with. All hope is not lost, however. It’s not too hard to upgrade a laptop hard disk–just backup to an external disk, swap drives, and then do restore to the new disk. I upgraded to a newer, faster, 500GB drive and got this performance:

The new disk doubled the average transfer rate (though the access time actually dropped a bit). It sped up the computer noticeably. So, if our laptop seems slow, upgrading the disk might be the right choice.

Got a fast or slow disk? Need some advice about hard disk performance? Tell me about it in the comments.