Robert McLaws: Windows Edition

Blogging about Windows since before Vista became a bad word

Windows Vista SP1 Performance - The Real First Look

Earlier this week, Microsoft officially announced that they have finished Windows Vista SP1 and released it to manufacturing (RTM). Unfortunately, it's not going to be released to the web (RTW) for another 30 days, save for the 15,000 or so testers that are lucky enough to have the RTM bits (AKA SP1 RC Refresh 2).

Just before all my server BS went down, I ran a bunch of benchmark tests on my local systems running Refresh 2. Since the code is finished, and some of the bits are out in the wild, I thought it would be a good time to talk performance in SP1, with some real-world results. And I have very good news for you.

To get started, I uninstalled the SP1 RC Refresh 1 from all of my Vista machines (I currently have 4 at my disposal). After fully-patching the machines, I then downloaded PCMark Vantage, which is a benchmarking program specifically designed for Windows Vista, and ran it against all 4 machines. This gave me a baseline of what the performance would look like on my machines as of today. I then installed the SP1 RC Refresh 2 across them all (which took 1-2 hours per box) and re-ran the benchmarks (which also took 2 hours per box). That gave me a basis for comparison.

Now, before you start complaining about my methods, let me talk about a couple of important points:

  • Comparison Points: I'd LOVE to have had the time to do re-images of all the machines, and benchmark them against the unpatched RTM build of Windows Vista. Besides the fact that it wasn't practical, the Windows Vista experience has improved quite a bit since then. People aren't going to compare in their heads to the way Vista was a year ago, they're going to compare their computer performance the day before SP1 to the day after. Fully patching the system before doing my baseline benchmark allows for a more real-world comparison.
  • PCMark Vantage: Because this tool was designed to benchmark Windows Vista, I believe it is a LOT more valid than the benchmark that got all the discussion before. It leverages specific features in Windows Vista (Windows Photo Gallery, Windows Contacts, Windows Mail, Windows Media Player) to provide performance metrics that are more meaningful to end users. The only thing that it didn't test was network copying, but Microsoft already has published benchmarks in that area.

Now, onto the machines. Here are the details on the two that I am able to share benchmark data on:

Dell XPS 410 (x64)

Samsung Q1 Ultra (x86)

The other two machines I can't discuss just yet, but as soon as I am able, I will publish the data. Without going into detail, I can say that the results were comparable, however. But the PCMark links that I've posted show full hardware specs, as well as the individual test results, so feel free to peruse and compare at your leisure, and draw your own conclusions.

So there you have it folks. While some people are willing to spell doom and gloom for SP1, I think most people will see a decent performance boost in real-world scenarios. For machines like the Q1 Ultra, the boost is much-needed enhancement to the hardware. For beefier machines, you might even get more bang for your buck. The 15% performance boost on the XPS 410 is much better than the 10% boost that XPSP3 has over XPSP2, and not even close to the numbers that Devil Mountain came up with for SP1.

But remember, take ALL benchmark claims with a grain of salt, because as always, YMMV.