A Close Look at Windows 8 Performance: Winners and Losers (Part 1)

Don’t let the shiny new Metro interface of Windows 8 fool you—the new operating system (OS) has been improved not just with a spanking new outfit but also in the performance department. For instance, Microsoft’s future OS needs to work with both high-end PCs as well as small-factor devices like ARM and low-powered Intel tablets, which is why its overall footprint needed to be reduced drastically.

As VentureBeat noted, “Windows 8′s secret feature [is] resurrecting old PCs.” Microsoft promised that Windows 8 would run equally well or better on low-powered machines than Windows 7. This is a bold statement, but the real question is, did the company deliver? In this blog post, we’ll explore that with early benchmarks,  showing if and how much Windows 8′s performance compares to Windows 7′s on the very same machines.

Windows 8 has been optimized from the ground up to work great with low-powered mobile devices. But how fast is it really? Our lab tests will give you the answers.

Our Test Beds: A Broad Range
To get comparable results, we put Windows 8 through a variety of tests on several very differently equipped systems. We used:

Two Desktop PCs: We installed the Windows 8 Consumer Preview on both a 2007-era desktop PC (Core 2 Duo, 2.66 GHz) and a recent Alienware gaming rig (Core i7 930, 8 GB RAM).

Two Laptops/Tablets: We also tested a low-powered Core i7 1.8 GHz, 4 GB RAM 13″ laptop, which includes a 256 GB SSD, as well as an Asus EP121 tablet sporting a Core i5 processor, 4 GB of RAM and a 64 GB SSD.

Netbook: We used a very common Samsung NC10 netbook running a 1.6 GHz Intel Atom and 2 GB of RAM.

We didn’t install any tools on those machines except for the applications that we frequently use for benchmarking. As usual, we performed all tests three times to get a good idea of the results. As usual, we put both OSs through the typical rounds of analyzing boot time, raw processing power and application launch speed. However, since this is a Consumer Preview, we just want to give a quick impression rather than perform a full-blown review—we’ll save that for later!

#1 – Boot Time Performance

Windows 8 is widely known for booting up extremely quickly. I used a stopwatch to determine how long it took until the desktop under Windows 7 and the new Metro user interface under Windows 8 was visible. The results will blow you away.

On all machines, Windows 8 was able to surpass its predecessor—right out of the gate, the new OS booted much more quickly. Once the final version hits, we’ll go through our usual paces of running Windows Performance Analyzer to determine if and how much background activity during boot was reduced.

#2 – Processes on a Fresh System

Last September, Microsoft boasted about less processes and reduced RAM consumption. Obviously, the less things that are running in the background, the better performance and longer battery life will be, as idle times are much higher. Right after we clean installed both Windows 7 and Windows 8, we compared the usage.

What we found was not necessarily a big achievement, but an achievement nonetheless. We saw a decrease of five processes as well as a drop in RAM usage by about 100 MB. In your day-to-day routine, this won’t  be of particular help. What is helpful, however, are the drastically reduced idle activities. On Windows 8, Microsoft managed to reduce overall background activity noticeably—both on the software and on the driver side. By default, the bare OS’s processes and services rarely cause any noticeable spikes.

#3 – Application Start-up Times

Application start-up times give us a good indication of the OS’s ability to pre-load data and quickly manage smaller file chunks. To test this, we “trained” Windows’s own SuperFetch feature by loading Outlook 2010 every morning at exactly 9 a.m. and keeping it running on both the Windows 7 SP1 and the Windows 8 OSs all day long. After a couple of days, Windows had adapted. Then, we were able to use the trusty AppTimer, which automatically launches applications and measures start-up times down to the millisecond.

Now, how long does it take? We only performed this test on the Core i7 1.8 GHz laptop.

There was  a difference in so called “cold” and “warm” startups – cold refers to the first startup of the application right after boot, and warm refers to subsequent launches that are all completely loaded from memory and, thus, faster. The differences were hardly noticeable, but there was a clear trend towards Windows 8 loading applications a tad slower; however, this could be attributed to hard disk controller drivers.

#4 – Office Performance

To test day-to-day performance, we used the dependable PCMark 7 benchmark tool that automatically performs tasks such as website rendering, virus scans, photo manipulation and video editing. It should easily prove how well Windows 8 can handle both CPU and hard disk heavy tasks.

We found mixed results. While Windows 8 achieved quite a leap in performance on the faster machines, the lower-end devices, especially the netbook, actually suffered a bit. This ran noticeably slower during the tests, which was also noticeable during day-to-day usage.

#5 – Gaming Performance

The S.T.A.L.K.E.R. benchmark is a very well established test program designed to measure frames per second in DirectX games. As usual, we performed all tests three times and averaged the results.

The gaming front seems to profit from Windows 8′s reduced memory usage and  apparent  optimization to DirectX. But, those results should be taken with a grain of salt: Windows 8 installed some more recent beta drivers that were not available for Windows 7 SP1.To compensate for the effect, we downloaded the most recent beta drivers for  nVIDIA and Intel onboard graphics. Overall, the netbook suffered a small amount again, while the faster machines gained quite a bit of performance.

#6 – Overall Verdict: Good Improvements, Some Losses

The Windows 8 Consumer Preview runs well on higher-end devices, but we were surprised to see that it lost some performance on lower-end machines—the very exact devices that it’s supposed to run well on. However, I’ll wait for the release candidate to make a final judgment. We may have an odd combination of unfinished drivers and code and bugs that may affect benchmarks (which is exactly why I only used a handful of my tests to get a general feel for new OS’s performance).

Overall, Windows 8 is going in the right direction. However, judging bare operating systems can only reveal so much.  We’ll find out how it really works when we’re able to test third-party applications and  the Metro-style apps. Do these negatively impact performance over time? We’ll find out in part two of our Windows 8 performance series.