Windows 8 Stress Test: How 3rd Party Programs Ruin the new OS’s Performance

In last week’s blog post, we learned how Windows 8 managed to beat Windows 7 in the performance department in almost every discipline. Combine that with the new user interface with its tiles and apps, and you might be thinking that it’s an entirely new operating system (OS). And while a part of it is definitely new, it is still yet another Windows version. Microsoft makes this very clear by simply adding another 0.1 in the kernel version number; Windows Vista was 6.0, and Windows 7 was 6.1). Despite all of its improvements, it’s the same architecture with all of its inherent weaknesses.

Windows 8 gets slower with almost each and every installed program.

It’s an age-old and proven fact: the more programs you install on your PC, the slower it will eventually become. Plus, a lot of programs don’t just install on your system and sit quietly in the “C:\Programs” folder. Many of them add icons to the tray area of your taskbar.

They automatically launch applications after boot, and they add even more background services and scheduled tasks that are active in the background.

On today’s systems, a handful of these programs are certainly not capable of bogging down your system, but add a couple dozen of these programs—and you’re quickly looking at a massive performance, reliability and usability problem. Unfortunately, you can’t really avoid it. First of all, your new PC already comes with at least two dozen built-in applications that offer “value-add”,and you also need your set of programs. So how does Windows 8 gets slower over time and with each application you install?

Test scenario: 150 programs. 2 months.

Windows 8 has “RTM’d” (Released To Manufacturing), and although we couldn’t really put it through a long-term test, we tried to simulate heavy usage as much as we could. We installed 150 programs (basically a mixture of programs we clean up with TuneUp Disk Cleaner, a list of our personal favorites and some of the applications we installed in our 200 programs experiment two years ago).

It took us about four hours to install all of these programs, and I noticed a performance and reliability drop after adding the first 50. The test system I installed this on is no slouch either. I used a 2009 Core 2 Duo 3 GHz, 4 GB RAM and a 256 GB SSD drive. It’s not a high-performance Core i7 rig, but also no age-old XP machine; its fast SSD and 3 GHz processor make it midrange.

Clean vs. 150 programs vs. optimized with TuneUp Utilities 2013

So, what was the overall impact on performance? Did all of the additional files, registry entries, background processes and startup items noticeably slow down the computer? We compared the performance of we experienced on day 1 when the programs were just installed with the performance eight weeks later after heavy usage. We were also curious to see if and how much basic optimization helped. For that, we used our own software including TuneUp Program Deactivator, Turbo Mode as well as Startup Manager.

#1 – Resource Usage: Processes and RAM Utilization

Before we jump into actual performance tests, we want to talk a bit about resource usage. First, we looked at how the numbers of active processes increased after we installed all of these programs when compared to a clean Windows 8 install. Every active process (idle resource usage, CPU spikes, RAM utilization, handles, etc.) adds its own weight to the system, and this doesn’t just lower performance but also puts a toll on battery life. For that, we had a look at Task-Manager and compared the number of processes exactly 10 minutes after boot. These are the results:

A clean install of Windows 8 showed us exactly 61 processes. Keep in mind that this number may vary on your machine depending on the driver situation and whether you’ve used Windows Update to download some “optional” components. Once we installed the 150 programs, that value jumped to an average of 131 (this varied between 129 and 133 with each test run). However, once we used all of TuneUp’s optimization tools (especially TuneUp Program Deactivator, which essentially pauses these processes), we saw that number go even below that of a clean install.

Memory utilization is another area where a lot of processes have a heavy impact. While one process that takes up 10 MB of RAM isn’t a problem, 100 processes are when today’s systems have an average of 4 GB of RAM. Keep in mind that Windows and your active applications could easily eat up at least half of your memory; add another 1 or 2 GB taken upfrom background proceses running, and you’re quickly looking at a low memory situation, which leads to Windows 8 swapping memory to disk and, thus, losing performance.

A clean install took up about 1.4 GB of memory (“Commit charge“), which included all of the basic drivers, OS files as well as the entire cache (SuperFetch). Once the 150 programs were installed, usage increased to a staggering 2.6 GB –this barely left 1.4 GB free and available for our day-to-day use. Keep in mind that, in average, a browser with about 10-15 open tabs can often be found consuming about 1 GB of memory. Then, after we optimized the system using TuneUp Program Deactivator, we were able to bring reduce the 2.6 GB down to 1 GB.

#2 – Boot Time (Windows Logo to Start Screen)

Could the 150 programs add to the overall boot time of the blazing fast Windows 8? Let’s find out! We measured the time it took from the second the blue Windows logo appeared until the moment the new start screen was visible. These are the results:

Boot time decreased to about 29 seconds after we installed the programs – that’s a stark contrast to to the quick 20 seconds of the clean install. Even after we turned off all of the programs, performance didn’t quite get back to to their original levels. What we really noticed, however, is that Windows felt extremely unresponsive long after we saw the start screen. That’s why we looked behind the scenes of the boot procedure using the brand-new version of Windows Performance Toolkit (WPERF), which is now part of the Windows Assessment and Deployment Kit (WADK).

According to WPERF, the original boot time of the machine came in at an impressive 61 seconds—this was the entire time it took for Windows to boot to the desktop and load all start-up items and services and perform internal checks. Once we installed the applications, overall boot time decreased to a staggering 170 seconds. That’s thrice as much. It was only once I optimized the PC with TuneUp Utilities 2013 that WPERF returned a great 55 seconds.

#3 – Application Startup Time (Outlook 2013)

As we’ve previously mentioned, launching applications is an excellent benchmark for any OS, as it requires heavy parallel operations and tons of CPU cycles. Let’s see how these were affected after we tortured the poor test bed with the 150 programs installed.

We used the brand-new Office 2013 Preview and ran through several test runs with AppTimer.

Once again, the heavy load and constant CPU thrashing of all the additional third-partystart-up items, scheduled tasks and services reduced the startup time of Outlook 2013 from an “okay” 2.66 seconds to a fairly long 5.9 seconds. Good news though: Once we temporarily turned off all of the programs using TuneUp Program Deactivator, we were able to get startup time back to its normal levels (and even a tad faster, might I add).

#4 – MP3 to AAC conversion (iTunes)

Next up, let’s look at how these 150 programs impacted a truly CPU-heavy task. For that, we converted a 30-minute podcast from an MP3 into the AAC format. The results are here:

Luckily, we only lost  12 seconds despite the heavy load. Still, that’s an increase of about 20%, and if you’re starting to convert DVDs or even Blu-rays, we’re talking about a pretty hefty impact. Thankfully, once we turned off all of the applications and optimized the PC, these values returned to normal.

#5 – PCMark 7

Our all-round benchmark favorite, PCMark 7, tests the performance of actual Windows applications, such as the responsiveness of multiple browser tabs, notepad pages and malware scans. It simultaneously performs dozens of these tests and calculates how fast your machine is when conducting these typical tasks. We wondered what the overall impact on our Windows 8 test bed really was.

Due to its relatively weak GPU and the older-generation CPU, PCMark 7 showed an overall score of 1512 after a clean install. Once we used the machine for about eight weeks and installed the 150 programs, performance actually went down to 1151 points—the benchmark even aborted several times due to excessive background activity. The results also varied quite heavily because, during each test run, some processes were active, while others were idle. It was utter chaos. Fortunately, optimizing the system improved performance. It became more and more clear that even Windows 8 was not capable of putting an end to or even lowering the priority of these (mostly non-essential) tasks.

#6 – Battery Life

These days, to many users, battery life is as important as performance. That’s why it’s crucial to measure the impact on Windows 8′s (usually very good) battery life once we put the Core 2 Duo laptop through our 150 programs marathon. The tests were performed with backlight set to 50%, Wi-Fi enabled and the power mode set to “Balanced”. We played a movie, measured the time it took for the battery to run dry and repeated this three times to get exact results. By default, since the battery has had plenty of charge cycles, the laptop went dark after only three hours and 15 minutes.

Due to the excessive background activity (we could hear the CPU fan rev up several times as usage increased to more than 50%), battery life dropped from 195 minutes to a mere 145. We lost almost an hour due to these programs.

Windows 8 has some pretty nifty power management techniques in place, but they had no cure for the constant CPU, RAM and HDD thrashing that occurred with more than 130 processes in place, all fighting for the PC’s resources. Fortunately, optimizing the system reduced brought back the original values but even increased battery life by a couple of minutes.

#7 – Photo Conversion Performance

Next, we used Image Resizer to convert a batch of 20 photos with the resolution of 3888×2592 to 1920×1080. The combination of CPU operations and hard disk activity should prove how much of an impact the 150 programs had.

Yet again, the constant thrashing reduced performance noticeably – we noticed that, often times, the entire operation just stopped about halfway through, probably due to a lot of other processes fighting for CPU time.

The entire operation resumed after a pause of about three seconds. This happened three out of our usual fivetest runs. Once we turned off all of the resource-hungry programs using TuneUp Program Deactivator, activated Turbo Mode and followed all of TuneUp Utilities 2013′s recommendations, performance went back to normal.

#8 – Gaming – Grand Theft Auto (GTA) IV with iCEnhancer 2.0

Surprised that we’re testing a game from 2008? We’re doing this on purpose—don’t worry. First of all, GTA IV is still a pretty bad console conversion that is sucking up a lot of CPU cycles and GPU power, even when running on today’s systems.

Plus, we added the iCEnhancer 2.0 mod to improve the game’s texture quality and lighting, and makes this look (almost) as good as a modern game. However, since we couldn’t really test performance of this resource hog on a 2009 machine, we performed the GTA IV benchmarks on our gaming rig, which we also tortured with our 150 programs experiment. Let’s look at the results before and after.

Even though we recently installed an overclocked ATI Radeon 7950, the Windows 8 clean install (combined with the latest AMD/ATI drivers) produced only 42 frames per second (FPS), which is smooth but surprisingly low. However, that number decreased to a near-choppy 34 FPS with the 150 programs active. We observed during the test run that in most of the scenes of the default benchmark run, performance was almost on par with the clean install (minus about two FPS). However, the average FPS returned was drastically decreased due to some processes cranking CPU usage up to 100% and—for a few seconds—lowering FPS to below 10. That’s unacceptable for gamers. Luckily, once the system was optimized, performance returned to normal and even gave an extra frame with each of our five test runs.

Verdict: It’s the same old Windows underneath.

Windows 8 might be the first step on a path towards a PC world in which apps barely consume resources and don’t run constantly in the background. Unfortunately, we’re still living in a world in which traditional desktop programs rule the market—and your PC’s performance! What was true in 1995 is still true today with Windows 8: the more programs you install, the slower your system becomes.

If your system is bloated with 50, 100, 200 or even 300 applications, it doesn’t really matter. Windows 8 will not tell you that your system is under heavy usage and will not recommend to uninstall or turn off applications. What we’re left with is a faster system when brand-new,  but it will just get slower and slower over time and with each application that adds services, background tasks, processes and startup items. This is even more noticeable on low-powered tablets or budget-oriented ultrabooks. Our advice is the same as it was two years ago when we first performed such an experiment:

• Before you install any piece of software, think long and hard if you actually need it or plan to use it regularly. If you just need it once, remember to uninstall it.
• Go through the list of all of your installed programs and decide if you still need all of them.
• Disable programs you need using TuneUp Program Deactivator.
• Clean up your system using our tips and our new TuneUp Disk Cleaner and TuneUp Browser Cleaner.
• Be sure that only the most necessary start-up applications load automatically when you turn your PC on

What do you do to keep your shiny new Windows 8 PC, tablet or ultrabook clean and fast from day 1?