Surprising results from Win10 benchmarks

The column reprinted below was originally published in the August 6, 2015, Windows Secrets newsletter.

Today, it supplements a new column, posted today, here: https://langa.com/?p=2460

(I’m reprinting selected Windows Secrets columns here to help ensure readers can find and access information I’m referencing in new columns; until older Windows Secrets/LangaList columns are moved to their new home at AskWoody.com.)

The original, verbatim, un-updated text follows:

Surprising results from early Win10 benchmarks

By Fred Langa on August 6, 2015 in LangaList Plus

Windows 10 might be the first Windows upgrade ever that doesn’t exact a performance penalty. In fact, based on these preliminary benchmark tests, Win10 actually makes some systems run slightly faster than they did under Win7 or Win8!

Preliminary speed results for Win10 installs

Microsoft has stated repeatedly that — unlike many past Windows upgrades — Win10 is designed not to require major hardware upgrades.

In fact, Win10’s minimum hardware requirements are decidedly modest:

• CPU: 1 GHz or faster
• RAM: 1GB for 32-bit Win10; 2GB for 64-bit
• Hard-disk space: 16GB for 32-bit Win10; 20GB for 64-bit
• Graphics: DirectX 9 with Windows Display Driver Model (WDDM) 1.0
• Display: 800 by 600

Note that these are the requirements to run Win10 — but not necessarily to run it well.

So, to see how Win10 performs on hardware in real life, I upgraded six PCs — a mix of Win7 and Win8.1 systems.

On two of the systems — one Win7, one Win8.1 — I ran formal performance tests using the well-regarded, synthetic benchmarking tools PCMark 7 1.4.0 (site), NovaBench 3.0.4 (site), and SiSoftware Sandra Lite 21.42 (site).

For the other four machines, I used a more subjective test: the overall feel of upgraded systems.

After running the initial “baseline” tests, I upgraded each PC using the Win10 Media Creation tool available on the “Download Windows 10” page. For the two formal test machines, I then ran all three benchmark suites again to measure the effects of the Win10 upgrade.

I also performed clean Win10 installs on those two systems, again using the Win10 Media Creation tool. (To get a clean install, I selected the “keep nothing” option during the Win10 setup. This completely replaces the existing OS with a virgin copy of Win10 — no active files, settings, drivers, etc. are carried over from a previous version.)

Then I again ran the full suite of benchmarks, measuring the effects of a Win10 clean install.

Picking older and newer PCs for real-life tests

In this case, “real-life” doesn’t mean pristine, laboratory-perfect systems. Rather, I picked two machines I use routinely for work.

PC 1 is a modest, high-mileage system that started life running Vista! It was subsequently upgraded to Win7 — and now to Win10. I use it as a live backup, essentially a data lifeboat I can switch to in short order, should my main system fail.

PC 1 specs:

• OS: Win7 SP1 x32
• RAM: 3GB
• Hard drive: 232GB (mechanical)
• CPU: Intel Pentium, 2-core/1GHz

PC 2 is my “daily driver” — in fact, I’m using it to write this article. It’s a fairly average, two-year-old system that originally came with Win8.0 and was subsequently upgraded to 8.1. It’s now running Win10. (I keep solid backups, so I was ready to quickly roll back to Win8.1 if needed.)

PC 2 specs:

• OS: Win8.1 x64
• RAM: 8GB
• Hard drive:476GB (solid-state)
• CPU: Intel Core i7, 3.2 GHz

Both systems have been well maintained using the techniques and tools documented in several Windows Secrets articles, including the July 2 LangaList Plus, “Prepping a Win7 PC for the Win10 upgrade”; the Jan. 16, 2014, Top Story, “Keep a healthy PC: A routine-maintenance guide”; and the Jan. 10, 2013, Top Story, “Let your PC start the new year right!”

Note that these are exactly the same tools and techniques available to all PC users; I used no “secret sauce” or undocumented maintenance techniques to get these systems ready for the upgrade.

The usual caution: Your mileage will vary

Though there’s nothing weird or unusual about my test systems, my prep work, or the upgrade procedures, it must be said that my test results are not indicative of all systems. Your setup could have very different test results.

Let me briefly explain.

Benchmarks can offer a guide to performance changes, but they truly apply only to the machines tested. The more your PC, your maintenance techniques, and your upgrade procedures differ from mine, the more likely you’ll get different results.

Running your own tests, on your own PCs, is the only way to know exactly how Win10 will affect your specific combination of software and hardware.

With that said, here are the benchmark results I got.

Some surprises in the benchmark results

The following figures show the benchmark results for the original baseline Win7 and 8 setups, for those same systems after an upgrade to Win10, and after a clean install of Win10.

Figures 1 and 3 show the raw scores in whatever units each benchmark tool uses. But in all cases, a higher score means better performance. Figures 2 and 4 show the results as percentage changes, where the baseline scores are assigned a value of 100 percent.

Figure 1 shows the raw benchmark scores for the Win7 PC — and they surprised me.

Figure 1: Upgrading my aging Win7 PC to Win10 changed system performance insignificantly.

I expected Win10 to significantly decrease the performance of the older Win7 system. But instead, the upgrade had almost no effect at all. Before and after the upgrade, the results were either the same or showed very small increases or decreases — in the range of 1 percent to 5 percent. Those differences are beneath the level of most human detection. Confirming those results, my Win7 system felt the same after the upgrade.

Figure 2 graphically shows the same results expressed as percentage change, where the original system’s score is set to 100 percent. Despite jumping two full OS levels, the older machine performed essentially the same after being upgraded to Win10 — an impressive result!

Figure 2: This graph clearly shows the minor performance changes after upgrading my older PC to Win10.

The Win8 results were even more surprising. As Figure 3 shows, the Win8 PC gained a smidgen of speed, both upgrading the existing configuration and creating a clean install. However, as with the Win7 system, the changes were small — not anything that would be obvious in real-life use. Still, that’s an achievement for a new OS.

Figure 3: Upgraded to Win10, the Win8.1 PC gained a smidgen of speed in two out of the three benchmark tests.

Figure 4 graphically shows benchmark results for the Win8.1 to Win10 upgrade.

Figure 4. This graph shows that upgrading my newer system from Win8.1 resulted mostly in speed improvements — though, again, the changes were very small.

In 30 years of covering Windows, I have never seen results like these. Previously, it was axiomatic that installing a new Windows version on an older PC always meant a significant and noticeable decrease in performance.

In fact, it was almost a rule of thumb that a new version required a hardware upgrade to maintain good performance.

But Windows 10 appears to break that mold. On the systems I formally benchmarked (above), and on all the other PCs I gauged subjectively, Win10 felt about as fast as Win7 or Win8 — or even a skosh faster.

It is, of course, early days: Win10 is brand new, and we’re just starting to explore all its nooks and crannies.

But so far, at least in terms of raw performance, it’s looking like Win10 could be the best Windows upgrade ever.

Permalink: https://langa.com/?p=2366

COMMENT / QUESTION on THIS ITEM? See the Comment box at bottom of this page!

NEW QUESTION? Ask here!

(Want free notification of new content? Click here!)