Windows 8 - Audio Performance Testing
Windows 8 has arrived and the team at Rain UK have been testing it out and comparing it to Windows 7.
We've spent the last few weeks investigating the performance of Windows 8 in comparison to Windows 7 for music and audio production. We used three different tests across three different systems running Windows 8 and Windows 7 and here are our results.
Now, please note that all our testing has been on brand spanking new awesome Rain computers with fresh installs of Windows 8 and that’s the experience we’re talking about – if you’re upgrading or installing it on an ancient laptop then your experience might be very different.
Below is a video showing how we tested the systems and what the results were, plus Robin shares his thoughts on what's good and bad about Windows 8 generally. Or read on for the test details.
Opinion on the internet about Windows 8 couldn’t be any broader – everything from "WOW!" To “it’s a catastrophic disaster” and if you’ve been around a while like I have you’ll know that this happens every time there’s a major change to your operating environment. And on the surface there’s been some pretty big changes. The obvious one is the Metro styled Start screen where you are faced with a fluid wall of living tiles that move and beep and tell you all sorts of interesting things about the weather and who was the last person to poke you. The Start screen has also replaced the Start menu which first appeared on the bottom left of our screens with Windows 95 – this seems to be the biggest cause for concern for many people as it does require quite a shift from an accepted way of working. Doesn’t mean it’s a bad thing – just different.
First let me introduce the test systems and explain a bit about what we’ve done. We took three of our systems. First a top of the range Livebook Pro, i7 Quad core processing and all that jazz. Then we tried out our best selling Nimbus Z desktop, an Ivy Bridge based overclocked power house. Finally there’s the Livebook SE which is a special edition of our Livebook LE – it’s smaller, thinner, cooler and the test model is running a entry level dual core i3 processor. All of them are installed with 8GB of RAM and an SSD system drive that comes as standard on all our systems.
We used three tests which attempt to provide some useful, measurable data that’s consistent and has some relevance to music production. They are all based in Cubase using its inbuilt effects and VST Instruments. The first test involves sample based instrument called HalionSonic - multiple instances are loaded and sustained notes are run through each one until the system crackles. A measurement is taken of how many instances and the total polyphony being played without glitching, or if the maximum is reached then we take the CPU percentage as our figure. The second test is similar but uses the virtual synth Spector. The third test is the celebrated Dawbench “Cubendo” test from dawbench.com – it essentially loads up the CPU intensive multi-band compressor as an insert on each of the 40 tracks of audio. Keep adding them until it crackles and then backup to get stable playback and measure the total number loaded. It doesn’t sound very complicated, which it isn’t, but it can be terribly time consuming.
Of course you can spread this across different bits of host software and use different plug-ins and methodologies but ultimately we’re just trying to find something that measures system performance in a relevant way and that can be compared from OS to OS on the same machine – so this will do!
For consistency we used the same audio interface through-out – the increasingly impressive Steinberg UR28M.
Before we get into the tests there's been lots of talk about the new Start Screen and how it gets in the way and why can't we just go straight to the desktop. I can sympathise with those complaints but actually the Start Screen makes things very easy. It's easy to see what's installed, easy to find and run the program you want to run and as the system boots faster than Windows 7 then you're already ahead of the game. Another common complaint is about it being focused on touch, which is true, however, using a mouse wasn't a problem at all - it was easy, natural so not something worth getting worked up about in my opinion.
Onto the tests:These three graphs show the comparative results on each computer of Windows 7 and Windows 8. Have a look and I'll go into more detail below.
DawbenchThe more compressors you can insert before the audio crackles the better the score. Both the Livebook Pro and the Nimbus Z favoured Windows 8, only the Livebook SE found Windows 7 to be the better performer. Maybe this is down to Core count and the speed of the system, the Livebook SE being far slower than the other two with an i3 Dual Core processor as opposed to the Quad Core i7 offerings in the other two. But we can't say that Windows 8 performs less well on a lower powered machine because the other two tests suggest the opposite. Perhaps it's a freak result - difficult to say - we did retry it a number of times with the same result. What it does show is that the performance is very similar in either direction and we shouldn't put too much weight on a single result.
SpectorWith the Livebook Pro and Nimbus they both maxed out the 64 instrument slots in Cubase and so the results were based upon the Cubase ASIO usage meter - lower is better. The Livebook SE couldn't manage that many instances and so a measurement was taken of how many Spectors it could manage at the point of audio glitching (higher is better). In the Livebooks Windows 8 wins out but with the Nimbus Windows 7 takes the lead by a hairs breadth. We'll come onto the reason for this after we look at the Halion results.
HalionSonicThis paints a very similar story to Spector. The Livebooks both show Windows 8 to be better while the Nimbus awards it to Windows 7. So what's up with the Nimbus? The Livebooks were both installed and setup with a "Local Account" login - this is the most basic login available in Windows 8 and doesn't connect to the Microsoft Store or other Live Tile apps for which you need a Microsoft Windows Live ID. The Nimbus on the other hand was fully integrated into the Store, Facebook, Twitter and other online goodies. It had sucked in photos from all sorts of places, connected to my music collection and media server, and synced up to email and messages. In other words the Nimbus was setup as fully functioning Windows 8 world of everything all at once. That is probably the reason why the performance was not quite as good. Let's be absolutely clear though - the difference is tiny - the impact of all that background stuff going on is very very small, which is good news for those of us who use their computers for everything.
Before we get too excited let’s make sure we interpret these results properly. Microsoft had made claims about the reduction in background processes and so the potential for improved performance was already known. So it’s of no great surprise that Windows 8 wins out in these tests. However, it’s only by the narrowest of margins. the Nimbus results show that's it actually very easy to erode the performance back to Windows 7 levels - but that's still very very good. The best news is that it works and works really well and so performance is not a reason to shy away from Windows 8.
The great news is that I've yet to see a piece of software or hardware fail to work or install. Albeit I've only tried gear that already runs on Windows 7. The UR28M was using Windows 7 drivers and worked brilliantly. I've installed a ton of software now including Reason, Adobe CS6, Native Instruments Komplete, Omnisphere, Office 2010, Arturia V Collection, Pro Tools, all without a glitch or complaint. I imagine that Windows 8 specific installers will be forthcoming as currently each install makes a bit of a mess of your Start Screen by scattering short-cuts all over it that used to be hidden away under the start menu - there should be an option for selecting which items appear on the Start page - maybe that'll happen in time. Hardware could be a bigger cause for concern but so far anything with Windows 7 drivers seems to work fine.
Windows 8 seems to work really well for me and while I'm still getting used to certain aspects of it, I'm totally enjoying the experience. Windows 7 feels a bit old and clunky in comparison - funny how that happens. We'll be offering it with all our systems as an option - Windows 7 as well - no plans to retire that just yet. Quite how we'll set it up for customers is another question and we'll experiment with that over the coming months and would welcome any feedback on what you'd like to see. Tweaks and settings are being discovered all the time, but all our usual magic is doing its thing in providing a good, stable and flexible system.