@davepermen - You're right, my info was out of date. I had done earlier tests with a pre-release Windows 8 build, and what I said in my post was accurate for that release.
Anyway, I've retested with the RTM build, and here are my current findings. Win 8 is indeed better, but only slightly
so- and the lower process count is still a trick.
I did a head-to-head comparison between 64-bit Windows 7 Professional SP-1 and 64-bit Windows 8 Pro (RTM) on a Dell Latitude E6430. I did a fresh install of each operating system and only installed the following on both PCs:
- All Windows Update patches
- Dell Touchpad drivers
- Intel HD Graphics Driver
- Microsoft Visual Studio 2010 Remote Debugger
I did not make any other customizations to the systems, so it’s a comparison of the default out-of-the-box Windows setup on equivalent Windows SKUs, with the same number of 3rd-party processes/services running.
Here’s the raw data:
- The smaller number of total processes on Windows 8 is due to an administrative change; they bundled more services into “host processes” which each manage multiple services. So Windows 8 lists nine fewer processes than Windows 7- but also shows nine more services!
- Windows 8 uses only 4.4% less RAM than Windows 7. I believe that all the people reporting that Win 8 uses 14-15% less are comparing a Windows 7 setup with tons of 3rd-party cruft installed to a pristine Win 8 install that hasn’t gathered any barnacles yet. Win 8 simply isn’t that big an improvement in terms of memory savings.
- I measured “power-on to desktop” because it was easy to start the stop watch at the same time I pressed the power button. This notebook takes around 13 seconds before it actually begins kicking off the OS install, so you can shave that much off of each score. Windows 8 boot time is 3 seconds faster than Win 7. Almost 8.5% faster than Win 7 if you shave off the 13 seconds for the hardware boot.
- I also monitored CPU usage in Process Explorer while the system was idle on both OSes. Windows 7 varied from 0%-0.77% CPU usage during this period, while Windows 8 regularly went from 0.00%-1.54%. Still not bad, but twice as busy as Win 7 when it was active.
There's also the matter of audio interface latency and DAW performance. Regarding audio interface latency, I did round-trip latency tests on four different interfaces (MOTU PCIe-424, MOTU UltraLite-mk3 FW, NI Komplete Audio 6, and Virus TI Desktop interface mode), and the results were virtually identical across operating systems. In terms of actual interface performance on Win 8, however, (how many plugins and notes of polyphony the interface could support before audio broke down), the MOTU interfaces stumbled pretty hard on Win 8 whereas the KA 6 and Virus ran pretty close to the same as on Win 7. I believe the performance differences are down to drivers, and MOTU has yet to adopt Win 8 as a supported environment, so beware. And yeah, that DPC Latency tool simply doesn’t work properly on Win 8 yet- an update is supposedly in the works.
Windows 8 offers very modest resource usage improvements over Windows 7. There are some performance tweaks here and there that aren’t reflected in my observations, and results will vary by application. You should also keep in mind that very few hardware vendors have released Win 8 drivers yet, and I and others have observed wildly varying results on numerous devices that work great on Win 7 and don’t fare so well on Win 8. If you decide to test out Win 8, put it on a separate partition, or at least make a full, working backup of Windows 7 before attempting to upgrade.
Also, while there ARE some performance improvements in Win 8, none of them apply in any way to day-to-day DAW usage. DAW performance will be exactly the same as Win 7 at best, or if your drivers aren't Win 8-certified, it may suffer. Upgrade to Win 8 if you like the new UI, or the non-DAW related features it offers. But if the PC is primarily for making music... I don't see any compelling reason to upgrade at this point.
EDIT: I guess I should also mention that those Cakewalk/SONAR Windows 8 numbers were misleading; they were simply meter readings that said nothing about the DAW's ability to function. Actual DAW benchmarks of SONAR have shown that it performs exactly the same on Win 8 as on Win 7.