Benchmarking Windows XP on Parallels Desktop for Mac

Having made the switch to Mac and wanting to still use Windows occasionally, I've been trying out Parallels Desktop For Mac, running Windows XP with SP2. Trying to decide if I might want to run a dual boot (with Bootcamp) rather than in virtualization, I decided to benchmark Parallels.

After some searching on Google, I decided on the CrystalMark Benchmark suite. I benchmarked XP SP2 on Parallels and on my old Pentium IV (3.2 GHz, 1 GB RAM, SATA 150 hard drives) to compare.

These are the results from XP running virtualized in Parallels Desktop. I used the latest build of Parallels Desktop, build 3188 (March 7, 2007) with the latest version of the Parallels tools.

And these are the results from XP running native on my old Pentium IV.

You may notice that the processor masks are different. In Parallels, it only allowed me 0001, while on my native XP box, it allowed me 0001 and 0011. As 0011 provided the best results on the XP box, I went with that to give it as much opportunity to shine as possible.

In hard drive, GDI graphics, and OpenGL 3D tests, the native box beat the virtualized box. Poor Open GL performance is a known issue for Parallels, and though I thought the hard drive speeds might be a little closer, the Mac's running a virtualized partition in a file, plus doing it on a 5400 RPM laptop drive, versus a native 7200 RPM SATA desktop drive on the native Windows box.

Where it really shined was in computation and memory speed. In these three measurements it blew the native box out of the water, turning in scores an average of 4.67 times higher.

Now some people will ask why I didn't benchmark it against a version running native in Bootcamp. I may well do that one of these days. My main reason for running it this way was to determine whether I still got a speed boost running it virtualized in Parallels versus the machine that was my production system up until December 2006. If I lost speed, then Bootcamp would definitely have to be explored. But at the speeds I got, those things where I feel a need for XP again will run plenty fast.

If you're running XP via Bootcamp on a MacBook Pro 2.33 GHz, run Crystal Mark and send me a screenshot.

UPDATE:

My friend, Gary Rosenzweig, of Developer Dispatch ran CrystalMark on his MacBook Pro under Bootcamp and came up with some interesting results.

While XP under Bootcamp blew Parallels's graphics scores out of the water, it came in significantly lower on the memory and computation scores. It still doubled the Pentium IV 3.2's scores on computation, but memory access was only about 25% greater. There may have been some background process in Gary's Windows install that was eating up processor cycles without him knowing it, so if you're running XP in Bootcamp on a stock MacBook Pro 2.33 GHz, please feel free to try your own benchmark and send it in to me at burgerguy@gmail.com.

This entry was posted on Tuesday, March 13th, 2007 at 1:46 am and is filed under Techno Thoughts.You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.