MacTech has just published an extensive benchmark comparison of the latest versions of Parallels Desktop and VMware Fusion. The article outlines in great detail the performance of each, not just regarding running popular programs and boot times, but also on file I/O, optimal host configurations, Retina display support, and a number of other important performance considerations for the virtualization packages.
While bigger may be better in some things, faster arguably takes the cake when it comes to computing. Be it booting a system or sheer number crunching, if you can do more of it at once or do it in less time then you likely have the edge.
Benchmarks have always been the standard for comparing different set-ups, but often when new products hit the market, their benchmarks include only a basic overview but often don't detail every aspect of the system's capabilities.
There have been several comparisons between VMware and Parallels, but in the latest one, MacTech tries to cover as much ground as possible. It delves into multiple aspects of these virtualization packages, looking at everything including system launch times and application launch times, memory footprints and host CPU usage, file I/O performance, graphics capabilities, and optimal configurations for managing guest installations of Windows 7, Windows 8, and OS X. In addition, MacTech looks at how each VM solution implements and handles Retina display support for Apple's laptop systems.
The performance tests were done on two MacBook Pro systems, one iMac, and a Mac Pro that were configured with common configurations of Core i5 and i7 processors (the Mac Pro used a Quad-Core Xeon), and RAM configurations of between 4 and 16GB.
MacTech has done an excellent job at evaluating the performance of these two products, so which package is the winner?
Ultimately, it appears Parallels Desktop comes out ahead, since it wins in most categories. For 3D graphics, Parallels bests VMware in 62 percent of its tests by a factor of 10 percent or more, and is faster overall slightly more than 71 percent of the time. Taking into account several graphics performance test where VMWare did come out ahead, Parallels topped VMware by 10 percent or more in 56 percent of the tests.
In addition to raw numbers, MacTech found some interesting details about optimal VM setups. For instance, allocating more RAM than is required to run the needed tasks in the VM may in fact slow it down a touch. MacTech found that, for Windows 7 and 8, allocating 1GB of RAM was the optimal amount for best performance. The same goes for the number of virtualized CPUs, where MacTech recommends you assign multiple CPUs only if needed (my approach has been to give the VM at least two but no more than half the number of CPU cores your system supports).
Overall, MacTech's findings are a very thorough effort that gives a well-rounded look at the performance of the two most popular virtualization packages in OS X. If you are interested how each stacks up in a particular area, then check out the MacTech article and dig into the results. Additionally, if you are interested in developing and implementing virtualization, MacTech has a number of upcoming events that focus in part on the performance and integration of vitalization solutions.
While this analysis does delve into the packages' performance, remember that it is not a full review. Even though performance is important, it is not the only factor that you should consider when choosing a virtualization solution. For instance, performance aside, how one package sets up the VM and the way it integrates into OS X may simply fit your workflow better than the other.
Both VMware and Parallels offer free trials for their software, so you can download both and do a head-to-head comparison to see which works best for you.