John Protopapas ran a set of synthetic benchmark tests on Parallels Desktop 7 and VMware Fusion 4 running Windows XP SP3. He ran an older copy of SiSoftware's Sandra benchmarking tool in Windows. Parallels Desktop 7as faster in two out of the three tests, while VMware Fusion 4 edged out Parallels 4 in the third test. It is difficult to extrapolate how this translates into actual usage, but it is roughly consistent with real-world tests on previous versions of Parallels and VMware. In Protopapas' tests results, higher numbers indicate faster performance.

Protopapas also found that assigning the maximum number of cores to the virtual machine produced faster benchmark results. Protopapas' report:

I have both Parallels and Fusion (current versions). I have a MacMini Lion Server with the 4 cores and 8GB RAM. Both benchmark faster with 8 Cores selected rather than a 4 physical core option. Menu meters shows eight cores four of which are virtual cores. The benchmarks using SiSoft Sandra (2001se) under WIN XP PRO SP3 are as follows:

	Parallels 7	Fusion 4
Dhrystone	41631 MIPS	37,287 MIPS
Whetsone	25,098 FPU	18,262 FPU
	45,459 MFLOPS	46,136 MFLOPS

One problem with Fusion - it sees to want to reinstall VMware Tools on each boot.

Parallels seems to have a somewhat quicker response than Fusion. Overall, I prefer the ergonomics of Parallels, although I mostly use full screen or windowed mode as I am used to it from my work environment: Win PCs. I hope that someday Apple allows the desktop version of Snow Leopard to run on a virtual machine. Right now I have to use my old MacMini with Snow Leopard integrated with my Lion server to run my few PowerPC apps via Rosetta through a two port KVM.

Have you tested Parallels 7 and VMware Fusion 4? your results.