I know this is going to sound strange, but I just benchmarked compiling a project using Adobe Flex Builder under both XP running on Parallels and natively under Bootcamp, and I'm getting better times under Parallels! What I really don't understand about this is that when running natively, Windows has access to all 2GB of memory, has complete access to both processors. Under parallels, I've given it a max of 832 MB, and the Mac activity monitor is showing not even full processor utilization for 1 processor! Any ideas how this could be happening? Thanks! Mark
You can think about Parallels virtualization engine as additional caching layer for all devices (cpu,hdd) - that is why in some conditions it can really be faster with it than without. Virtualized does not always mean slower...
Let me just chime in and say that I'm so tickled at how fast this all works under OSX on a MacIntel machine that its been refreshing. I'd just about given up hope that VPC would ever announce anything so I'm really looking forward to whats in store for the future.
Thanks, Tacit_One, that helps explain it. I thought that that might explain it, especially for things like disk access, which compiling probably stresses. I've also noticed that start up time for the OS is much faster. Thanks and it's looking really good!
I 've got the same feeling using a medical soft : 4 mn to analyse 24 hours ECG under VPC 7 on a imac G5; 40 sec with Boot camp (imacIntel) and 15 sec with parallels ! That's sounded me very strange and I thought something was wrong... Anyway i'm gonna check that point. Otherwise bootcamp supports USB 2...
The current issue of Maximum PC disspells that XP is faster in Parallels Workstation. I may have misread it, but I thought it was testing it on a Mac with Bootcamp and a Mac with PWS.
I think it depends on what you're doing. Operation in general seems slightly more sluggish. However, startup times are much much faster and my compile times are faster. It could be that if you're I/O bound, it can speed things up.
Memory and cache operations are definately an order faster but everything else is slower - about 20-25% slower. See this thread for some benchmarks I ran to verify the performance gains in RC2. http://forum.parallels.com/thread2451.html
