Currently I have Parallels setup to use my boot camp partition with Win XP Pro and am quite happy with the performance on my MacBook Pro. What are people’s experience with just using Parallels native virtual Windows partitioning and not BootCamp as the partition?
I noticed that Parallels with native XP is much faster than BootCamp with its own XP. After several tests, I booted BootCamp.
It is hard to understand how XP running inside a Parallels VM which has an extra layer of software in the middle can be much faster than XP running in a Boot Camp partition. How did you make the comparison? Which areas are faster? Network access? Graphic? Disk access? Curious mind would like to know more.
the only thing i noticed is boot time for BC part is slower because it has to unmount the disk and initialize parallels tools. not a big deal to me
I am inclined to "boot" bootcamp and just have Parallels run XP so I can save on disk space. I don't need to play video games, just some Windows only apps specific to the work I do.
Sorry, but this is silly. Win XP on Bootcamp alone has access to your whole machine and all the ram etc. It is considerably faster in any operation than any Parallels VM.
That's seems to be an evidence, isn't it? That's wrong. Sorry. Some behaviours of XP running in the Parallels VM on Mac OS X run faster than native (bootcamp). And there is nothing very surprising here. You will of course find things slower in the VM, too. But that is a misconception that all of XP and what you do with it will run faster when native (bootcamp) than in the VM. For one thing remember that the Intel chips on which this runs are designed to allow that (hardware extensions / instructions supporting it). Mac OS X I/O and caching can also have a positive effect. In the other way, if the application is capable of significantly increasing its performance when there are 2 or more processors available, it will be faster in native boot because it will indeed have access to all the cores of the chip. Inside the VM, with current Parallels Desktop version, the XP will see a single cpu. But Parallels and Mac OS X (the host) itself will make good use of the other to 'support' the VM.
What could make a difference is if OSX and/or Parallels do a better job of using free RAM as disc cache than native Windows - which sounds very, very feasible. This could be a huge advantage when using a virtual hard drive - a bit like having a hard drive with 512MB-1GB of cache and a very intelligent controller. If you're not really pounding the CPU and RAM - e.g. just using basic office stuff - then this might give you significantly faster load times. As you start to run CPU/RAM intensive stuff then rebooting into native windows and using all your RAM and processor cores is going to give an end-of-argument advantage. If not, then it sounds like its time to check that your motherboard and HD drivers are all properly installed. If what this thread is actually comparing is Parallels running from the BootCamp partition vs. Parallels running from virtual HD then I'd be totally unsurprised if the latter was faster. Does Parallels+BC even support DMA/UDMA disc access? Another no-fair comparison would be a new, clean install of Windows in a VM versus an 8-month-old BootCamp install - Windows is well known for gunging up over time and really needs a 6-monthly spring clean (that's why we're using Macs, folks!).
I use Parallels with a VM image, do not and have never used a boot camp partition with Parallels. Windows runs very well and I have no issues, and I do work it pretty heavily from time to time. One thing I have noticed is a performance hit when in coherence mode vs. window mode. I run some large complex spreadsheets and have discovered that I just have to be in window mode to work with them - coherence just makes them too slow. Right now I've only got 512Mb memory allocated to the VM, I may raise that.
Feeling and guessing something is fast doesn't make it fast. To satisfy my own curiosity I did a little benchmark during my lunch hour. I ran it on a 20" iMac with core duo and 2G RAM. The XP VM has 1.2G RAM and 16M video RAM allocated, and it points to the Boot Camp partition. Opening Photoshop Element 2.0 took 7.2 sec in XP Boot Camp and 9.8 sec in XP VM. Resizing a 3,000x3,000 pixel image to 12,000x12,000 in Photoshop Element took 13.8 sec in Boot Camp and 3 min 32 sec in XP VM. I think what happens in this case is that the VM runs out of memory. The progress bar moves quickly at the beginning and then it slows down and the HD icon starts flashing. I will try to do more test after work.
Maybe some test should be run--my experience suggests what I have said above. If, however, what the original poster meant was Win XP in a Parallels VM vs Win XP in a Parallels VM using Bootcamp, then he is certainly correct. Using the Bootcamp-based VM in Parallels is considerably slower than Win in a Parallels VM.
