Posted on 11/06/2009 9:31:58 PM PST by Swordmaker
Parallels on Wednesday released Parallels Desktop 5 for Mac, a new version of its virtualization software that allows Intel-based Macs to run Windows and Linux alongside Mac OS X.
According to Parallels, the new version of its software is up to 300 percent faster than the previous version, Parallels Desktop 4. The company also hired Crimson Consulting Group to do performance testing on Parallels 5 and said it was 22 percent faster than its nearest virtualization competitor when running Windows 7 64-bit on a MacBook Pro.
While the competitor wasn't named by Parallels, it is widely thought the virtualization market is ruled by Parallels and VMWare's Fusion for Mac.
Beloussov said the company achieved those goals by adding support for gestures, supporting a broader range of 3D features, and optimizing Parallels to handle virtual hardware and drivers more efficiently.
Parallels 5 also adds a new viewing mode called Crystal, which is similar to Coherence, where Windows completely disappears from your desktop. However, Crystal adds the Windows taskbar items on the top of the Mac OS X menu, allowing easy access.
Apple also offers a way to run Windows on an Intel-based Mac. It's called Boot Camp, but it requires the user restart the computer and run Windows separately from the Mac OS. While that may be inconvenient for many users, the native speed of running on the hardware was always seen as a major advantage. Parallels says they can now beat that.
"Parallels is faster than Boot Camp," Beloussov said. "Most slow downs with Windows is not because of computational problems, it's handling drivers. We offer optimized virtualized hardware and drivers, making it faster."
Parallels Desktop 5 for Mac is available for $79.99 or $49.99 for an upgrade.
If you want on or off the Mac Ping List, Freepmail me.
Astonishing if true. I might just switch again if it is— I started with Parallels but tired of their abysmal customer service and switched to VMWare. VMWare’s customer service isn’t much to write home about but Parallels’ was in a class of its own. Still, for this kind of performance increase... I’d switch. I truly would.
Yes.
I am running Windows XP as the virtual machine on my MacBook Pro with OSX. I am assuming if I upgrade that I will still be able to do that. I will upgrade if possible because I have not been happy with the speed and am not willing to dedicate more memory to the Windows side to improve it. Any thoughts? What do you run?
Humor me... You need more memory to run this? It is memory intensive?
Of course. A VM is basically a large application running on the native operating system. Whenever you run an application, the operating system must allocate memory to it. With me so far?
With Parallels you're running an application that is another computer (guest), inside the native one (host). The host must have enough memory that it can assign the application (the guest OS) enough for -it- to run satisfactorily.
I have a MacBook (the little one) with 2GB of RAM. I regularly run a WinXP VM, with 1GB assigned to the XP VM, and 1GB remaining in OS-X. If I had 4GB of RAM, I'd probably give the XP VM a bit more, but XP runs well in 1GB.
Win7 likewise does well in 1GB. Vista, no, but I'm not running Vista.
So in general, figure you need total physical RAM equal to what you want in BOTH operating systems. I'd call that memory intensive, but it's no surprise that it is so.
That's gotta be BS. I have yet to see -any- virtualization scheme that can run a VM (Parallels, VMware, etc.) faster than on native hardware (BootCamp). Even ESX, a totally stripped down mini-Linux with no USB support, etc. for big server use, doesn't make that claim.
Somebody in their marketing department is doing some "creative" testing to come up with this data, assuming there is any. I call BS, except maybe for some special cases they manipulated so they could make this claim.
If in fact Apple's BootCamp drivers are so slow that VM drivers beat them, is speed them up. Yet, the fastest native Windows machines have often been MacBooks in BootCamp, over the past few years. I don't think it's a problem.
Anyway, it's impossible for a virtual machine to run faster than on native hardware, since you're going through more software layers. So this has to be some special case crap argument.
This one's gonna look awful silly when it's debunked...
I upgraded earlier this week. Version 5 is faster than version 4 with XP. I have not compared benchmark speeds tho.
I don't doubt the claim that they sped up their code, and good for them.
I'm only saying that (except for a rigged case) virtualization can't run faster than native code on the hardware (unless the native code is really stupid-slow). If Parallels has figured out how to make software that runs so fast that it actually beats the hardware, somebody will be in for a Nobel in Physics -- they'll have invented the equivalent of a perpetual motion machine!
I've generally used VMware products, but a few co-workers have used Parallels and say it historically has lagged behind VMware in speed and features. I suppose this means Parallels is catching up, which is good.
Thanks! Much appreciated. I have never used VMware or whatever it is called
You're quite welcome. VMware is the current industry leader among perhaps four viable VM vendors (Parallels, Microsoft, Zen). To give you an idea of the flexibility this technology provides, here's a condensed list of what guest OSes I run these days:
I'm not associated with VMware in any way, BTW. Just a happy customer. ;-)
Ok, yeah that sounds about right. Good call.
And to tell you the truth, I'm not concerned with a few percentage points difference in speed. I care more about solid, stable support of peripherals and no crashes/BSODs. Drivers that are optimized to the brink of disaster for speed often are more trouble than they're worth, if they destabilize the system under stress.
Marketers love comparative benchmarks, but I just want the sucker to run without having to worry about it.
I can see how the claim could be true if the virtualization software does a better job of managing the disk cache than the native OS. In such a case, real-world application benchmarks would finish in less elapsed time, although actual CPU time would still be larger, due to the virtualizer having to trap system calls in the VM and emulate them to a certain extent.
That's an interesting point. I gotta think about it...
I assume there are three things going on with VM:
Anyway, I suspect it will all come down to how fast the BootCamp drivers for Win7 turn out to be, once they're released.
With so much power and memory etc it is obvious (as you said) that an operating system is just another application and you then run an application within it
Yeah, in the context of a VM, the guest OS runs as if it's an application on the host OS. And then other applications within the guest, as you said.
Example: When I want to run Microsoft Office for Windows on my Mac, I have these layers going:
Alternatively I can have the Windows desktop be the application that appears on OS-X, and have the Office app appear on the Windows desktop.
So exactly as you said -- the guest OS is just another application, and you run an application within it.
Of course, my example only happens to use a Mac for the host. You can do the same thing on Windows or Linux hosts. It's kind of a kick to run Win95/98 (as a VM) on speedy modern Windows PC hardware. It's unbelievably fast, and it makes for some interesting double-takes from friends.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.