Published: November 23, 2008 in articles » technology by Thomas Baekdal
My new MacBook Pro arrived a few days ago, and while I have always had a Mac, my work is centered on Microsoft technologies. So my main computer, for the last 8 years, has been a Dell computer (at home), and an IBM/Lenovo ThinkPad at work... But now I can do everything on the Mac thanks to virtualization, and it is so cool!
Switching to Mac, as the primary place to work, is not easy when you rely on Windows only products. For instance, our main development tool is Visual Studio 2008, and you cannot get that on a Mac, nor can you switch to an alternative. The other problem is that I currently do not want to buy Mac licenses of all the products I got. Adobe CS4 alone costs a gazillion greenbacks. Office is cheaper, but still costs $135.
On my old Mac I had Parallels. It's a virtual machine that allows you to run Windows in a "virtual window", and it was terribly slow. This was mostly because my old Mac only had 1GB of RAM, and it was a Mac Mini.
With my new MacBook Pro I wanted to solve this particular problem. And what I did was to buy a high-speed hard disk and add 4GB of RAM. That should solve the hardware performance issues.
I also turned away from Parallels and instead bought VM Fusion. In all the reviews I have seen on the net, Fusion is not only faster but also more capable than Parallels.
...and it works brilliantly!
The great thing about VM Fusion is that you can run Windows applications in a mode called "Unity". This allows you to run Windows applications like it was a native application on your Mac. Here is a screenshot where I am listening to iTunes (Mac), writing this article (Word 2007/Vista), checking something in Chrome (Vista), and keeping tabs on my RSS feeds in NetNewsWire (Mac).
And apart from the Vista border in the Windows programs, it all looks like the Windows applications was a natural part of Mac OS X.
Here is another screenshot where I am running Photoshop CS3 (Windows version) while finding the images I need in Safari (Mac).
...or what about this one. Managing files in the Mac OS X Finder and Windows Explorer completely simultaneously (it even supports drag and drop between them).
And the last example is the dock, where Windows applications are handled just like any other application.
It is unbelievably cool!
Of course, the big question is... is fast enough? Yes, it is incredibly fast.
I ran the PassMark Performance Test to see just how fast it was. It felt pretty snappy, but I wanted to see some figures.
I tested the following setups:
...and here are the results:
I am completely blown away. Running Windows Vista, in a virtual machine, in Mac OS X, on a MacBook - is faster than running Vista directly on a Lenovo ThinkPad T61P (and "P" is the high-end model).
How can this even be possible? How can Windows be faster on a Mac, running inside a virtual machine? It defies any form of logic. My MacBook is now officially the fastest Windows computer I own... and it is not even running Windows directly.
There are a few downsides to all of this. While Windows applications are overall faster, 3D performance is severely lacking. You cannot play high-end 3D games, or use 3D design applications. It simply cannot translate 3D performance through VM Fusion fast enough. Not that I care about that, that is what Xbox 360's is for.
The same goes for video. Video in a window is not a problem, but video in full screen is. But why use Vista for playing video? Why not just use VLC directly in Mac OS X.
There are few other quirks, mostly related to Unity mode (not in full screen). One is that there is an occasional "stutter" when you start to drag and drop your mouse cursor. And copy/paste rarely works (but it does in fullscreen).
So the top tip of the day is...
If you want a really fast computer for running Windows applications...
get a MacBook Pro.
Writer, Project Manager and Interaction Designer
follow me on Facebook, Twitter, Flickr, or visit my website
Mar, where do you have that information from.
I see no indications that you are right. The macbook feels faster, and performed the test faster than any of my pc laptops.
It is the same story when I compare speeds in Photoshop.
'Once you eliminate the impossible, whatever remains, no matter how improbable, must be the truth.'
I believe it's hard to tell if it's slower in VM without measuring because it is almost as fast.
But the truth is a timer within a guest OS is not accurate.
You might want to read 'Timekeeping' and 'Performance Measurements' in the following document by vwware:
http://www.vmware.com/pdf/vmware_timekeeping.pdf
Writer, Project Manager and Interaction Designer
follow me on Facebook, Twitter, Flickr, or visit my website
Mar, quotes from the document you refer to:
"Performance measurements reported by a guest operating system running inside a virtual machine are meaningful and in some cases just as accurate as when the operating system is running on physical hardware. However, you must interpret these results with care, because the guest operating system does not have the full picture of what is happening on the host."
"Memory usage counts taken by the guest operating system accurately reflect the guest operating system's usage of its virtualized physical memory."
"CPU usage measurements taken by a guest operating system generally give an approximately correct measure of the relative CPU usage of various processes running in the guest operating system, but they do not reflect how loaded the host system is."
Or to put in in another way. Performance measurements test in Windows can unly test the performance for the available ressources made available by VM Fusion. It does not show you the real performance of the hardware (which would be even faster).
I must admit that I still do not see any indications that you are correct in your assumptions.
In any case, I decided to do a visual test. To perform the exact same task on both my Thinkpad and the Macbook.
As you can see the difference is not much (except for the 200 MB file), but the Macbook is repeatedly faster.
And this is really what matters to me. Even if the performance tests are inaccurate, the real world performance still puts the Macbook out on top.
1. The Mac Harddrive must be much faster. Try Cinebench.
2. A Comparison with Bootcamp on the macbook would be more viable
vmware states the virtual machine _will_ be slower,
and that it _can_ deliver a fairly accurate timer they call
'apparent time' but benchmark software has to be aware of that ('Pseudoperformance Counters').
So, as empirical evidence shows that the performance of a guest OS in a VM will be reduced by overhead from the VM, the more likely case of a benchmark not being aware of timer issues must be true.
Writer, Project Manager and Interaction Designer
follow me on Facebook, Twitter, Flickr, or visit my website
The MacBook's hard disk is faster, much faster actually (see the disk mark in the article), and that definitely makes a difference.
And yes, a comparison with BootCamp vs. PC would be interesting. But, I do not have time to install windows all over again...
Although I also use VMWare Fusion, it is worth pointing out that Parallels works exactly the same: you can run your windows applications side by side (Parallels calls this 'Coherence' mode instead of 'Unity'), supports drag-and-drop between both OSs and also shows your running Windows apps on the Dock).
I have been switching between both VM prodcuts and honestly, I still haven't found which one is better. Both are very performant and full of features.
Writer, Project Manager and Interaction Designer
follow me on Facebook, Twitter, Flickr, or visit my website
Sergio, I am aware of that, but thank you for pointing it out. Although my version of parallels is an older one.
I chose VM Fusion based on several reviews I found on. All of them saying that VM Fusion is the fastest one. Parallels does have one advantage over Fusion, and that is that it supports OpenGL (not that I have any such need).
Thomas-
Your point is well taken. There is really almost no reason to buy anything but a Mac these days. I have a good friend who designs for Dell, and several others manufactures. He recently switched from the world of a windows machine and a Mac to just one new MacBook Pro. He runs windows in boot camp to run such tools as Pro E and the rest of his life in OSX. He has been very satisfied with the results and says that he has no regrets and that his MacBook Pro is noticeably faster than any alternative options out there with the same CPU and Ram. I personally will not be buying a Windows only machine for any foreseeable future. Vista and XP are just not worth the driver and virus hassle they inevitably become at some point not to mention that the are about as behind the curve in regard to what consumers want as the big three automakers.
you are right about the 3d programs, but if you stay with mac native 3d app you will be plesently supprised. I use maya, animation master & i'm testing the new mac version of rhino it's pretty slick.
Nice article. I am almost convinced in switching to a mac as well. I have one question though....
I am a Windows user since the start of my career, so I really don't know that much about macs.
On apple's website I read about a software built in to leopard called "Bootcamp".
In what ways are VM Fusion better than this app?
thanks!
Mickel, you may have already found this answer, but here it is anyway. Bootcamp, Parallels, and Fusion all obviously allow you to install Windows on your Mac. The benefit of Fusion (and Parallels) is that it allows you to run them simultaneously - as shown in the article above - while Bootcamp requires you to reboot your machine to switch between OSs.
Something worth noting - and something that you already Fusion users will already know - is that if you take your Windows partition, be it physical or virtual, online, you still want to have Norton or some other anti-virus software installed.
The real determining factor of which to use is your convenience. If you need to use a PC software like Quicken, where you just need it for a couple minutes here and there, use Fusion. If you're running CAD software or something else that's more of a time commitment, Bootcamp is free so use that. Bootcamp also has supported more drivers (usb 2.0 being one) than Parallels, but with the new version I'm sure they thought to update that.
Hi Thomas
What's that disk mark you refer to in comment 6? i can't find it? It might come handy if I decide to buy a new macbook pro.
9 months ago I bought the macbook air and have been very satisfied with it. However it's running Fusion very slow. (I suspect that the limited RAM, the lack of graphics processor and the rather slow harddisk is the reason).
So my current windows usage (from mac) is very limited. I too also have a windows photoshop license. And changing it to a mac license is a non-reversible decision: You have to destroy your windows license and they will convert it. And you cannot buy a mac license as an upgrade from windows. Rather unflexible.
Writer, Project Manager and Interaction Designer
follow me on Facebook, Twitter, Flickr, or visit my website
Hi Jesper,
The disk mark is the long line in the second graph (3rd from the the top). Part of the six "marks" that make up the pass mark score.
About Photoshop, then you might want to consider Photoshop Elements (price: USD 119.00 / DKK 668.00).
I had the same problem. I have a Windows license of CS3 (and I got tired of having to start fusion every time I just needed to tweak a photo. So I tried out the new version of Photoshop Elements, and was really surprised how much you can do in it.
Elements covers about 98% of what I use Photoshop for.
BTW: Another issue. I never shut down my computer, and I have found that after 3-4 days, Fusion starts to get insanely slow. A restart solves it, but I never had this problem with my Dell, or my Mac Mini. My guess is that it the combination of 2 OS's sharing resources.
Hi,
it would be nice to confirm this peformance. for me it takes around 30 mins to boot Vista under fusion 2. any idea what is going wrong?
tx
markus
Baekdal.com is (nearly) everywhere. Check out our social profiles:
FaceBook
Twitter
LinkedIn
YouTube
FriendFeed
Delicious
Nov 23
2008
mar
It's not faster.
The timer information vmware delivers is wrong.
That's all there is to it.