Benchmarking Parallels, Fusion, and VirtualBox vs Boot Camp - The Mac Observer(macobserver.com)
macobserver.com
Benchmarking Parallels, Fusion, and VirtualBox vs Boot Camp - The Mac Observer
http://www.macobserver.com/tmo/article/benchmarking-parallels-fusion-and-virtualbox-against-boot-camp
8 comments
As a MacOS virtualizatiion user, I appreciate the value of these comparisons. But the graph coloring is crazy-making.
First each product has its own rainbow color, then everything is blue, then BootCamp changes to VirtualBox's purple, then Parallels8 takes Parallels7's blue and BootCamp takes Parallels8's green?
Please, please, please use consistent mappings of colors-to-products throughout your graphs.
First each product has its own rainbow color, then everything is blue, then BootCamp changes to VirtualBox's purple, then Parallels8 takes Parallels7's blue and BootCamp takes Parallels8's green?
Please, please, please use consistent mappings of colors-to-products throughout your graphs.
Good point, gojomo. Sorry about that! I used Apple's Numbers app to make the graphs and it doesn't do too well at offering color customization. It's often an all or nothing approach.
I'll make future graphs with Excel.
Thanks for the feedback!
I'll make future graphs with Excel.
Thanks for the feedback!
Hey man don't blame Numbers. Changing those colors is a simple task.
After years of using VM's to virtualize Windows I finally broke down and got a real, physical Windows box.
On my space-constrained system, a VM often ended up feeling more of a parasite than a useful convenience. Initially I was just using a VM for running Office, but I eventually moved onto Visual C++ and running games from Steam. In the end, for my typical use case (developing windows software) performance in all areas paled in comparison to a real system. There was no real advantage to not using a real system.
On my space-constrained system, a VM often ended up feeling more of a parasite than a useful convenience. Initially I was just using a VM for running Office, but I eventually moved onto Visual C++ and running games from Steam. In the end, for my typical use case (developing windows software) performance in all areas paled in comparison to a real system. There was no real advantage to not using a real system.
I feel comfortable running Windows as host and Linux as guest with VMWare Free Player.
With regards to the title: once you stop sugar-coating it, this is just a comparison of virtualization and native execution. Boot Camp does not apply any/many restrictions to the system and is just a thin BIOS wrapper over EFI for bare metal use.
Benchmarking virtualized solutions against native boot is a good numbers show but proves nothing for virtualization customers except, potentially, that their virtualization software of choice is "finally approaching near-native performance."
Benchmarking virtualized solutions against native boot is a good numbers show but proves nothing for virtualization customers except, potentially, that their virtualization software of choice is "finally approaching near-native performance."
I really don't trust benchmarks running inside VMs. Toyed around with that for a bit & concluded that the results are entirely useless.
All of these benchmarks factor in time somehow (mb per _seconds_). Its like time slows down & speeds up at random in a VM. And heaven forbid you run some tests with Intel VT-x on & some off (In some cases I got VTx off = 100x faster - lol).
All of these benchmarks factor in time somehow (mb per _seconds_). Its like time slows down & speeds up at random in a VM. And heaven forbid you run some tests with Intel VT-x on & some off (In some cases I got VTx off = 100x faster - lol).
The only thing that bugs me about Fusion is that it is usually one hell of a buggy mess right after release. They are usually very quick and thorough to fix that though.
Overall, I quite like Fusion and really see no reason to switch. Any comments from people who have used both maybe?
Overall, I quite like Fusion and really see no reason to switch. Any comments from people who have used both maybe?
Weird: clinking the link reliably BSODs my box. Win2k, Opera 12.02 or Firefox 11.0! I remember seeing the same thing last week with the original article, and wrote it off as a fluke. First hard crash I've had in months.
Could it be some plugin?
I doubt it. I tried again under Opera, with plugins, javascript, animated images, cookies, and pop-ups all disabled. Yet another BSOD and reboot.
I also tried grabbing the site with "wget -p" on a linux box, but I didn't end up with anything I could view as an entire page. Too dynamic, I guess. I did go through and open all the image files inside Opera without a BSOD, so I can rule those out at least.
I also tried grabbing the site with "wget -p" on a linux box, but I didn't end up with anything I could view as an entire page. Too dynamic, I guess. I did go through and open all the image files inside Opera without a BSOD, so I can rule those out at least.
I've found Parallels (7) to be unstable, and I didn't appreciate how it installed crap all over the place with no easy uninstall. Fusion may not be quite as fast, but it is a more robust and professional product. My 2c.
why is vb so slow, and does anyone know if it is slow on all hosts? i use it on linux because it's free and (very) easy and reliable. but maybe i should be switching?
[edit: just poking around on the 'net, reading any benchmarks i can find, the differences don't seem to be that great between the two - seems to depend on the particular benchmark with the general view (my impression) being that virtualbox is easier to use but vmware being slightly faster on average.]
[edit: just poking around on the 'net, reading any benchmarks i can find, the differences don't seem to be that great between the two - seems to depend on the particular benchmark with the general view (my impression) being that virtualbox is easier to use but vmware being slightly faster on average.]
I can't answer your question about why it's slow in these tests, but I do own Parallels (7), and have used VirtualBox extensively.
Some background: I use Parallels for local testing of build-scripts that I use to turn up Rails hosting infrastructure. A "full stack" build script turns up some basic compile tools and utilities, Ruby 1.9.3, Apache, Passenger, and MySQL. All of this is built from a base snapshot: a Debian netinst installation with nothing but root and a non-superuser account. No other packages added or configured at install.
My scripts write timestamps at the beginning and end of the process. With "nothing" running on my host MacBook Pro, Parallels and VirtualBox both take within seconds of each other (around 14m45s) to build an app server, so for my usage, Parallels doesn't represent much of an advantage. I use Parallels for my hosts mostly because I'm familiar with it's configuration tools and supporting application config. I wouldn't spent the money on it unless you use one of the operating systems for tasks they demonstrated will run much faster.
Some background: I use Parallels for local testing of build-scripts that I use to turn up Rails hosting infrastructure. A "full stack" build script turns up some basic compile tools and utilities, Ruby 1.9.3, Apache, Passenger, and MySQL. All of this is built from a base snapshot: a Debian netinst installation with nothing but root and a non-superuser account. No other packages added or configured at install.
My scripts write timestamps at the beginning and end of the process. With "nothing" running on my host MacBook Pro, Parallels and VirtualBox both take within seconds of each other (around 14m45s) to build an app server, so for my usage, Parallels doesn't represent much of an advantage. I use Parallels for my hosts mostly because I'm familiar with it's configuration tools and supporting application config. I wouldn't spent the money on it unless you use one of the operating systems for tasks they demonstrated will run much faster.
I noticed that in the "computation" comparison, Vbox was tied with the others. That's what I care about.
So it seems like Vbox isn't as good at I/O and graphics. Like you said, vbox installs easily and works well.
I don't plan on switching even after reading this article.
So it seems like Vbox isn't as good at I/O and graphics. Like you said, vbox installs easily and works well.
I don't plan on switching even after reading this article.
Short answer: Boot camp is the champ (obviously) and Parallels 8 is second. Fusion is third, but not by much, and VirtualBox is a distant fourth ("you get what you pay for").