PDA

View Full Version : Boot Camp vs Parallels vs VMWare Benchmarks




MacRumors
Dec 19, 2007, 03:58 PM
http://www.macrumors.com/images/macrumorsthreadlogo.gif (http://www.macrumors.com)

MacTech performed (http://www.mactech.com/articles/mactech/Vol.24/24.02/VirtualizationBenchmark/) an exhaustive set of benchmarks comparing Parallels (http://send.onenetworkdirect.net/z/13532/CD90867/), VMWare Fusion (http://www.vmware.com/products/fusion/) and Boot Camp (http://www.apple.com/macosx/features/bootcamp.html) to run Windows on a Mac. To tackle this problem, MacTech undertook a huge benchmarking project starting in September. The goal was to see how Boot Camp, VMware Fusion, and Parallels performed on different levels of Mac hardware, covering both Windows XP and Vista, and comparing that to a baseline PC running Windows.
Doing such an exhaustive comparison resulted in 19 configurations tested with over 2500 tests to be completed. They tested 3 different broad scenarios: one step tests, multi-step tasks between Mac OS X and Windows, and quantitiative benchmarks on a MacBook, MacBook Pro, Mac Pro and a Fujitsu Lifebook A6025.

One Step Tests: In XP, Parallels is 17% faster than VMWare Fusion on XP and 1% faster than Boot Camp. In Vista, VMware Fusion ran 46% slower than Boot Camp, and Parallels ran 44% slower than VMware Fusion.
Multi Step (Cross platform) Tasks: Parallels was 6x faster than VMWare on XP, and 5.2x faster on Vista.

A number of application specific benchmarks were also undertaken using Microsoft Word, Excel, Outlook, Internet Explorer and more. These results are detailed in their article (http://www.mactech.com/articles/mactech/Vol.24/24.02/VirtualizationBenchmark/) along with relevant graphs.

Their final conclusion, however, sums up the results as follows:
... both VMware Fusion and Parallels performed well, and were a good user experience. That said, Parallels was somewhat faster in general than VMware Fusion for XP. If you want the best virtualization performance for Vista, then VMware Fusion is your choice.

Article Link (http://www.macrumors.com/2007/12/19/boot-camp-vs-parallels-vs-vmware-benchmarks/)



Eidorian
Dec 19, 2007, 03:59 PM
Definitely worth knowing these benchmarks.

Good work Parallels.

finchna
Dec 19, 2007, 04:01 PM
i suppose as a comparison of the 3 mac products it makes sense, but i don't understand a benchmark of a 1.86 machine when the Apple machines are substantially faster. calling any of the apple solutions faster than windows seem inappropriate unless the benchmark machine is the same speed--or at least much closer to it. and testing a desktop against a laptop? why not benchmark a desktop win machine?

triskadecaepyon
Dec 19, 2007, 04:05 PM
I originally thought that I had made a mistake when buying parallels as VMware was a lot cheaper. Guess i did make the right decision.

natallica
Dec 19, 2007, 04:06 PM
Very interesting. I was about to purchase VMWare based on this: http://crave.cnet.com/8301-1_105-9760910-1.html

-- N

chillywilly
Dec 19, 2007, 04:06 PM
The article says this was done in September. And from the article it states:

"We tested VMware Fusion 1.0 (51348)"

Since then, VMware Fusion 1.1 has been released and I've noticed a huge speed increase.

I used Parallels at one point (before version 3.0 came out) and it wasn't something I was happy with. I never used version 3.0, so maybe the user experience is better now.

But having put VMware Fusion through my own paces, XP, Fedora 7 and Vista are all very fast on my 2gb MBP 15" CD. Having Windows for the few things I need it for, available at just a few clicks, is nice.

I'd like to see them go through the same benchmarks and tests with the latest version of Fusion. I bet the results will be a lot better when compared to Parallels.

macfan881
Dec 19, 2007, 04:08 PM
the only key thing they left out was gaming i think thats another key thing people will wanna look at too :confused:

Muzzway
Dec 19, 2007, 04:09 PM
http://www.macrumors.com/images/macrumorsthreadlogo.gif (http://www.macrumors.com)
In XP, Parallels is 17% faster than VMWare Fusion on XP and 1% faster than Boot Camp.

Wait... faster than Boot Camp?

trainguy77
Dec 19, 2007, 04:11 PM
Wait... faster than Boot Camp?

Yeah i noticed that...it doesn't make sense.

Lancetx
Dec 19, 2007, 04:12 PM
This is really good to know. Both virtualization products have their strong points, but since I only use XP Pro and I have absolutely no plans of "upgrading" (I use that term very loosely) to Vista, Parallels is definitely the way to go for me then at least anyway.

CHSE
Dec 19, 2007, 04:14 PM
>In XP, Parallels is 17% faster than VMWare Fusion on XP and 1% faster than >Boot Camp. In Vista, VMware Fusion ran 46% slower than Boot Camp, and >Parallels ran 44% slower than VMware Fusion.

Ohmy! can someone do the math for me? Who wrote this? Thanks for making us all think very hard.

coolbreeze
Dec 19, 2007, 04:14 PM
I wanted really bad to like Parallels, but it has a serious bug that prohibited me from purchasing it. I had to unplug my external hard drive every time I wanted to start up Parallels. I tried beta versions, stable, etc. All gave this error and would not boot until I unmounted my drives.

That was a showstopper and drove me to Fusion, which I admit is slower (even the latest version).

Oh well. Easier than having to reboot into BootCamp every time.

raskar
Dec 19, 2007, 04:14 PM
I'd like to see them go through the same benchmarks and tests with the latest version of Fusion. I bet the results will be a lot better when compared to Parallels.

I've been using Fusion since i bought my macbook last month and i am quite pleased with the performance so far. I am sticking to what I have.. no need for a change.

MarlboroLite
Dec 19, 2007, 04:15 PM
All I kept hearing was that VM Fusion was faster than Parallels as well as more stable...:confused:

I am a prospective post-MWSF Mac owner, and I was sure I was getting Fusion..but this casts it into doubt.

Do people recommend starting using Vista after SP1? I don't want to "downgrade" by buying XP which will probably within a year start to really become obsolete and then have to "upgrade" to Vista all over again....

qtx43
Dec 19, 2007, 04:16 PM
So, on average:

Under Vista "...Parallels runs 44% slower than VMware Fusion."
i.e. Parallels time = (1+0.44) * Vista VMware Fusion time
But later in the article "VMware Fusion is 44% faster than Vista under Parallels"
i.e. VMware Fusion time = (1-0.44) * Parallels time
or rearranged: Parallels time = 1.79 * Vista VMware Fusion time

I'm too lazy to look in the actual spreadsheet to find out what the sloppy writer actually meant.

phytonix
Dec 19, 2007, 04:17 PM
Parallels was a rip off from v2 to v3.
v2 is like crap compared to VMWare and awfully slow in my experience.
We did not upgrade to v3 and switched to VMWare.
1.1 is good. I bet VMWare will catch up with the speed, if Parallels is faster now.

ChrisA
Dec 19, 2007, 04:19 PM
So they compared an old Version of VMware to Parallels and found that a few months ago Paralles was faster. I wonder which is faster if you compare the current version of each? I think the new version of Fusion s better. Also VMWare is multi-platform if you need that.

There are other solution they did not test. "wine", I imagine would be the best if it worked for the programs you need to run. And what about QEMU http://fabrice.bellard.free.fr/qemu/about.html read more about Wine at http://www.winehq.org/

chillywilly
Dec 19, 2007, 04:24 PM
I am a prospective post-MWSF Mac owner, and I was sure I was getting Fusion..but this casts it into doubt.

VMware Fusion has a 30 day trial. I spent most of that pushing the limits of what I was doing (mostly MS Access, Visio and a few other Windows programs) and was quite pleased with having that month before making the purchase.

cpatch
Dec 19, 2007, 04:26 PM
Horse puckeys. I own the latest versions of both Parallels and Fusion and can say without a doubt that Fusion is faster and more efficient than Parallels when used with XP. Maybe if you only use Fusion with one core (Parallels is limited to a single core) then it's slower, but with the option to use both cores enabled it's not even close. Plus the amount of CPU Fusion uses when idle is significantly lower than Parallels.

TurboSC
Dec 19, 2007, 04:30 PM
the only key thing they left out was gaming i think thats another key thing people will wanna look at too :confused:

definitely would have helped. I'm about to get a MBP after MacWorld SanFran, and gaming is quite an important thing for me :)

I'll stick with bootcamp hehe.

chillywilly
Dec 19, 2007, 04:32 PM
Form the article:

"VMware Fusion is designed from the ground up to be separated from the host OS. Some have called this "sand boxed". It has the benefit of keeping things very clean, and intuitive ... but most importantly secure. You have to make real efforts for one OS to affect the other."

I understand what they are saying, but reading this from an interactive perspective could be misleading.

Fusion has a feature called Unity, which allows you to run Windows programs alongside the OS X, even placing them in the dock. Once you get the hang of how this works, it's nice to be able to run a single program you need, it sits there right next to Entourage and Safari and allows both to run pretty seamless.

I know Parallels has a similar feature, but having tried both, Fusion still wins out for me.

I understand the functionality they are doing in their testing, showing how certain file formats interact with the various programs under OS X or Windows, but to me, it's seems they are leaving a few stones unturned.

I know these tests and benchmarks are meant to show how the various solutions work, but in this day and age of quickly updated software, results can seem outdated in a matter of months.

Music_Producer
Dec 19, 2007, 04:35 PM
Horse puckeys. I own the latest versions of both Parallels and Fusion and can say without a doubt that Fusion is faster and more efficient than Parallels when used with XP. Maybe if you only use Fusion with one core (Parallels is limited to a single core) then it's slower, but with the option to use both cores enabled it's not even close. Plus the amount of CPU Fusion uses when idle is significantly lower than Parallels.

I agree.. VM Fusion is much, much faster than Parallels. I own both.. have deleted Parallels since I purchased Fusion.

APPLENEWBIE
Dec 19, 2007, 04:35 PM
I own the latest versions of both Parallels and Fusion and can say without a doubt that Fusion is faster and more efficient than Parallels when used with XP.


OOPS.... HIT THE RETURN KEY TO FAST.....

psxndc
Dec 19, 2007, 04:38 PM
What I'd REALLY like to see is a way for my "hardware" not to change when I switch from Parallels to Bootcamp or vice versa. I've already had to call Microsoft once due to "too many activations" and I'm getting really sick of it.

And lets talk about what really matters: which allows you to run WoW the best? Native OS X, native under bootcamp, or *shudder* virtualized?

-p-

AlBDamned
Dec 19, 2007, 04:41 PM
So with updates to both programs (not Boot Camp) are these results useless? Or would they be a marker?

APPLENEWBIE
Dec 19, 2007, 04:41 PM
I own the latest versions of both Parallels and Fusion and can say without a doubt that Fusion is faster and more efficient than Parallels when used with XP.

Me too. I run VM ware and Parallels (using XP) on two identical iMacs, and it sure SEEMS like VM is snappier. Not sure what to make of this. For certain I much prefer VMware's interface and it seems more stable than Parallels.

For what it's worth, I recently got Dragon Naturally Speaking 9 Preferred to work in VMware Fusion /XP (took a little doing...) and it is actually quite useable. I think it is faster and more accurate than it was on the WinPC I tried DNS (v.6) on a few years ago.

GSMiller
Dec 19, 2007, 04:46 PM
I find my combination Boot Camp/Parallels to work quite nicely with XP Pro :)

ezekielrage_99
Dec 19, 2007, 04:51 PM
n00b question....


So which should I get Parrallels or VMWare :confused:

kaltsasa
Dec 19, 2007, 04:54 PM
Parallels had a nasty habit of grinding my system to a halt for minutes at a time for no identifiable reason. Fusion does not suffer from this. My team at work has been using these virtualization solutions since the release of Parallels, we have purchased and deployed several Parallels installations but we've migrated to Fusion after getting a few licences and testing it out for a couple months. The general consensus among the 4 of us using them both is that Fusion generally feels snappier (especially when running multiple VMs but when running a single it still feels snappier), Fusion seems to be more stable, Fusion seems to have better support for USB devices. We've migrated our distribution plan of virtualization to Fusion rather than Parallels. It's also my understanding that VM has a much more solid base with 64 bit support and multi core support. The scuttlebutt is that Fusion is a Cocoa app from the ground up with VMs solid VM tech behind it and that Parallels Desktop is a Carbon app, I can't verify this, just what I've heard.

My vote is for Fusion having used both extensively.

ryanasimov
Dec 19, 2007, 04:58 PM
Then Fusion is the only reliable choice. I used Parallels quite happily until I got a Blackberry and encountered extreme frustration trying to install Desktop Manager; the install hangs before finishing. Since I had zero problems with Parallels prior to that, I hadn't thought to check the forums at Parallels website for conflicts. The official word from Parallels is that Blackberry support is "coming soon". Some people have managed to get Parallels to work with their Blackberry, but connectivity is sporadic and usually requires a lot of plugging and unplugging their Blackberry.
So I tried Fusion and it works flawlessly. I've found it to be faster than Parallels and more stable.

-Alan-
Dec 19, 2007, 05:00 PM
Interesting, I always thought XP running through Parallels was surprisingly fast. It's nice to know I wasn't blinded by my Apple Newbie love affair. :)

chillywilly
Dec 19, 2007, 05:07 PM
I'll echo ryan on connecting up a PDA or phone under Windows.

Fusion is the first virtual app that allows Microsoft Activesync to function without any problems. I started off using MS Virtual PC 7 on my G4 Mac mini and it wasn't pretty when trying to hook up a Treo or Windows Mobile device.

First time installing and hooking up my Treo 750, it worked, synced and hasn't had any issues since.

To be honest, I've not tried Parallels for this.

~Shard~
Dec 19, 2007, 05:08 PM
This is great info to have, as I have been trying to make up my mind on which solution to use. The resultant feedback in this thread is most helpful as well, as it appears to paint a somewhat different picture...

arn
Dec 19, 2007, 05:10 PM
This is great info to have, as I have been trying to make up my mind on which solution to use. The resultant feedback in this thread is most helpful as well, as it appears to paint a somewhat different picture...

Ya, I wonder if 1.1 really made substantial speed increases, or if VMware just "feels" faster.

Regardless, both are available for free trial, so you can make your own comparison. But I found MacTech's methods to be particularly precise so I wouldn't discount their results off hand.

arn

azentropy
Dec 19, 2007, 05:10 PM
I had both Parallels 2.x and VMWare 1.0 and couldn't really tell a difference in Windows XP.

However VMWare was MUCH faster running Linux VM's and provided Linux drivers from the beginning. Also since VMWare does have a free player for both Windows and Linux I could move/clone my VM's between all 3.

VMWare 1.1 feels much faster than 1.0 did... I didn't "upgrade" to Parallels 3.0...

davidwinogad
Dec 19, 2007, 05:13 PM
I've used parallels v2 and 3 and found it so slow that it was unusable.

Boot camp is waaaaaaaaaaaaaaaaaay faster than parallels

thestaton
Dec 19, 2007, 05:14 PM
Horse puckeys. I own the latest versions of both Parallels and Fusion and can say without a doubt that Fusion is faster and more efficient than Parallels when used with XP. Maybe if you only use Fusion with one core (Parallels is limited to a single core) then it's slower, but with the option to use both cores enabled it's not even close. Plus the amount of CPU Fusion uses when idle is significantly lower than Parallels.

Couldn't have said it better myself.

Kashchei
Dec 19, 2007, 05:33 PM
I know CrossOver isn't in the same category as these other programs, but how does it compete with them? I really like the idea of running Windows programs without going through the hassle of installing Windows, even if it means a hit in speed.

chillywilly
Dec 19, 2007, 05:36 PM
I tried Crossover (which uses the Wine technology) and it's limited to a list of programs they have tested with. I ran IE 6 ok, but it wasn't without some limitations.

If you are just looking for a program or two, it's not a bad solution. But needs some tweaks to make it work (at least when I tried it out).

lkrupp
Dec 19, 2007, 05:39 PM
Let's see what these posters are saying when virtual machines and boot camp kill off OS X and Apple becomes a high end Windows OEM. This coming year will make the path clear. Intuit has already dumped Quicken for the Mac. They say they're working on a Mac version for 2008 but those in the know say it will be a browser based, online version. Adobe will be next. Heck Photoshop Elements 6 is out for Windows for goodness sakes. The Mac version is 4. Just wait until the first big developer announces that Mac users will run the new version under bootcamp or virtualization. The rest will soon follow. Apple itself is worried about it and said so in recent guidance releases.:mad:

cogsinister
Dec 19, 2007, 05:40 PM
Vmware 1.1 is great.....runs XP faster than many PC'c ive owned......

I gave up with Parallels.....just wouldent work.......

Westside guy
Dec 19, 2007, 05:45 PM
After reading through that article... all I can say is, I'm not particularly sure the authors know what the heck they're talking about. Some of what they said sure sounded like they were attempting to translate something that'd been explained to them without actually understanding the explanation.

I'd also have liked to see more specific details regarding their methodology.

jt2ga65
Dec 19, 2007, 05:48 PM
I had Parallels until they came out with the latest update, which broke my install, so I had to do a reinstall anyways. With Parallels, it worked just fine with the virtual HD image, but I could not get it to work right with bootcamp while allowing seemless booting in Parallels or native bootcamp.

I downloaded the latest VMWare Fusion, and it is flawless with bootcamp, even to the point of not having to select a hardware profile when booting. That never quite worked out with my Parallels install. I'll probably go ahead and purchase Fusion, and ditch Parallels.

As for games... If you are going to run WoW or Guild Wars, teh option is pretty clear. I've ran Guild Wars in both Parallels and Fusion (You have to enable activeX support), but neither was really enjoyable. This is the primary reason why I wanted to use Boot Camp to begin with, so I just boot native to play, and then will load Fusion when I need Windows for other things, like editing visio drawings, or other things I don't want to or can't do in OSX.

-jt2

thedesigncrew
Dec 19, 2007, 05:51 PM
If you need the F12 key for certain applications in Windows, Parallels won't be able to help you if you are running Leopard. In Tiger it works just fine. Our company just switched to VMware Fusion which takes care of this issue and runs great on Leopard.

chillywilly
Dec 19, 2007, 05:51 PM
Apple itself is worried about it and said so in recent guidance releases.:mad:

They did? Where?

Quite the slew of predictions.

Mac users that run these virtualization tools are not sending the message to sofrtware developers, "Hey, we don't care about the Mac" It's the money and Windows marketshare that drives companies decisions to reduce their development efforts on the Mac.

Mac users are increasing their numbers. Other developers besides Intuit and Adobe are making Mac software.

Personally, I don't think it's as bad as your post makes it out to be. But I do agree that companies need to keep loyal to the Mac platform instead of simply looking at the bottom line and the number of Windows users.

irun5k
Dec 19, 2007, 06:12 PM
Parallels has a malicious feature called SmartSelect. It is apparently "ON" by default, or at least I didn't specifically enable it when I brought some Parallels 2.x VM images into Leopard with a fresh Parallels 3.0 install.

The result was that OS X was now registered three times to open a bunch of windows file types- one time for each of my three VMs. You can turn this feature off, but in my case I was left having to repair OS X after the damage was done. I had to do some manual cleanup including rebuilding my launch database.

Damage done to my VMs is also uncertain since Parallels registered all my OS X file types in the Windows VM images! I can guess my registries are a wreck after this.

This acted a lot like a virus and in my opinion goes too far in attempting to blur the line between OS X and Windows. I view Windows as an uncontrolled beast and I believe Parallels should be a tight sandbox. When I have three versions of OS X Application launcher stubs for garbage like Notepad, Wordpad, and all their nasty Windows relatives, I get upset.

moutian
Dec 19, 2007, 06:12 PM
I first tried pd and it caused several kernel panics on my mac pro. Then I have to switch to fusion, which so far has been extremely stable.

omnivector
Dec 19, 2007, 06:22 PM
This study seems a bit flawed.
1. They didn't use the newest version of fusion
2. They didn't activate fusion's multi-core support
3. They didn't do any gpu benchmarks (all of which would show fusion being faster / actually working)
4. It's impossible for parallels to be faster than bootcamp. At anything. Virtualization just doesn't work that way.

coolbreeze
Dec 19, 2007, 06:22 PM
What I'd REALLY like to see is a way for my "hardware" not to change when I switch from Parallels to Bootcamp or vice versa. I've already had to call Microsoft once due to "too many activations" and I'm getting really sick of it.

-p-

Aint that the truth. At one point I was running BootCamp, Parallels, and Fusion (evaluating) and each time I booted into one or the other, I'd have to re-register Windows and Office 2007 (even Adobe Acrobat at one point) due to "significant hardware changes." I had to call MSFT numerous times and walk through that annoying million digit authorization code nonsense. At one point I just went to a MSFT operator and hammered her with questions as to why I get that error each time I boot into another program (Fusion, then Parallels, then BootCamp, etc). Every single time I had to re-register and it triggered their "piracy alarm." Of course, MSFT support had no answer other than "you must have changed your hardware." :rolleyes:

Anyhow, those of you running Fusion and say it's so much faster than Parallels, how much RAM are you allocating it? My WinXP/Fusion is kinda slow (MacBook current version w/2GB RAM, fresh install of everything). I have allocated 512MB to Fusion.

SolRayz
Dec 19, 2007, 06:31 PM
I've tried both and I have stuck with VMware Fusion. It's more reliable and quicker and far less buggy especially ver. 1.1. Parallels seemed slower and glitchy, so this benchmark makes no senseto me.:confused:

Manatee
Dec 19, 2007, 06:34 PM
I've used both Parallels and Fusion since their first Betas, and I've found the performance of both to be very good. I used to use Virtual PC on my G4 machines, so I'm pleased as long as the VM runs at the same quantum level as bare metal -- which both Parallels and Fusion do.

My main concern -- far more important to me than speed -- is the integrity of the virtual machine and its interface with the host OS. That's the sort of comparison that I'm really interested in between Parallels and Fusion.

I've found the current versions of both VM products very stable, and very flexible. I've been using Vista for several months, and it has worked well with both Parallels and Fusion. I miss the Aero, though. I find much of the Vista interface overly complex and unintuitive. Just try finding where to click to shut down Vista. ;)

Hemingray
Dec 19, 2007, 06:52 PM
Wow, I never would have expected that... Parallels has always seemed to be the slowpoke to me. When I beta tested VMWare Fusion, it seemed a lot faster! Parallels just seems to keep eating up the CPU.

(P.S. Using XP, mind you.)

RedTomato
Dec 19, 2007, 07:00 PM
Thanks to Mac Tech for putting a lot of effort into this, but the article seems to be summarised as:


Strange tests, incomprehensible graphs, and garbled explanations.

Strange tests

Open a message in outlook, save a document, vertical scroll in Excel etc, are these things really worth benchmarking?

Incomprehensible graphs

Take a look at the first graph:

http://www.mactech.com/articles/mactech/Vol.24/24.02/VirtualizationBenchmark/Chart-OutlookonXP.jpg

There's a section titled 'Open Messages'. Are the relevant graph bars below or above the section title? It isn't clear. Looking at 'Print email' at the end, it seems the bars are set above the legend, which means bizarrely you have to read from right to left to understand the graph.

But looking at the left end of the chart, does that section show 'launch times' or an average of all the measurements? If not, why is no average shown? And who's Adam?

Furthermore, due to their presentation, it's hard to extract any useful information from the graphs.

Because the bars float about in some sort of x,y,z swimming pool, detached from their scales, it's hard to tell if, for example, the last yellow bar on the right in the above graph is higher or lower than the red bar behind it. Ditto for many of the bars on the other graphs.

Looking at most of the red bars, I have no idea where on the scale they are. Truly shocking. I feel like I'm looking at one of these examples of how NOT to make a graph.

Garbled explanations

This and the corresponding strange maths has already been pointed out several times in this thread. I won't beat the dead horse.

Di9it8
Dec 19, 2007, 07:08 PM
I have found VMWare completely reliable running on a MacBook Pro, compared to the constantly crashing pc that is virtualised.
No speed issues whatsoever, and switching between Mac an PC is no problem.
I would really recommend VMWare Fusion;)

chillywilly
Dec 19, 2007, 07:12 PM
Anyhow, those of you running Fusion and say it's so much faster than Parallels, how much RAM are you allocating it? My WinXP/Fusion is kinda slow (MacBook current version w/2GB RAM, fresh install of everything). I have allocated 512MB to Fusion.

I'm running 2gb on my MBP and I've given Fusion 768mb. If I'm doing a lot, I'll increase it to 1gb.

If I end up using it more often, I may have to upgrade to a new MBP and upgrade the RAM to 4gb, so I can allocate at least 1.5gb to Fusion.

But so far, the 768mb works for most stuff I do.

Towhead
Dec 19, 2007, 07:17 PM
FWIW, Ive installed PSoC designer under fusion on my macbook and without any issues it started working with the usb debugger/programmer automatically. Niiice.

Have not tried Parallels, and after reading all the other testimony here why should I, eh?

APPLENEWBIE
Dec 19, 2007, 07:49 PM
Anyhow, those of you running Fusion and say it's so much faster than Parallels, how much RAM are you allocating it? My WinXP/Fusion is kinda slow (MacBook current version w/2GB RAM, fresh install of everything). I have allocated 512MB to Fusion.


On iMac with VMware Fusion, about 700mb of ram allocated out of 2.5gb. On Parallels about 1000mb of ram allocated out of 2.5gb.

ks-man
Dec 19, 2007, 08:01 PM
I'll be soon buying a mac (longtime PC user) and had a couple of questions that seem to apply to this thread.

My wife and I have been planning on getting a laptop (probably MBP after a hopeful Macworld update) b/c I wasn't yet ready to retire the PC desktop (I use a desktop more than a laptop). I figured getting a Mac laptop would be a nice transition to using a mac and assuming that I can do everything I need to on my mac, getting an iMac in a few years (could get a MP but I do like the look of the iMac).

After reading this thread I'm wondering if Virtualization should lead us towards getting an iMac now. I like the idea of bootcamp but I still do enough on my PC that I don't want to have to restart the computer whenever I want to use Windows.

My main question is if people here can give me some real world examples or explanations on what I would notice comparing:

A 2 year old Dell running XP with 1GB of Ram and I think a 1.5GHZ processor vs. an entry level MBP with 2 GB of Ram vs. an iMac with 2GB of Ram.

How would these 3 computers stack up head to head running Boot Camp on the macs and how would they stack up running one of these two programs (probably Fusion) on the mac?

Thanks for the help.

GNice
Dec 19, 2007, 08:24 PM
Like many (most?) I have some real questions about this test. My firsthand experience is that VMWare is more stable and less resource intensive. Hmmmmmm.....

mgclayton
Dec 19, 2007, 08:40 PM
"In XP, Parallels is 17% faster than VMWare Fusion on XP and 1% faster than Boot Camp."

Wait... faster than Boot Camp?

Parallels was 1% faster than the Baseline PC, and NOT Boot Camp.
Boot Camp was 12% faster than the baseline PC, and therefore the fastest as you'd expect.
Hope that clears that up.

(Sorry, new to this, don't know how to quote things correctly)

BornAgainMac
Dec 19, 2007, 09:08 PM
I've used parallels v2 and 3 and found it so slow that it was unusable.

Boot camp is waaaaaaaaaaaaaaaaaay faster than parallels

Probably because of memory. Remember with Boot camp you are using all the available memory on your Mac. Running Parallels or VMWare you are allocating memory for your guest operating system. If you allocate too little then it will run slow. If you allocate too much then both your Mac will run slow and your guest OS.

I wouldn't run Parallels on a Mac with less than 2 GB of total memory and less than 1 GB for XP or Vista.

cgc
Dec 19, 2007, 09:12 PM
I'm really curious if they used more than one CPU/core with Fusion or BootCamp. I find it IMPOSSIBLE that Parallels is faster then BootCamp...I would be suspicious if either virtualization were even close.

BTW, I've tried Parallels 3 and own Fusion 1.1 and use BootCamp, all good but in order of speed (I've done my own unbiased benchmarks) is: Bootcamp, Fusion, Parallels with Parallels coming in last place.

twoodcc
Dec 19, 2007, 09:27 PM
nice to know this stuff. so if you want vista, get fusion, and if you want xp, get parallels

ppcaric
Dec 19, 2007, 09:39 PM
First of all: BootCamp is not a virtualization way of running windows. It runs windows natively on your mac. It only introduces a bios emulation needed at boot time, but afterwards it's 100% native windows. So... HOW can parallels be faster than BC?! :eek:

Besides... I currently run the latest versions of vmware on a macbook pro with core2duo cpu and, believe me, vmware is by far faster than parallels! The only thing at which parallels works better is at win-mac integration. I'm not only talking about coherence/unity mode, but also about data exchange between the two OSes. As for this aspect, parallels is much better, also giving you the chance to see windows data as mounted network positions or as mounted image files of the windows hard drive when the virtual machine is shut down.

Last but not least: vmware is much better and faster at virtualization for linux OSes. But this wasn't in the aim of the benchmark, so let's keep it out.

Conclusions? Parallels guys must have paid these guys for publishing such a biased and false benchmark! :cool:

arn
Dec 19, 2007, 09:55 PM
Parallels was 1% faster than the Baseline PC, and NOT Boot Camp.
Boot Camp was 12% faster than the baseline PC, and therefore the fastest as you'd expect.
Hope that clears that up.

The summary is correct. Directly from the article:


When testing the one step tests described above, on average, Parallels is 17% faster than VMware Fusion when running Windows XP, and 1% faster than Boot Camp.

Later in the article


And, as we said before, if you want the best XP performance with the types of applications tested here, Parallels is not only faster than VMware Fusion, but it's faster than Boot Camp on average for the applications that we tested.

sundoggy
Dec 19, 2007, 10:08 PM
Let's see what these posters are saying when virtual machines and boot camp kill off OS X and Apple becomes a high end Windows OEM. This coming year will make the path clear. Intuit has already dumped Quicken for the Mac. They say they're working on a Mac version for 2008 but those in the know say it will be a browser based, online version. Adobe will be next. Heck Photoshop Elements 6 is out for Windows for goodness sakes. The Mac version is 4. Just wait until the first big developer announces that Mac users will run the new version under bootcamp or virtualization. The rest will soon follow. Apple itself is worried about it and said so in recent guidance releases.:mad:

You are seriously deranged, or at least ignorant and out of touch. Have you seen the Mac market share numbers lately? They are soaring up. Granted they're no where near Windows, but then we know that the masses (you included) are ignorant.

WildPalms
Dec 19, 2007, 11:09 PM
Yeah i noticed that...it doesn't make sense.

Thats because thats one of worst test setups I've seen. Useless data.

mgclayton
Dec 19, 2007, 11:26 PM
The summary is correct. Directly from the article:



Later in the article

Ah yes, I see you got that from the Overview, but the article contradicts itself in the main section:

"By comparison, when we compared XP configurations against the baseline PC, we got these results:

* XP under Boot Camp averaged 12% faster than the baseline PC running XP

* XP under VMware Fusion averaged 1% faster than the baseline PC running XP

* XP under Parallels averaged 19% faster than the baseline PC running XP"

pamon
Dec 20, 2007, 12:16 AM
I use parallels and am very impressed with it. Run Win XP Pro on it and have been very satisfied in the speed and ease of use for parallels. Downloaded VM trial and never could get it going right.

dante@sisna.com
Dec 20, 2007, 02:16 AM
I run Fusion and it runs fast and great.

Have Win 2000 and Vista on it now. XP is coming Friday.

Very stable and fast. Setup was easy and flawless.

All features work, even my MacBook Pro's wireless keyboard and mouse.

ppcaric
Dec 20, 2007, 02:16 AM
Ah yes, I see you got that from the Overview, but the article contradicts itself in the main section:

"By comparison, when we compared XP configurations against the baseline PC, we got these results:

* XP under Boot Camp averaged 12% faster than the baseline PC running XP

* XP under VMware Fusion averaged 1% faster than the baseline PC running XP

* XP under Parallels averaged 19% faster than the baseline PC running XP"

This is no contradiction! It says, in order of speed, Parallels is 19% faster than the reference PC, BootCamp is 12% faster than the ref. PC (hence slower than Parallels), VMWare is only 1% faster (so the worst).

That said, it's self-evident that something must be wrong with the benchmark procedure they used. Native speed XP (BootCamp) CANNOT BE slower than the same XP on the same machine sharing memory and CPU power with MacOS X running (Parallels and VMWare)!!!
This article is pure trash (even without entering the newest Parallels vs oldest VMWare issue). :mad:

aswitcher
Dec 20, 2007, 02:18 AM
Just got Vista today for VMWare so pretty happy to see this.

solvs
Dec 20, 2007, 02:29 AM
nice to know this stuff. so if you want vista, get fusion, and if you want xp, get parallels
I wouldn't say that. My experience has been exactly the opposite with XP (haven't tried Vista). Read the rest of the thread, and you'll see I'm not alone when I say that. We use Parallels at work and it sucks. Not as bad as it was in the beginning, but I still don't like it. I've been using VMWare myself since before it came out, and while not perfect, is much better IMO. Especially when run from my BC partition.

For the record, I know some of the people who have worked on it, so I'm not exactly unbiased, but I still call bunk on these tests.

jhande
Dec 20, 2007, 03:05 AM
Parallels has a malicious feature called SmartSelect. It is apparently "ON" by default, or at least I didn't specifically enable it when I brought some Parallels 2.x VM images into Leopard with a fresh Parallels 3.0 install.

The result was that OS X was now registered three times to open a bunch of windows file types- one time for each of my three VMs. You can turn this feature off, but in my case I was left having to repair OS X after the damage was done. I had to do some manual cleanup including rebuilding my launch database.


Seriously! SmartSelect is definitely *not* a feature, it's a nasty pernicious piece of bleep. I've been through the same route, and it pisses me off to no end that I've paid for it.

What really escalated this to incandescent rage, was that there were no forums listed on parallels homepage (I think they *still* don't have a direct link there). I only found them when googling for the SS problem.

Parallels as a kit is OK, but suffers from extreme featuritis. Result: over the holidays I'm buying Fusion. I've been running the trial, and it's been rock solid.

.02 etc.

iSamurai
Dec 20, 2007, 04:28 AM
i get bluescreens every time i use vmware, it's probably winxp's fault, so for now i'm using parallels until vista sp1 is out then i can do a clean install.

heiesuke
Dec 20, 2007, 06:16 AM
Vmware 1.1 is great.....runs XP faster than many PC'c ive owned......

I gave up with Parallels.....just wouldent work.......

Co-sign!
This article appears to be biased BS as I know 1st hand that Fusion is way faster and taxes my system less than Parallels by far. They just blew any and all their credibility for me.

Data
Dec 20, 2007, 06:23 AM
After reading the whole thread, i feel that the article wants to suggest that parallels is faster / better, and from the reactions i read here most seem to think that VMware is faster /better. Most of the bechmarks are a mess and not clear at all. Very happy to read the real life experience from users here,i'd go for VMware for sure.

RedTomato
Dec 20, 2007, 06:33 AM
Most of the bechmarks are a mess and not clear at all.

It gets worse (see my earlier post on this). I also just realised that nowhere does it say if a longer bar on the graph means it took more time or less time (higher score) than the 100 benchmark!

Data
Dec 20, 2007, 06:47 AM
It gets worse (see my earlier post on this). I also just realised that nowhere does it say if a longer bar on the graph means it took more time or less time (higher score) than the 100 benchmark!

Lol , need we say more .

iBunny
Dec 20, 2007, 06:52 AM
VM Ware Fusion has been substantially faster than Parallels IMO.

I have had both, and unfortunatly I own both; purchasing Parallels first, and DL'ed the Trial of VM Ware only to be blow away. So I bought it and replaced parallels.

Oh well, trial and error it seems.

ericthered
Dec 20, 2007, 07:58 AM
Whatever...

I really tried parallels but it just wouldn't work on my Mac Pro and their tech support was simply horrible.

Vmware worked, as advertised with very little configuration. Fast doesn't mean ****** if the program doesn't work.

~Shard~
Dec 20, 2007, 08:44 AM
After reading this thread I am leaning towards purcashing VMWare as opposed to Parallels. Thanks for posting this story arn & Co., I think it has provided people with vaule information and feedback.

coolmemin
Dec 20, 2007, 09:28 AM
I switched to VMWare from Parallels because it had a much smaller effect on my Mac experience at idle, so I could leave it on without a noticeable speed decrease on my mac. Parallels constantly brought my system down to a halt, even with 4GB of Ram. THey might have improved since then but I'm sticking with VMware.

brantfordbrands
Dec 20, 2007, 02:12 PM
I know I'm not the first one here to dispute the article, however in November, I did my own testing using trial versions of VMWare and Parallels on the 2.4 ghz. 24" iMac and VMWare won hands down above Parallels, and in no way could I see ANY virtualization solution to be quicker than Boot Camp. The only thing that Parallels did have above VMWare was a slightly better ability to run DirectX programs...Some ran, but way too slow...

I didn't want to spend $80+ to get the wrong virtualization program, and Boot Camp is AWESOME, but alas, you have to boot to it.

Somebody mentioned games: Neither solution is good for 3D Gaming, stick with Boot Camp for that...

If you're curious about the results and methods, you can check it out on my blog:

http://www.coffeecopy.com/CoffeeCopy.com/Jeffs_Blog/Entries/2007/11/11_Entry_1.html

...Jeff

~Shard~
Dec 20, 2007, 02:24 PM
If you're curious about the results and methods, you can check it out on my blog:

http://www.coffeecopy.com/CoffeeCopy.com/Jeffs_Blog/Entries/2007/11/11_Entry_1.html

Thank you for your personal feedback and the link, I'll definitely be taking a look... :)

GoodWatch
Dec 20, 2007, 02:47 PM
The only thing I don't understand is how a virtualisation layer (running in a host OS) can outperform an OS that runs straight on the hardware. Bootcamp is a bootmanager and once you boot into XP, you run that straight on the Mac hardware. :confused:

~Shard~
Dec 20, 2007, 03:55 PM
The only thing I don't understand is how a virtualisation layer (running in a host OS) can outperform an OS that runs straight on the hardware. Bootcamp is a bootmanager and once you boot into XP, you run that straight on the Mac hardware. :confused:

I was wondering the same thing. Surely you lose something by using virtualization software?...

GoodWatch
Dec 20, 2007, 04:06 PM
I was wondering the same thing. Surely you lose something by using virtualization software?...

You do. Even with VMware ESX 3.x you lose about 4%, according to VMware themselves. And that is a very thin layer that runs directly on the hardware. I don't get it :confused:

qtx43
Dec 20, 2007, 06:22 PM
You do. Even with VMware ESX 3.x you lose about 4%, according to VMware themselves. And that is a very thin layer that runs directly on the hardware. I don't get it :confused:When the benchmark is something like "print email" or "open messages", then the time it takes is mostly due to the speed of hardware (network connection, hard drive, printer). As long as the processor can keep up, and there's enough memory to account for the overhead of the virtualization (no disk thrashing), then any time difference could very well be within the margin of error of measuring it. Which is to say, if you're just doing non-intensive tasks, like word processing, ordinary business stuff, then these things are fine. On the other hand, if you want to do heavy video processing, or play the latest games, then neither one is probably satisfactory.

solvs
Dec 20, 2007, 10:59 PM
I also just realised that nowhere does it say if a longer bar on the graph means it took more time or less time (higher score) than the 100 benchmark!

Maybe that's the confusion, that would explain how they seem to have everything completely backwards.

bamadonk
Dec 20, 2007, 11:14 PM
I have used and owned parallels and VMWare Fusion since both were beta. At first Parallels worked, but a few "upgrades" ago it began crashing constantly. Each time they release a new "upgrade" I try it again, and it continues to crash. In my opinion Parallels is trash. VMWare is very stable and never crashes.

Speed be damned. If it won't run 5 minutes without crashing what difference does a speed test make.

My primary use of the above programs is to play duplicate bridge online. I have written to Parallels about the problem and never received a response. In my opinion the program to buy is obvious.

~Shard~
Dec 20, 2007, 11:18 PM
I find it amusing that the actual purpose of this article was to efectively promote Parallels, yet now the opposite effect is occurring, as many people are finding out that Parallels in fact appears not to be as good as Fusion. In addition, the credibility of the people whom carried out these tests and wrote the review is up for criticism as well.

Very bastardly, I approve! :cool:

solvs
Dec 20, 2007, 11:30 PM
Well, since you can download trials of them both, anyone can see which is better.

And with the exception of a few posters, most of whom only prefer Parallels because they had issues with VMWare, the consensus tells us there is something wrong with either the tests, or the testers.

SourApple
Dec 21, 2007, 01:29 AM
Hi all,

I've been watching reviews of Parallels vs Fusion with interest as I am looking for a virtualisation solution, esp. given the attractive Christmas bundles that both now have -- ie. Fusion 2-for-1 discount (the $20 rebate does not apply to non-US customers) and the Parallels Premium Edition.

I am using a MacBook (Black, May 07) w/ 2GB RAM installed. I have mainly been running Boot Camp 1.4 beta on it (it came with OS X 10.4.10). I have a two-screen setup, and Boot Camp 1.4 beta does not often detect my external monitor. (I hear that this is fixed in Boot Camp 1.5 FINAL -- ie. a US$129 upgrade.)

My primary OS is Windows XP SP 2, running:
* MS Office 2003 (need Outlook)
* SI Station (www.sistation.com - stock monitoring programme)
* Internet Explorer v7 (as stock trading, i-banking and e-govt websites in Singapore generally don't like Firefox, let alone Safari)

Basically, I could have just bought a Windows laptop, but have always preferred a Mac. :-P (My PowerMac G4 Gigabit Ethernet, vintage 2000 version, is still sitting below my desk, with a Performa 475 sitting in cold storage ... and anyone wants a SCSI Syquest drives?)

My main questions for the forum are:


1) If I use the Finder to access a Word (.doc) file, will it be opened in MS Office 2003 or in the pre-loaded Mac Office 2004 (Trial version)?


2) Assuming I trash the Mac Office 2004 (trial version), what steps does it take to get MS Office 2003 to open the Word file under Fusion?

Do I have to first transfer the file to a shared directory (like how it was done in Connectix's Virtual PC) before I can open the file?


3) If (2) is true, then would that mean that any Office document clicked on in Mac Firefox will require additional step of saving to a shared directory first before it can be opened?


4) Following from (1), if I click on a .pdf file attachment in an Outlook message, will it open Adobe Reader for Mac or Windows?


5) Can the shared folder in Fusion be pointed to the "~/Documents" folder in the Mac partition. (My set up is OS X running under a HFS+ partition, with Boot Camp loading Win XP under a NTFS partition, both accessing a FAT32-formatted networked drive attached to my Airport Base Station.)

Will doing so leave OS X-specific .DStore and .Trash files all over the place?

Alternatively, is there a better way to set up a shared folder such that I have minimal duplication between my Mac OS X and Boot Camp partitions?

It appears that Parallels' Smart Select function (or "virus") takes care of this issue. (I do seriously wonder what kind of "voodoo" it does to handle the HFS+ to NTFS file system translation, and the corresponding performance hit.)


6) Under Fusion's Unity, will programmes like Faststone Image Capture do screen shots of the Mac Desktop? (This is basically a free version of ScreenCapture Pro for me.)


7) If I install Parallels 3 (5th December 2007) version, how likely is it to mess up my Boot Camp partition? (Seems that Fusion pretty much leaves it alone in line with the use of fewer .DLLs.)


8) Are the Parallels Premium Edition bundled software worth it? Is Kapersky running in the background going to eat up a lot of resources, eg. vs. BitDefender? Is the Acronis stuff worth it, eg. do you have to boot into Boot Camp to run them? Will it slow down Parallels by much?


9) What are the ideal settings for Fusion or Parallels under my configuration? I am a pretty mobile user that needs Windows to be running 24/7 (due to stock trading), and I am particularly concerned about drain on battery life when I am on the move.


Basically, if I can get past the "need many manual steps" integration of Fusion and involves significantly less battery drain, I'd likely go with it. But if the latest version of Parallels makes life a whole lot easier (without any beachballs of death) ... then I'd likely go with that.

At worst, I upgrade to Leopard and get my second monitor back. (I assume that both running Win XP under Parallels or Fusion will give me back my second monitor -- which Boot Camp beta took away.)


Thanks for any feedback and comments.

Regards,
Sour Apple
Singapore

solvs
Dec 21, 2007, 02:07 AM
Both have trial version. I suggest giving VMWare a try and running it from your Boot Camp partition. You should be able to do what you want to do.

user13
Dec 21, 2007, 03:17 AM
Whatever...

I really tried parallels but it just wouldn't work on my Mac Pro and their tech support was simply horrible.

Vmware worked, as advertised with very little configuration. Fast doesn't mean ****** if the program doesn't work.
What does it mean it wouldn't work? It simply didn't turn on or what?

As for me, I think there's not much difference between the two rivaling parties. Parallels is what I've been using for some time already and I'm pretty satisfied with it. The speed is acceptable, there's no memory leaks and it works rather fine. don't know 'bout the article, but it seems to be a bit biased.

JFreak
Dec 21, 2007, 03:22 AM
I don't get it how some virtualization system can be faster than the real deal (Boot Camp). Either the benchmark has it wrong or I have missed something...

solvs
Dec 21, 2007, 03:47 AM
I don't get it how some virtualization system can be faster than the real deal (Boot Camp).
They can't be and aren't, just one of the many things that doesn't make sense about the article.

AidenShaw
Dec 21, 2007, 09:44 AM
I don't get it how some virtualization system can be faster than the real deal (Boot Camp). Either the benchmark has it wrong or I have missed something...

It looks like the tests that Parallels was better than native were graphics and display intensive (like scrolling text).

I'd suspect that the native drivers are poorer than the OSX drivers (does Apple supply the graphics drivers, or are they from the graphics manufacturer?), and that Parallels has a good implementation for their graphics.

sushi
Dec 21, 2007, 10:41 AM
Personally, I am very happy with Parallels and the environment that it provides. I like having a separate window to simulate a PC. When using on my iMac, which has an external monitor connected, I throw Parallels over to it, and then run Windows in full screen mode on it leaving the Mac OS on the iMac. Works well for me running typical MSFT Office applications and such.

I am currently running Tiger. However, after I upgrade to Leopard, I would envision running Parallels in a separate space in full screen mode.

FWIW, I tried VMWare in the past and did not like how it worked. I may give it a try again based upon the comments in this thread.

Spievy
Dec 21, 2007, 03:44 PM
Somebody mentioned games: Neither solution is good for 3D Gaming, stick with Boot Camp for that...

If you're curious about the results and methods, you can check it out on my blog:

http://www.coffeecopy.com/CoffeeCopy.com/Jeffs_Blog/Entries/2007/11/11_Entry_1.html

...Jeff

I was really interested in your results but your 2nd page isn't working.

bankshot
Dec 21, 2007, 06:48 PM
Hilarious to see the VMware fanboys jumping up and down about how the methods and numbers must be wrong. Kinda like what you'd expect around here in general if an objective benchmark came out showing that Windows/Linux/non-Mac is better than OS X in some task. :rolleyes:

While it would have been nice to test the newest VMware, the authors seem to have a very good reason for not doing so: they took their time and methodically ran a huge set of tests over and over to get the most accurate results. So the latest versions of each were not out yet when they started. Or in other words, those were the latest versions when they started. If they'd switched to a new version every time it came out, it's quite possible that they'd never finish.

Hopefully they re-run their tests with the newer versions, now that the initial results are out.


On the question of how virtualization was able to beat out the real thing, I'm surprised that very few people seemed to understand that this is, indeed, possible. AidenShaw got it:

It looks like the tests that Parallels was better than native were graphics and display intensive (like scrolling text).

I'd suspect that the native drivers are poorer than the OSX drivers (does Apple supply the graphics drivers, or are they from the graphics manufacturer?), and that Parallels has a good implementation for their graphics.

In addition to that, there's also the possibility that either Parallels or OS X (or both) is able to speed up things like disk accesses by doing a better job of caching parts of the virtual hard disk file than XP did of caching the actual hard disk. If OS X preloaded a piece of data into memory and XP did not, the version running under Parallels would still benefit from the OS X caching while the Bootcamp version would not.

And we see that Vista, which reportedly caches much more aggressively, performs much better under Bootcamp than when virtualized. It's probably a combination of this and the 3-d graphics requirements that make Vista run best on bare metal.


Also keep in mind that as the authors point out, this is a limited set of tests designed to simulate certain tasks. Like all benchmark tests, they perform better under certain conditions and worse under others; and they may or may not represent real world usage.

ppcaric
Dec 22, 2007, 02:19 AM
In addition to that, there's also the possibility that either Parallels or OS X (or both) is able to speed up things like disk accesses by doing a better job of caching parts of the virtual hard disk file than XP did of caching the actual hard disk. If OS X preloaded a piece of data into memory and XP did not, the version running under Parallels would still benefit from the OS X caching while the Bootcamp version would not.


Anything can be done with the "right" benchmark :cool:
If you know of some specific tasks in which parallels does better, using caching for instance, than xp alone on the all hardware (bootcamp), then you can "build" a benchmark that voluntarily gives an unfair advantage to one of the challengers...

The question is: would you trust such a benchmark? :eek:

I propose one thing. Let's do our own benchmarks! Trial versions of both softwares are available. Many of us own one or both of them. Let us create a "true" benchmark.
Let's put inside something like zipping a huge file (an ubuntu iso) with 7-zip under windows, opening and scrolling to the end a large word file with ms word and a large pdf file with adobe reader, etc.

Or, even better, let's download one of the many windows trustable benchmarks available for free, let's run them on our parallels and vmwares and bootcamps and then let's use this thread to publish the results.

Let's show these guys how benchmarks ARE to be done! ;)

solvs
Dec 23, 2007, 01:49 AM
It looks like the tests that Parallels was better than native were graphics and display intensive (like scrolling text).
But the graphics card is simulated in Parallels, while in BC, they are just like the real thing (because they actually are).

I'd suspect that the native drivers are poorer than the OSX drivers (does Apple supply the graphics drivers, or are they from the graphics manufacturer?), and that Parallels has a good implementation for their graphics.
Except that the Parallels drivers made by Parallels. Which really aren't that great. The BC drivers are Apple's, which are actually pretty good (as per PC World's comments that they're the fastest PCs), but you can use the manufactures as well. ATI or Nvidia. They should be, and are, faster.

One of the many reasons those of us who actually use them know this can't be correct. ;)

Hilarious to see the VMware fanboys jumping up and down about how the methods and numbers must be wrong.
Yes, of course. We're all just VMW fanboys. Every single one of us. Except for that 1 poster who had a problem with it and now uses Parallels. :rolleyes: Or those of us who've used them are crying foul because in real world experience, they don't match reality. If they did, we wouldn't be complaining. ;)

While it would have been nice to test the newest VMware, the authors seem to have a very good reason for not doing so: they took their time and methodically ran a huge set of tests over and over to get the most accurate results.
But it still doesn't make sense. I've been using Fusion since before it came out, and besides a few minor issues, it's been better than Parallels, which I've also been using since early beta days (as I said, we use it at work, and only still do so because it came out of beta sooner and corp is slow to move to upgrade). Nether of them are better than BC though, even with the older drivers. Something must be wrong with these tests, and we're right to question them. Looking at the vast majority who do, I can't say I blame them since my experience matches theirs.

Hopefully they re-run their tests with the newer versions, now that the initial results are out.
Looking around at other sites that also question their credibility, I doubt it would help, but maybe they could at least do another preliminary one with the current software, at least before trying to defend this one.

On the question of how virtualization was able to beat out the real thing, I'm surprised that very few people seemed to understand that this is, indeed, possible. AidenShaw got it:
Except he was wrong. Parallels drivers and Apple drivers maybe, but not native ones. And emulation is still going to be slower than the real thing. Which it is. Which again, is one of the reasons why we don't trust the tests.

In addition to that, there's also the possibility that either Parallels or OS X (or both) is able to speed up things like disk accesses by doing a better job of caching parts of the virtual hard disk file than XP did of caching the actual hard disk. If OS X preloaded a piece of data into memory and XP did not, the version running under Parallels would still benefit from the OS X caching while the Bootcamp version would not.
I supposed it could, but it doesn't.

Also keep in mind that as the authors point out, this is a limited set of tests designed to simulate certain tasks. Like all benchmark tests, they perform better under certain conditions and worse under others; and they may or may not represent real world usage.
They don't, which is why we're questioning them. :)

RedTomato
Dec 24, 2007, 05:33 PM
I personally don't have a problem with the methodology. Arn said he was impressed with it, which is good enough for me.

My beef is with the poor selection of tests, dreadful writeup, and appalling presentation of the results. As I've said, the graphs are almost meaningless. Where is a table, or list of actual measured times?

freejack86
Jan 2, 2008, 11:20 PM
Funny, So much said for either side. One camp gave up on Parallels back before 3.0 was released(Build 3218 ver 2.5, 4128 :3.0, 4560:3.0, 5160:3.0, 5582 3.0; 5 builds ago in one year) another sides with Fusion 1.x/1.1.

These two companies will be flip flopping you endlessly with development, options and better builds.

Having two companies pull for my money is better than either giving up.

The proof in the pudding will be with VT-D comes out from Apple. Windows machines already have it, called Vpro.

Parallels has it for 4.0 coming this year, Fusion, not sure.

You'll be able to assign a graphics card to Virtual machine and not deal with writing code hash to run your video with "experimental" shaders.

Being able to directly communicate with the hardware and not through the OS shell will greatly improve your experience with 3D and Games.

VT-x? Sorry that means new hardware for you.

Bootcamp? Try running 36 different Guest OS's. Or at least 5 or 6 at a time.

Try carrying Bootcamp on an external Drive, can't do that, Virtual Machines can. You can even run a virtual machine from a USB key (8gb or 16gb preferred). Virtual machines can be portable.

Apple only writes drivers for Vista and XP. No plans for all the 36+ other operating systems of the real world.

So bootcamp for games/3d software. Better yet, get Parallels or Fusion too! They both support Bootcamp and Virtual Machines at the same time.

Buy Both $160.00 because Parallels does run some VM's better than Fusion and vice versa. You paid as much for 2 of your Xbox 360/Wii games and even more for your iPod Touch.


Cheers,:)

Krafty
Jan 2, 2008, 11:25 PM
I could never get parallels or bootcamp to work.

Eidorian
Jan 2, 2008, 11:33 PM
The proof in the pudding will be with VT-D comes out from Apple. Windows machines already have it, called Vpro. Cute, but VPro is a hardware management system. Virtualization is a requirement (I'm looking at you E4500, E6540) but it's not the cornerstone. Windows has nothing to do with this.

VT-x? Sorry that means new hardware for you.Already implemented. ;)

Bootcamp? Try running 36 different Guest OS's. Or at least 5 or 6 at a time.

Try carrying Bootcamp on an external Drive, can't do that, Virtual Machines can. You can even run a virtual machine from a USB key (8gb or 16gb preferred). Virtual machines can be portable. Not that we don't know this.

I could never get parallels or bootcamp to work.That's very detailed.

ayeying
Jan 2, 2008, 11:37 PM
I could never get parallels or bootcamp to work.

How does it "not" work?

Bootcamp? Try running 36 different Guest OS's. Or at least 5 or 6 at a time.

On an average user, they're just gonna run 1 or 2.

Krafty
Jan 3, 2008, 01:16 AM
bootcamp could never partition it hard drive and parallels always froze when I tried booting XP. VMware was the only thing that worked but I perferred bootcamp.

Eidorian
Jan 3, 2008, 01:19 AM
bootcamp could never partition it hard drive and parallels always froze when I tried booting XP. VMware was the only thing that worked but I perferred bootcamp.Boot Camp does have issues partitioning when the file system lacks a large enough section of contiguous space to split off.

VMWare works so what's the problem?

acray
Feb 22, 2008, 09:11 PM
Boot Camp does have issues partitioning when the file system lacks a large enough section of contiguous space to split off.


To expand on this. Boot Camp has to have X Gigabytes of "empty" hard drive space at the end of your disk. Where X is the size you specify for the Boot Camp partition.

However, data is stored somewhat randomly on the disk at a physical level. When the disk is less full it concentrates data in the beginning and middle of the disk, but over time some will get stored toward the end of the disk. Boot Camp will not move this data and thus fails when trying to create the partition because the space is not empty. (Note: The process is anything but "random" in reality, but that's a good way to think about it if you don't want to get into the technical details.)

Two solutions: Format your entire hard drive and then install Boot Camp asap before any data happens to get stored out at the end of your disk.

Alternatively, you can purchase a program like iDefrag to do an offline defrag of your hard drive in "compact" mode to move all your OSX data to the beginning of the disc and leaving the end free and allowing the Boot Camp setup to complete.

P.S. I recently setup Vista under the latest Parallel 3.0 build and used it for a few days. I'm currently installing Vista in the latest Fusion 1.1 build and I'll comment on the two once I've had some time to test Fusion.

Spievy
Feb 23, 2008, 07:25 AM
I have both Bootcamp and Fusion running Windows XP. I am so surprised how fast Fusion does run. I am extremely impressed with the speed. That being said I have not done any benchmarks, nor do I have any concrete evidence that supports my claim. But I have not booted into bootcamp since I installed it. I personally could get away with just fusion. The only negative thing I have noticed with fusion, it takes about 15 seconds longer to load the windows desktop than bootcamp. But that is just the "Tools" loading.

acray
Feb 23, 2008, 07:31 PM
Here's my anecdotal evidence, take from it what you will.

First a bit of background to let you make your own assumptions on any biases I may have. I work in the Windows Universe and began counting down the days to carting around a Mac laptop as soon as the MBPs were announced. I started out using Parallels (Fusion wouldn't even be in beta for a while yet) and was amazed at the performance. My old laptop wasn't too shabby, but Parallels still out performed it by a notable margin. -- Largely due to the Mac hardware.

I tried the beta for Fusion, but didn't really get into it. It didn't have the performance Parallels did. So I stuck with Parallels for my production VM and left Fusion for the experimental ones. However, every new build with Fusion seemed to provide notable performance increases - which is to be expected in a new product.

Fusion was the first to support Boot Camp. As soon as that came out I went to that and have been using Fusion running the Boot Camp install of XP ever since. I have continued to notice performance increases as Fusion as evolved.

My point in mentioning performance thus far is to demonstrate that Fusion has made considerable progress since the beginning. While I used Parallels there were similar upgrades, and only assume that they continued once I retired my Parallels VM.

I'll not say much about the much debated benchmark in this thread because most of what's worth saying has already been said. One thing I didn't see explicitly mentioned was that I think it was good practice that they stuck with one build all the way through the test. But one must acknowledge that based on a typical software development time table version 1.0's often have performance issues. So one would expect greater improvements (in the general case) among the initial followup releases to a 1.0 package than a 3.0 package. The reason being that the latter has already had the time to release several versions, each improving on the first. One of advantages of being first to market. Meaning that one would expect "larger" performance increases and fusion as it went from 1.0 to 1.1.1, while one would expect "smaller" performance increases in the various builds of 3.0.

On to my non-scientific review of Vista SP1 Ultimate on the latest builds of Parallels 3.0 (5584) and Fusion 1.1.1 (72241). My MBP has 2.16MHz Core Duo processors, and 2GB of RAM. Both VMs are running off 1GB of RAM and 40GB expanding virtual hard drives.

I will preface this by saying this was NOT a benchmark, just my real world experience when using Vista on the two platforms. I was using OSX at the time for various tasks as well. Obviously conditions in OSX and the two copies of Vista were not identical, but they were comparable as both tests working under my "standard" operating load.

I used Vista Parallels for two days. I spent the first day loading Windows, Office and applying SP1, installing software, etc. I spent the second day doing my day's work on Vista. However, the performance so poor: delays when opening menus and documents, screen refresh rates, choppy scrolling, etc. I that spent some time turning off the majority of the eye candy Vista allows you to. -- I did stay with the Vista (non-Aero) theme though. I'm a fairly impatient person when it comes to general responsiveness (as mentioned above) and I deemed Vista on Parallels as a valuable experimental tool (I'm a sys admin,) but too slow for daily use.

So I setup Vista SP1 Ultimate in Fusion. The only notable configuration difference in hardware and software is that I was using 2 virtual processors in the Fusion setup. I'd like to reiterate this was not a benchmark, but a write up on what I experienced using this software in the "real world," so it may or may not be considered an apples to apples test. So far Fusion has performed much better under Vista than Parallels did. Install times of the various programs and updates, while not timed, seemed to go much faster. The responsiveness of Vista on Fusion is much better. Usable even. Audio is a bit choppy, but I assume that is a driver issue that can be corrected (haven't bothered to look into it as I don't do anything with audio in Windows.) I did notice that OSX did take a larger, but livable, performance hit when the VM was cranking away at full speed, like during boot up. - I attribute this to Fusion running in multi-threaded mode, where as before it would never use more than 50% of the Mac's cpu power. OSX had comparable performance during standard usage of both VMs though.

I have my Boot Camp XP install (that I run through Fusion) stripped down to just about bare bones, and the performance of XP vs Vista is what you would expect. My XP has better responsiveness in general than Vista. Interestingly Outlook 2007 was more responsive in Vista than XP. Vista does run fast enough for me to consider using it for my production VM, a few more days of testing will determine that.

I notice two problems with the Vista VMs. 1) The Netflix streaming movie player would not work on Parallels, the error message said the video card was not compatible. I only spent a few minutes troubleshooting and did not resolve. A problem just with the Netflix Windows Media Player plugin or possible DRM issues with Parallels and Vista? 2) The a fore mentioned sound problem in Fusion. -- I did not experience any stability issues with either, but also haven't tested them long enough to say anything conclusive about there stability.

Some of my non-performance related opinions on Parallels and Fusion:
-I use the Windows VM for work, and OSX for personal use. I like keeping them isolated or sandboxed from each other, I don't want anything more than shared network folders for the occasional file transfer. Fusion is built more towards that end than Parallels.
-Related to the previous item, I always work in full screen (putting Windows on my external display and keeping OSX on the laptop display) or windowed mode. However, I would use the Unity view on occasion if it supported spreading windows across multiple displays. Parallels' Coherence does just that.
-When running in Windowed mode, Parallels has a hardware icons that flash with activity. Fusion's icons are in one of two states, connected/disconnected. I like the activity monitors; one of those "little things."
-I run bridged network connections. When I connect my Mac to different networks Parallels would automatically disconnect/reconnect the VM's adapter, triggering a renewal of the IP address. Fusion does not do this and I have to renew my IP manually (via a simple script I wrote). Another one of those little things.

Conclusion: If I were going to be running Vista, I would only run it on Fusion using 2 virtual processors. I currently use Fusion for running my Boot Camp install of XP, and am quite happy with it. If I had a day to kill I would setup an XP install on Parallels to see if 3.0 holds up against Fusion (I've never ran XP in 3.0.) If Parallels performed well I would probably switch over and disable all of the integration features so I could use the dual screen coherence feature.

Rhinestorm
Feb 23, 2008, 08:26 PM
Don't give into this crap. Parallels is unstable, unreliable, and slower than VMWare Fusion.

chickenninja
Feb 24, 2008, 03:12 AM
native rocks the socks out of any emulation

chillywilly
Feb 24, 2008, 11:36 AM
native rocks the socks out of any emulation

Sure it does. But if you don't spend much time in Windows, and need it only to do a few things, it's nice to have it within a minutes reach, do what you need, switch back over to the Mac side, then switch back. It's beats having to reboot multiple times just to do a few things.

If, however, you do a lot of processor intensive work under Windows, then Boot Camp is your choice.

Fusion gives you the best of both worlds, by allowing you to use your Boot Camp partition and the choice to open a Windows session along side your Mac session.

acray
Feb 24, 2008, 02:35 PM
Fusion gives you the best of both worlds, by allowing you to use your Boot Camp partition and the choice to open a Windows session along side your Mac session.

Parallels advertises support for running the boot camp install now too. I've never used it though.

drditty
Feb 24, 2008, 02:52 PM
I have been using Parallels for approx. 10 mos. In the medical world, EMRs (electronic medical records) are windows only. In addition, I use Quickbooks pro, and this too has to be in Windows. Finally, I have to use Explorer to run payroll for the practice, and this too I can only use in Windows. Parallels worked well enough that I gave away my PC last summer. I was very happy to see it go. When Leopard came out, things were really bad for a few weeks or months with multiple crashes. However, with subsequent Leopard upgrades and Parallels updates, things are now wonderful for me. I never use windows except for the above reasons, and do everything else on the Mac side. I have been more than happy with the experience over the last few mos. Note that I'm not a gamer. In addition, in my group of 12 docs, 4 have purchases macs over the past 6 mos. These are people who have never owned a mac in their lives. I haven't tried VM, but I plan to try it on the MBA my wife just got to see how it compares. If your needs are similar to mine, then I think that Parallels has been a great product, despite the negative vibe it has received on this thread.

zenwhen
Feb 24, 2008, 06:32 PM
These benchmarks are very wrong. As soon as I saw them ranking emulated windows over it running directly on the hardware, I discounted everything else they said.

acray
Feb 25, 2008, 12:39 AM
These benchmarks are very wrong. As soon as I saw them ranking emulated windows over it running directly on the hardware, I discounted everything else they said.

You have to play close attention to the report, as it is written poorly, because it does not say that the virtualization solutions out performed the tests run in boot camp, or "bare metal."

What it did say is that they out performed their "base line PC," which they made some brief mention to being spec'd similarly. It is quite possible, likely actually, that different hardware with the same "specs" will perform differently.

So this doesn't point out an impossibility in the report (like others have mentioned in this thread,) but a bias/ignorance on how to conduct a sound benchmark.

The way I see it they could be out to intentionally misrepresent the results by introducing irrelevant and misleading stats like their "base line PC" and you should then assume the rest of the tests were designed to give the "right" answer. Or they could just be ignorant of the fundamentals of benchmarking and think the "base line PC" contributes to their findings. Even with pure intentions, you can't conduct an accurate test if you don't understand what you're testing.

In either case, the inclusion of the "base line PC" calls into question the validity of the entire test suite.

bloomberg
Feb 25, 2008, 09:53 PM
I tried vmware first since it's already familiar from running on linux - a version they have given away free for years. It worked well enough I never got around to trying out Parallels and just bought the trial version. It's not perfect, like how it gets confused when unplugging dual-screen monitors, or how unity-mode gets all kinds of confused sometimes and you have to kill it, for example, but good enough.

I use two setups, one for 'serious' work that uses both cores and 2GB ram that runs in its own 'space' in full screen mode with dual monitors. The other is mostly for browser testing with just one core and one GB ram that is kept suspended between uses. That setup works really well for me.

Cubytus
Aug 20, 2008, 10:10 AM
acray did a better test in my opinion. It is much more representative of real-world usage.

bpl323
Aug 20, 2008, 01:38 PM
Hmm... I wish WINE worked better.

Luvsthesun1
Feb 10, 2012, 03:29 PM
I am going to a vmware bootcamp (http://http://www.globalittraining.net/vmware-training/vsphere/vsphere5) next week for vmware training (http://www.globalittraining.net/vmware-training/vsphere/vmware-vsphere) and I will see what information they have on this subject.

thermodynamic
Apr 13, 2012, 04:51 PM
Hmm... I wish WINE worked better.

Ditto.

Or Crossover Office for that matter... those would be best, but Parallels 7 I've found to be formidable...

I am going to a vmware bootcamp (http://http://www.globalittraining.net/vmware-training/vsphere/vsphere5) next week for vmware training (http://www.globalittraining.net/vmware-training/vsphere/vmware-vsphere) and I will see what information they have on this subject.

Thanks!

I've been trying to find benchmarks of XP vs 7 in a virtual session and everything I've found compares XP and *ista. Vista fared poorly in VMs, but I'm keen on trying Win7 and seeing how Office 2010 (w/Visio) interacts. XP is okay, but if Win7 offers a more refined and faster experience than XP, I'd eat up the extra disk space and use it instead... especially on my MBP!

acray
Apr 16, 2012, 02:18 AM
Everything else is definitely not equal in my case, but I have an XP, 7, and Ubuntu VMs on my MBP.

Out of those three Windows 7 performs the best. And it is what I use the most.

It has it's glitches now and then, but it runs better than I remember XP running back when it was my primary. Though comparing Win7 running in Parallels 7 and on a new MBP to running XP in Parallels 3 on a MBP that is a few generations older/slower...

If I took the time to do semi-controlled test like my write up in 2008 between XP and 7, I don't know which one would be faster. But frankly I'm not interested. Win7/Parallels works well enough so that I'm not cursing the hourglass too often, and I prefer 7 over XP.

Office 2010 works well. I don't have Visio installed, but I know i've used Visio 2007 in a VM so I'd bet 2010 works as well.

adder7712
Apr 16, 2012, 04:38 AM
Nah, I would run Windows in Boot Camp. Native hardware support.