Home > Blogs > VMware Fusion Blog

Performance Matters

We've recently seen yet another round of claims about VMware Fusion’s performance. We take performance very seriously, so we dug into the published claims as well as their testing methodology.

We actually appreciate it when someone points out that they think we're not the best because it gives us an opportunity to verify our own assumptions, and revisit the always important question of where performance matters most to our users. When it comes to running Windows on a Mac, we find that our customers care most about stability, reliability, and getting work done.

So, when we run performance benchmarks, we focus on measuring tasks that are the kinds of things that our users actually do. We prefer tests that don't focus too intensely on one specific application type or another. Our experience is that benchmarks like PCWorld's WorldBench 6 and Microsoft's own Windows Experience Index give us a more balanced view of overall performance. This leads us to what we believe are more objective measurements of what a Windows user is likely to experience in the real world.

The Windows Experience Index
The Windows Experience Index performs five specific tests that measure processor, memory, general graphics, gaming graphics, and hard drive performance. When we ran these tests on three different Macs we found that VMware Fusion delivers world-class performance, with the added benefit of reliability and stability users have come to depend on from VMware.

The following chart shows the performance differences we measured on one of our test machines, including test runs with Boot Camp. We configured the two different test VMs as similarly as we were able to, making sure that both used the same settings for CPU, memory, and same type of virtual hard drive. We also ran each test 3 times; the chart is based on the average result of the 3 runs. (The results were highly consistent for all three test configurations; the averaging process didn't obscure any outliers.)


You may notice that hard drive performance is–counterintuitively–identical for Boot Camp. By default, Boot Camp partitions are placed by the Boot Camp Assistant at the very end of a disk. With hard drives, ie non-SSDs, there's a big performance difference between accessing data at the beginning and the end of a disk. Based on the hard data we measured, we could legitimately claim that our hard drive performance is equal to native performance. While factually accurate, such a claim would be thoroughly misleading. The reality is that if you put your Windows data at the very end of a hard disk, performance won’t be as good.

This is a clear example why we feel it's critical to both compare performance characteristics that really are equal as well as present the resulting data in context. This isn’t just important, but we feel it’s the only upfront way to do it.

In all fairness, we acknowledge that we have a little work to do on gaming graphics in order to meet the needs of hard core gamers. That said, VMware Fusion 3 successfully passes the Autodesk testing suite for AutoCAD 2010 and AutoCAD 2011. This should give VMware Fusion users the confidence to run important Windows- based 3D apps reliably on their Macs.
WorldBench 6
WorldBench 6 is the latest version of PCWorld's application-based benchmark suite that includes tests performed within Microsoft Office, Adobe Photoshop, Autodesk 3ds Max, Roxio VideoWave Movie Creator, and Mozilla Firefox. Each app is used as it would be in real-world situations, and the individual results are aggregated to derive a final, combined WorldBench score.
With WorldBench 6, VMware Fusion 3.1.2 is between 4 and 15 percent faster than Parallels Desktop 6, depending on the test hardware:


We know that users depend on VMware Fusion for running the apps they need for their work. To that end, we will continue to focus our energy on building the best Mac virtualization product for our customers. We remain very interested in hearing constructive criticism about VMware Fusion, and are delighted to be able to give you a little insight into how we think about performance, and why. 

Buzz Around Town, Welcome Macworld 2011
This is always a fun time of year. Mac enthusiasts (sometimes with their families) come to San Francisco for a few days of mingling with other Mac folk and checking out the latest and greatest.
The city is buzzing and it reminds us fondly of the conversations we had with many VMware Fusion users at previous Macworld events. We put together a little video with some highlights from last year:


While we're not exhibiting this year, don't think we're not wandering the show floor like everyone else, drinking in the atmosphere. We're very proud both of our Mac product, and of being part of the Mac community.

And for the token marketing plug, the VMware Online Store is hosting a special 20% off promotion in celebration of Macworld.  Combined with a $30 rebate, you can get VMware Fusion for $33.99 through 11:59PM PT on Saturday night (January 29).  Happy Macworld!

7 thoughts on “Performance Matters

  1. Martin

    “So, when we run performance benchmarks, we focus on measuring tasks that are the kinds of things that our users actually do”
    Don’t mean to be rude, but how do you know what I do with your product?
    Just so you know, I bought Fusion right from version 1 (was with it right from the beta actually) but I’ve not used it in probably 6 months as it’s too slow.
    Don’t get me wrong. I’m sure any virtualisation would be just as slow (or fast). Thing is, when you add the continual Office activation nagging every time I boot my bootcamp partition under fusion and the awful unity glitches it’s just far easier to reboot these days.
    Shame really. I’d love it to work well but it’s just easier to reboot!

  2. Tom

    VMWare Fusion is amazing.
    As an Apple Consultant, I spend a lot of time with both Parallels, and Fusion. There are a number of problems that make Parallels not the optimal choice. in a good number of situations. Binding to an AD Domain in a virtual machine “just works” in VMWare, not so much in Parallels.
    Then there is the cross-platform support. I can take my VMWare VMDK to any pc, install vmware, boot it there and move on,. that and that alone makes it worth 50x more than Parallels.

  3. Mikero

    Martin, the issues you describe all sound like anomalies.
    I’d recommend filing a Support Request so our engineers can help you with that.

  4. Matt

    I believe this is something not true…. At least on my i5 MacBookPro Parallels wins if compare Windows Experience Index.
    It also feels to boot VM much faster and Quicken starts faster… So artificial benchmarks like that seem to be some kind of a marketing bullshit.

  5. Pingback: Compare: Fusion 4 vs Parallels 7

Leave a Reply

Your email address will not be published. Required fields are marked *