Another year, another version of Parallels to (hopefully) bring us closer to achieving Boot Camp levels of performance whilst remaining within Mac OS’s loving embrace.
Parallels Desktop 9 for Mac is now available, and just like last year, we’ll see if it provides a substantial improvement in games performance over version 8. Again, we’ll record the performance of several Windows-only games running in three environments:
- Parallels Desktop 8
- Parallels Desktop 9
- Native (Boot Camp)
We’ll be using the ubiquitous Fraps to capture various data, like frame rate over time, and record the output.
Set up
We used the following configuration to run the tests:
- Mac Pro 2010 2.8 Quad-Core Intel Xeon (4 CPUs allocated to Parallels)
- 10 GB RAM (4 GB allocated to Parallels)
- NVidia GTX 670 (512 MB video RAM allocated to Parallels)
- DIY Fusion Drive (Mac OS)
- Combination of SSD and 7200 RPM disk drive (depending upon game) for Windows
- DirectX 10
- Vertical synchronisation enabled
- Full-screen (1920×1080) in all cases
- Windows 7 (Boot Camp configuration)
- Parallels Desktop 8.0.18608
- Parallels Desktop 9.0.23036
- All games tested had settings optimised to run smoothly in Windows rather than Parallels.
Some important notes before we begin:
Testing methodology
For each of the tests, we used a built-in benchmarking utility if one was available, otherwise loaded up a save game from the mid-point of the game and recorded playthroughs where we tried to perform the same actions each time. Slight variations in gameplay are therefore to be expected.
Parallels Desktop 9 Supported Platforms
One of the purported benefits of Parallels 9 over previous versions is compatibility with Windows 8.1 and OS 10.9 “Mavericks”. We didn’t take advantage of either of these, being that most Windows gamers are currently running Windows 7, and OS 10.9 has not been released yet.
DirectX Version
We ran DirectX 10 versions of every game (where available), as that’s the latest supported DirectX version in Parallels. Though we did our best to ensure that none of the games tested used DirectX 11 when doing the Boot Camp versions, it’s possible that some of the games do this automatically without letting you know.
It’s also worth pointing out that where DirectX 11 versions of games are available, performance in Boot Camp may be better because of this, in addition to the differences seen in our benchmarking process.
Overall Performance
As noted above, we optimised the settings to run in Boot Camp. Boot Camp results are only used as a baseline reference. Just because the overall frame rate appears low in any of these tests, doesn’t mean that you wouldn’t be able to change the settings to get a decent frame rate in either versions of Parallels. Frame rates recorded are therefore for the purpose of relative comparison only, and should not be taken as an indication of real-world performance.
Likewise, having FRAPS record and measure gameplay itself has a significant impact on performance, so expect real-world performance to be better than the figures below.
Differences between Parallels Desktop versions
In terms of graphical configuration, there was very little difference between either version of Parallels.
This is the configuration screen for Parallels Desktop 8:
And here’s the same screen in version 9:
To the test-a-tron!
Far Cry 3 Test
The game that some called game of the year, Far Cry 3 is no slouch when it comes to visuals. As with most of Ubisoft’s titles, even though they’re happy to feature Macs in their commercials, no-one there thinks to actually support the platform. Parallels then might be the only way for Mac gamers to give Far Cry 3 a spin.
Benchmark video
Far Cry 3 frames over time (lower is better)
Far Cry 3 frame rate over time (higher is better)
Test | Far Cry 3 PD8 | Far Cry 3 PD9 | Far Cry 3 BC |
Min | 7 | 12 | 24 |
Max | 23 | 21 | 37 |
Average | 16.638 | 16.897 | 28.999 |
So a very slight improvement over PD8 there, but still a ways to go to catch up to Boot Camp.
Crysis 2 Test
Crysis was always a hallmark of how detailed PC games could be. Crysis 2 was a little more consolified than its predecessor, but it is a graphical powerhouse nonetheless.
Benchmark video
Crysis 2 frames over time (lower is better)
Crysis 2 frame rate over time (higher is better)
Test | Crysis 2 PD8 | Crysis 2 PD9 | Crysis 2 BC |
Min | 0 | 0 | 28 |
Max | 19 | 18 | 48 |
Average | 12.644 | 11.526 | 30.39 |
It seems that if anything, the frame rate is slightly better in version 8 here, but they are both well below the Boot Camp rate (which appears to have been capped to 30). Also worth noting that the Parallels versions both looked rather blurry.
Mass Effect 3 Test
Another popular game franchise that has failed to materialise on Mac OS, the controversial space opera features lots of pretty lasers to point our calculators at.
Benchmark video
Mass Effect 3 frames over time (lower is better)
Mass Effect 3 frame rate over time (higher is better)
Test | Mass Effect 3 PD8 | Mass Effect 3 PD9 | Mass Effect 3 BC |
Min | 1 | 1 | 54 |
Max | 28 | 28 | 62 |
Average | 17.848 | 17.277 | 59.408 |
Again we’re seeing slightly better performance overall in Parallels Desktop 8, and again we should note that the Boot Camp frame rate has been capped, this time at 60 fps.
Kingdoms of Amalur: Reckoning Test
A fast-paced fantasy RPG that didn’t work out too well for its developers, but is a pretty good game on the whole, this game is like a modern take on World of Warcraft.
Benchmark video
Kingdoms of Amalur: Reckoning frames over time (lower is better)
Kingdoms of Amalur: Reckoning frame rate over time (higher is better)
Test | Reckoning PD8 | Reckoning PD9 | Reckoning BC |
Min | 19 | 19 | 28 |
Max | 36 | 37 | 85 |
Average | 27.836 | 26.778 | 58.08 |
Really not much in it. Again we’re seeing a hairline improvement in version 8, and frame rate capped at 60 in Boot Camp.
Dawn of War II Test
Relic’s Warhammer 40k-themed real-time strategy has a built-in benchmark test, so that’s a useful way to give us directly comparable results.
Benchmark video
Dawn of War II frames over time (lower is better)
Dawn of War II frame rate over time (higher is better)
Test | DOW2 PD8 | DOW2 PD9 | DOW2 BC |
Min | 0 | 0 | 25 |
Max | 26 | 28 | 66 |
Average | 11.471 | 12.313 | 39.982 |
Marginal improvement with Parallels 9 again.
Warhammer 40,000: Space Marine Test
Staying in the Warhammer 40k universe, we move from RTS to third-person shooter with Space Marine.
Benchmark video
Space Marine frames over time (lower is better)
Space Marine frame rate over time (higher is better)
Test | SpaceMarine PD8 | SpaceMarine PD9 | SpaceMarine BC |
Min | 15 | 9 | 29 |
Max | 33 | 29 | 78 |
Average | 22.085 | 19.627 | 49.18 |
Parallels 9 is doing a bit worse here, and the Boot Camp test is capped at 60 fps.
L. A. Noire Test
Another rather good game from a now defunct studio, L. A. Noire took the GTA formula to the 1940s.
Benchmark video
L. A. Noire frames over time (lower is better)
L. A. Noire frame rate over time (higher is better)
Test | LANoire PD8 | LANoire PD9 | LANoire BC |
Min | 0 | 0 | 10 |
Max | 31 | 31 | 32 |
Average | 12.484 | 10.211 | 25.153 |
Although the game ran in Parallels, it was basically unplayable compared to the Boot Camp version (which itself performed surprisingly badly).
S.T.A.L.K.E.R Shadow of Chernobyl (modded) Test
STALKER is a highly praised open-world first person shooter. We applied the essential Stalker Complete 2009 mod and gave it a burn.
Benchmark video
Stalker frames over time (lower is better)
Stalker frame rate over time (higher is better)
Test | Stalker PD8 | Stalker PD9 | Stalker BC |
Min | – | 3 | 0 |
Max | – | 25 | 90 |
Average | – | 15.316 | 33.134 |
Of all the games we tested, this was the only one that ran on Parallels Desktop 9 but not version 8.
The Elder Scrolls V: Skyrim (modded) Test
We tested Bethesda’s ubiquitous RPG loaded up with several graphical mods.
Benchmark video
Skyrim frames over time (lower is better)
Skyrim frame rate over time (higher is better)
Test | Skyrim PD8 | Skyrim PD9 | Skyrim BC |
Min | 0 | 0 | 22 |
Max | 22 | 18 | 44 |
Average | 5.792 | 9.903 | 30.252 |
Although Parallels Desktop 9 seems significantly better here, it’s worth noting that something caused the game to stall completely in both versions.
Super Street Fighter IV Test
SSF4 also has a built-in benchmarking test, but sadly it doesn’t run the same test every time.
Benchmark video
SSF4 frames over time (lower is better)
SSF4 frame rate over time (higher is better)
Test | SSFIV PD8 | SSFIV PD9 | SSFIV BC |
Min | 20 | 19 | 55 |
Max | 56 | 36 | 78 |
Average | 28.138 | 28.565 | 62.248 |
Although the performance was nearly identical in both Parallels 8 and 9, Parallels Desktop 9 seems to show less variance.
Fallout: New Vegas (modded) Test
Obsidian’s guns-and-conversation first-person RPG also gets a mod makeover, and run through our analytical visors.
Benchmark video
Fallout: New Vegas frames over time (lower is better)
Fallout: New Vegas frame rate over time (higher is better)
Test | FalloutNV PD8 | FalloutNV PD9 | FalloutNV BC |
Min | 0 | 7 | 25 |
Max | 22 | 21 | 54 |
Average | 11.6 | 14.091 | 30.189 |
This has showed the greatest improvement between 8 and 9 of all our tests, but even then it’s not that significant.
Renegade Ops Test
An old-school top-down shooter with new-school graphics. We lock and load some data and find out how it performs.
Benchmark video
Renegade Ops frames over time (lower is better)
Renegade Ops frame rate over time (higher is better)
Test | RenegadeOps PD8 | RenegadeOps PD9 | RenegadeOps BC |
Min | 7 | 7 | 56 |
Max | 16 | 24 | 62 |
Average | 13.156 | 13.188 | 59.939 |
And that’s about as close as it gets. Both Parallels versions were extremely blurry.
Failed Tests
For the sake of completeness, we also tried to test the following games, which did not run properly with either version of Parallels Desktop:
- Battlefield 3
- Just Cause 2
- Metro 2033
- Tomb Raider (2013)
Conclusion
It seems that Parallels Desktop 9 isn’t the significant upgrade in terms of graphical performance that Parallels Desktop 8 was. There’s a marginal improvement in frame rate at best, and a decrease in performance at worst; and really not much difference in terms of compatibility. Compared to our Boot Camp base line, there’s still room for improvement. It will be interesting to see how OpenGL 4 in Mavericks changes the balance once it’s released.
We did note (but did not qualitatively test) significantly improved loading times for many of the games tested however, so you might feel that’s worth the upgrade cost, or you might simply want to make use of the 6 months Parallels Access (that we also took a look at) that’s included.
That said, really I was pleasantly surprised by the number of games either version of Parallels would run now, and all the above games can certainly be tweaked to run adequately. If you’re not already a Parallels Desktop user, now might be the best time to take a closer look.