[Macworld]

Post your comments for Parallels 5 (build 9308) here

[People_Eater]

Something's wrong when you need to mention the build number of an application in the title of a review.

[griffman]

That's how Parallels denotes their releases, so we want to be sure people know which version we tested. (There's a newer build, 9310, which came out just as we went final with this.)

-rob.

[People_Eater]

griffman, on 26 January 2010 - 08:39 AM, said:

That's how Parallels denotes their releases, so we want to be sure people know which version we tested. (There's a newer build, 9310, which came out just as we went final with this.)

-rob.

Right. Something's wrong. Why do they need to release so many builds?

[EDronkert]

Seems like Parallels supports bleeding edge features while sacrificing some stability (sleep/MacLook issues).

[griffman]

Parallels would have to answer that question . But they haven't released any more than has VirtualBox, and only one more than did Fusion in the same time frame, so it's not like it's outrageous.

They just choose an odd naming convention; I'd much rather see 5.0.1, 5.0.2, etc., but they use a much more complex sequence. Be thankful I didn't include the full version number, which is:

5.0.9310.546799

Urgh.

-rob.

[perdygood]

Looks like the spelling and grammar checkers are broken under Parallels. ;-)

I am also dismayed that Parallels and VMware use build numbers instead of a more standard versioning scheme (x.x.x). When confirming I have the latest version, it's much more difficult or error prone to compare build numbers. I really don't care how many times the developer had to build the software. Build numbers tell me the software is beta quality.

[griffman]

EDronkert, on 26 January 2010 - 09:02 AM, said:

Seems like Parallels supports bleeding edge features while sacrificing some stability (sleep/MacLook issues).

I wouldn't characterize it as sacrificing stability, but sacrificing some functionality. I never experienced an outright crash in Parallels during my entire time testing the program, while I did manage to bring down both Fusion and VirtualBox at one point.

But yes, a good summary of the differences between Fusion and Parallels is that Fusion is a more refined, seemingly better tested product that gives up something in the speed department to make sure everything works well. Parallels is loaded with features, and is amazingly fast, but those features and speed come at the expense of some rough spots in implementation and upgrading.

-rob.

[People_Eater]

griffman, on 26 January 2010 - 09:04 AM, said:

Parallels would have to answer that question . But they haven't released any more than has VirtualBox, and only one more than did Fusion in the same time frame, so it's not like it's outrageous.

They just choose an odd naming convention; I'd much rather see 5.0.1, 5.0.2, etc., but they use a much more complex sequence. Be thankful I didn't include the full version number, which is:

5.0.9310.546799

Urgh.

-rob.

It's more than just a naming convention. "Builds" are for developers and beta testers. This really indicates that they're just throwing releases out, without putting them through proper testing and QA procedures.

[griffman]

perdygood, on 26 January 2010 - 09:05 AM, said:

Looks like the spelling and grammar checkers are broken under Parallels. ;-)

Wow, that's odd. I fixed all those mistakes once yesterday, too. Sigh. Fixed again now. Sorry about that.

Quote

I am also dismayed that Parallels and VMware use build numbers instead of a more standard versioning scheme (x.x.x).

Fusion uses a standard naming convention -- it's now Fusion 3.0.1, without any need for a build number.

-rob.

[TheBum]

Does antialiasing work in any of the games under Parallels 5?

I use the Phoenix RC flight simulator and even though the antialiasing option is present and selectable in the program settings, I don't see any evidence in the actual rendering that it's being done. I'm just wondering if it's my old hardware (2.0GHz Core Duo MacBook Pro w/ ATI Radeon X1600 Mobile graphics) or a deficiency in Parallels.

Antialiasing works fine in Boot Camp.

This post has been edited by TheBum: 26 January 2010 - 09:27 AM

[pkay]

So, how much of the speedup in Parallels 5 over Fusion 3 is due to the use of more cores (eg, all 8 vs. max of 4, for Parallels and Fusion, respectively) and use of OpenGL 2 vs. OpenGL 1.4 for Windows 7, again respectively? Can we attribute all of the performance gain to hardware acceleration or are there other implementation differences that contribute significantly?

[pkay]

BTW Rob, thanks for reviewing both Parallels and Fusion at the same time. That's by far the best way to help your readers make informed decisions, and I appreciate it.

[griffman]

pkay, on 26 January 2010 - 09:30 AM, said:

So, how much of the speedup in Parallels 5 over Fusion 3 is due to the use of more cores (eg, all 8 vs. max of 4, for Parallels and Fusion, respectively) and use of OpenGL 2 vs. OpenGL 1.4 for Windows 7, again respectively? Can we attribute all of the performance gain to hardware acceleration or are there other implementation differences that contribute significantly?

Given I tested on a four-core Mac Pro, and I only used two cores, none of the speedup is due to support for more cores. OpenGL is faster in Parallels than in Fusion in XP Pro (where both have it). Tomorrow you'll see a summary of all the speed tests we performed, and you'll see that Parallels was faster pretty much in every test we ran (including WorldBench, PCWorld's official Windows benchmarking test). My tests involved file copy operations, unzipping, sleep/wake/start/shutdown tests, and gaming and Windows HD media playback.

The actual Excel spreadsheet I used to track each test's results will also be available for download, so you can see the nitty-gritty details if you like.

-rob.