I was rather surprised at the poor graphics performance of Linux because Apple isn't known for high performance graphic. Honestly I was expecting Linux to do better,there. Probably should be some testing done with closed source drivers.
What I find most pleasing is the excellent performance of XCode with Clang. I would have to say the Apple should be very happy that a compiler that has barely matured does so well.
you right but its not the point if you do benchmarks it must be done in "neutral computer" you can't create benchmarks that the computer (mac) from apple and also the osx from apple its not "real" benchmarks or its not objective.
What exactly do you mean?
Michael is benchmarking the performance of Linux to the performance of OSX running on two examples of Apple hardware. True OSX is in its "native" territory and has optimizations in place to take advantage of the hardware, but Linux has other optimizations that Apple may not have considered. If benchmarks running Windows 7/8 or Solaris were included, they would have other optimizations that their respective publishers found important for the hardware at hand.
My Apple hardware is a pc like any other that can be purchased from HP, Asus, et al. Sure the hardware can be a little pricey, but I find it extremely rugged and capable for my needs. I do not need to limit myself to running OSX and contrary to popular belief, can run any OS that supports my hardware even eliminating OSX is I desire.
Articles like this are great to see and are the reason I keep visiting Phoronix. I see Linux as being a "neutral" platform for testing hardware, specifically, processors.
Last edited by heinsj24; 08-25-2012 at 03:48 PM.
Reason: Kudos for Phoronix
Linux wins at nearly everything but graphics related work. Just goes to show how far behind Linux drivers are compared to OS X.
Most people buy a computer to use for a) writing, b) browsing the net, c) mailing d) chat. Some insignificant number of users also do it for e) games and f) software development.
Graphics at 1080p is more than adequate, it is superb. Higher resolution buys very little to nothing.
If you are a gamer, than speed of graphics is important. Here it is important, but for very very few individuals.
Now look at the difference in costs. A standard laptop or desktop is about half the price of an Apple device for hardware.
With Linux, the software is free, and includes the a-f above, and if you need more, you may choose from thousands of free programs, from database to games to statistical packages.
The cost of ownership for a business or a student, over 4+ years does not make the Apple a winner.
Linux users are delighted with what they have. Apple users are delighted with what they have.
How reliable are tests with one and only one computer?
Normally, for statistically reliable results, you need between 18 (low side) and 20 (somewhat better) devices to use for comparisions.
Then you could say that you have consistant results with a 5% confidence interval.
Each laptop will have variances in components. Even the clock speeds may vary slightly, as would logic switching speeds. These would allow some laptops to favor some results and not others.
Therefore, you should take the Phoronix results as measure of results based on a single machine but not as a measure of a batch of machines, or as other than driver differences executing on that specific machine.