What Linux vs. OS X Performance Tests Should Be Run Next?

[dungeon]

And performance per euro if you can, because things are much different from that POV As most Europe does not use OS_X, and things like nVidia Titan X is even more overpriced here

Nvidia Geforce GTX TITAN-X Will Be Overpriced at $1300 MSRP in Europe

[peppercats]

Performance per dollar? I'm not sure what you mean by price per dollar.

woops yep
I shouldn't be posting when I should be working

[Guest]

Well, we can only test stuff that is common to both systems and exclude things like responsiveness (which is fairly hard to measure).
If we are comparing kernel performances, hdd performance tests actually make quite a bit of sense. You could also try some browser benchmarks,
since browsers are basically the same across platforms and only system level implementation details change.
Comparing gaming performance makes sense, but tests themselves should be carefully chosen (preferably games that released on same day for both linux and macosx).
If we have wanted to make linux look good, we could run some server tasks, but since mac is likely setup to deliver desktop experience,
it would be hardly fair comparisson (agressive scheduler optimized for responsiveness rather than troughput and increases priority of currently active window, something that no linux distro does at the moment).

Overall, I don't think that we can make any impartial comparison, we can make mac look good or we can make linux look good, just by choosing "right" tests.

[brownsr]

Lmde 2

I'd love to see what the new Mint LMDE2 on Debian Jessie (in RC) could do. Compositing on full screen windows should be turned off ('general' system settings) I would presume.

[r1348]

Drop tests.

[gens]

HINT is a good OS benchmark

[adakite]

First I have to admit that I never use the Phoronix bench test, so I'm not aware of what kind of test it's already implemented, but
what I'd like to see here is the following:

- OpenCL 1.2/2.0 (FP64) compute tests on BLAS/FFT stuffs like that, pure computing in double precision using CPU and the GPU part.
- The reason is that OpenCL is possibly better supported under OS X than on any other system, could be good to check that

- OpenGL 4.0, which is on the contrary less supported under OS X, specifically compared to Linux (I wouldn't be so sure with Windows)

- CLang compilation time


- And then some "Office" oriented stuff, I'm thinking of generating complex pdf from .tex or .ps sources and compare the cpu-time as well as the capability of each system to use (or not) the Adobe's standard (this is usually not the case under OS X, and I'd like to check if this has an impact).

Those are really specific, I would understand if you do not have time for that.

Thanks though

[wizard69]

I thought about this a lot and still don't know the answer!

The problem is I don't use a LINUX box in the same way that I use my Mac. I don't even use the VM contained LINUX in the same way.

I do believe that the standard graphics tests are in order but I'm not even sure Apple has updated the GPU drivers fully to support Broadwell. Developmental tools tests are in order as that is one area where both machines can get a work out. You could of course test productivity apps but I rarely use such under LINUX.

One place that testing might be helpful is with web oriented clients and server soultions. Along with this maybe testing the Python oriented end of the machine, or maybe more accurately the supplied scripting environments. Eventually I'd like to see iPython integrated into your test suite.

The fact of the matter is I don't expect Mac OS to excell at anything in particular.

[Thirsty]

Encryption Test

I would like to see more encryption tests.
Encryption is more and more important.
like AES 128, 256, SHA 1 2 256.

[mitcoes]

I qwould use the mac mini as base 100 or 1000 for benchmarking

Every benchmark you use has its different base
Years ago benchmarking had a base 100 or 1000 with IBM PC 8086
Mac Mini is a great product for establishing a modern base in 100 or 1000 and changing it every new Mac Mini launch
And making more understable and easy to read every benchmark
Also giving a weight for each test (as you like with a certain logic) to make a CPU score, a GPU score and a total score

This way NUCs pendrive computers, even intel based tablets and other compact devices, including chromebooks laptops Steam Machines, and even any CPU+GPU+MB combination would be easily compared that is what benchmarking is about.

Also Phoronix benchmarking total scores would be able to establish price/performance benchmarking and MB, CPU and GPUs GPU, CPU MB and even RAM and HDDs and SDDs and total rankings, in pages with tables that would score a lot of visits that will give you more funding