Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Just accept that no matter what Apple does, it won't be better than Mavericks.

2013 MacBook Air with Mavericks would have 13 hours of battery with regular use (last the whole day).

That being said, at least El Capitan is noticeably smoother than Yosemite which just crapped my Mac.

I think upgrading beyond El Capitan will be a huge mistake. Seems like the last straw for a 2 year old laptop.
 
Unigine Heaven 4.0 on a MacPro 5,1. Preset Extreme, quality Ultra.
a) Yosemite 15,2 fps.
b) El Capitan 11,6 fps and no sound.
El Capitan is slower.

You can't really compare Yosemite vs El Capitan if your benchmark software hasn't been updated to use Metal. Metal is where the vast majority of the performance increase comes from.

When El Capitan is meant to and runs off Metal you need to compare Yosemite on OpenGL vs El Capitan on Metal.
 
You can't really compare Yosemite vs El Capitan if your benchmark software hasn't been updated to use Metal. Metal is where the vast majority of the performance increase comes from.

When El Capitan is meant to and runs off Metal you need to compare Yosemite on OpenGL vs El Capitan on Metal.

Its apples and oranges. Yeah, Metal is faster. Comparing OpenGL implementations still makes perfect sense, especially as all games are still using OpenGL.
 
Holy ****. What mac do you have?
it says right on the screenshot :)

Anyway, my 2007 iMac is a lot better on El Cap than it ever was on Yosemite.
Tried going back on mavericks (which seemed a little bit faster and smoother than yosemite) a couple of months ago but can't use it, need new yosemite features.
 
Holy ****. What mac do you have?

Per my sig it's a "2012 cMBP 15"/2.6Ghz i7 Quad/16GB/hi-res antiglare/480SSD/750HDD/OS X 10.11.0".

It is a very fast one though! The highest scoring 2012 2.6GHz model on the Geekbench site is less than 100 points faster, so I guess I lucked out.
 
Its apples and oranges. Yeah, Metal is faster. Comparing OpenGL implementations still makes perfect sense, especially as all games are still using OpenGL.

But people are comparing the OS's performance (Marvericks/Yosemite vs El Capitan.) So right off the bat they are swaying the "test" in favour of the older OS because they aren't even testing El Capitan properly. El Capitan is built to run off Metal and only use OpenGL for the unfortunate on a 2011 or older computer. If you're testing the performance between an older OS to the brand new one you need to have the new being tested running what they were designed to run (Metal.)
 
But people are comparing the OS's performance (Marvericks/Yosemite vs El Capitan.) So right off the bat they are swaying the "test" in favour of the older OS because they aren't even testing El Capitan properly. El Capitan is built to run off Metal and only use OpenGL for the unfortunate on a 2011 or older computer. If you're testing the performance between an older OS to the brand new one you need to have the new being tested running what they were designed to run (Metal.)

This is nonsense. OpenGL and Metal are two different APIs that coexist peacefully.
 
This is nonsense. OpenGL and Metal are two different APIs that coexist peacefully.

Sigh.

You're missing the point entirely.

El Capitan is designed to run off Metal (it only uses OpenGL on older unsupported GPU's)

If you are testing the performance OF the OS between an older one and the new El Capitan. The test and comparison HAS to include Metal. That is where El Capitan gets nearly all its performance increase. Thus you are not getting the full result of what El Capitan is capable of.

That's like having two runners race and telling one he is only allowed to walk.
 
  • Like
Reactions: SRLMJ23
Sigh.

You're missing the point entirely.

El Capitan is designed to run off Metal (it only uses OpenGL on older unsupported GPU's).
No, it runs off openGL on all GPUs, new and old. In fact, Metal isn't used at all except for intel GPUs, since no Metal app exists yet.
 
Sigh.

You're missing the point entirely.

El Capitan is designed to run off Metal (it only uses OpenGL on older unsupported GPU's)

If you are testing the performance OF the OS between an older one and the new El Capitan. The test and comparison HAS to include Metal. That is where El Capitan gets nearly all its performance increase. Thus you are not getting the full result of what El Capitan is capable of.

That's like having two runners race and telling one he is only allowed to walk.

Metals Core graphics and core animation are only available for Intel graphics.
 
No, it runs off openGL on all GPUs, new and old. In fact, Metal isn't used at all except for intel GPUs, since no Metal app exists yet.

The OS El Capitan runs on Metal if you have a 2012 and up iMac, MacBook, MacBook Pro. It supports 7000 series and up AMD cards (I forget the cut off on Nvidia) and I think Intel 4000 HD and up. Anything below those cut offs runs off OpenGL.

It's true games need to be written to support it. Some apps can if they used CoreAnimation already.

But people are benchmarking the OS. It's not a fair comparison to benchmark El Capitan gimped on OpenGL when it is designed to use Metal. So if the benchmark software isn't written to utilize Metal to test El Capitan, then those scores are well below what the OS is actually capable of.
 
Since when? Early in the beta it was proven that 7000 series and up AMD cards were supported. Same for Nvidia (forget the cut off point)

http://arstechnica.com/apple/2015/09/os-x-10-11-el-capitan-the-ars-technica-review/7/

"Even if your Mac is on the support list, there’s a chance that it doesn’t use all of Metal’s advertised features. For example, Apple tells us that the Core Graphics and Core Animation acceleration is available only on Macs that use Intel integrated graphics exclusively. It’s not available on higher-end Macs with dedicated GPUs or in laptops that switch dynamically between integrated and dedicated graphics. The Intel GPUs are the ones that need the most help driving OS X’s UI, but it still seems a bit odd. That said, all of the Macs on the support list can use Metal for gaming and GPGPU tasks, integrated or dedicated."
 
Here is another test: XBench
Results 175.52 Yosemite
System Info
Xbench Version 1.3
System Version 10.10.5 (14F27)
Physical RAM 16384 MB
Model MacPro5,1
Drive Type OWC Mercury Accelsior PCIe SSD
CPU Test 239.14
Memory Test 484.35
Quartz Graphics Test 297.40
OpenGL Graphics Test 232.71
User Interface Test 54.28
Disk Test 539.03
Results 88.67 El Capitan
System Info
Xbench Version 1.3
System Version 10.11 (15A284)
Physical RAM 16384 MB
Model MacPro5,1
Drive Type OWC Mercury Accelsior PCIe SSD
CPU Test 239.54
Memory Test 514.86
Quartz Graphics Test 181.64
OpenGL Graphics Test 245.31
User Interface Test 19.82
Disk Test 661.28
 
http://arstechnica.com/apple/2015/09/os-x-10-11-el-capitan-the-ars-technica-review/7/

"Even if your Mac is on the support list, there’s a chance that it doesn’t use all of Metal’s advertised features. For example, Apple tells us that the Core Graphics and Core Animation acceleration is available only on Macs that use Intel integrated graphics exclusively. It’s not available on higher-end Macs with dedicated GPUs or in laptops that switch dynamically between integrated and dedicated graphics. The Intel GPUs are the ones that need the most help driving OS X’s UI, but it still seems a bit odd. That said, all of the Macs on the support list can use Metal for gaming and GPGPU tasks, integrated or dedicated."

That's really lame if that is true. It does say "chance" though :p

Still doesn't explain why newer systems (in the 2012 and up requirement) with dedicated GPU's see a much bigger gain than 2011 and down. (Slight gain but very very small, at least on my machines.)

But if that's the case, then Metal at least for now is pretty crap and barely a feature to a huge percent of users.
 
Here is another test: XBench
Results 175.52 Yosemite
System Info
Xbench Version 1.3
System Version 10.10.5 (14F27)
Physical RAM 16384 MB
Model MacPro5,1
Drive Type OWC Mercury Accelsior PCIe SSD
CPU Test 239.14
Memory Test 484.35
Quartz Graphics Test 297.40
OpenGL Graphics Test 232.71
User Interface Test 54.28
Disk Test 539.03
Results 88.67 El Capitan
System Info
Xbench Version 1.3
System Version 10.11 (15A284)
Physical RAM 16384 MB
Model MacPro5,1
Drive Type OWC Mercury Accelsior PCIe SSD
CPU Test 239.54
Memory Test 514.86
Quartz Graphics Test 181.64
OpenGL Graphics Test 245.31
User Interface Test 19.82
Disk Test 661.28

Do you have any screenshots yet?

I question that last test too. Everything is higher in El Capitan except quartz graphics test and user interface. And user interface appears to be showing the most noticeable improvements in real life.
 
That's really lame if that is true. It does say "chance" though :p

Still doesn't explain why newer systems (in the 2012 and up requirement) with dedicated GPU's see a much bigger gain than 2011 and down. (Slight gain but very very small, at least on my machines.)

But if that's the case, then Metal at least for now is pretty crap and barely a feature to a huge percent of users.

I agree. Its unlikely I'll even own my current Mac by the time any noteworthy game dev uses it.

Last I checked with the devs of Elder Scrolls Online, Zenimax they had no plans whatsoever to support it in any updates. From my interpretation it even further limited an already limited customer base since all Macs supported OpenGL. And the obvious was the amount of work required to code for it.

I love to see a game update for it just to see the improvements.
 
Here is another test: XBench
Results 175.52 Yosemite
System Info
Xbench Version 1.3
System Version 10.10.5 (14F27)
Physical RAM 16384 MB
Model MacPro5,1
Drive Type OWC Mercury Accelsior PCIe SSD
CPU Test 239.14
Memory Test 484.35
Quartz Graphics Test 297.40
OpenGL Graphics Test 232.71
User Interface Test 54.28
Disk Test 539.03
Results 88.67 El Capitan
System Info
Xbench Version 1.3
System Version 10.11 (15A284)
Physical RAM 16384 MB
Model MacPro5,1
Drive Type OWC Mercury Accelsior PCIe SSD
CPU Test 239.54
Memory Test 514.86
Quartz Graphics Test 181.64
OpenGL Graphics Test 245.31
User Interface Test 19.82
Disk Test 661.28
I too tried Xbench. I don't think it tells much. The user interface test went from 400 to 20 or something (5 times slower than the reference PPC G5), between 10.9 and 10.11. It's cleary bogus. Quartz tests should be more reliable, and strangely, numbers are quite a bit lower for 10.11. I don't think it maters as long as 10.11 is fluid.
 
I went from 10.9 to 10.11, as 10.10 never worked properly for me. I haven't run any benchmarks but el cap seems faster, if anything. Late 2013 13" rMBP.
 
I agree. Its unlikely I'll even own my current Mac by the time any noteworthy game dev uses it.

Last I checked with the devs of Elder Scrolls Online, Zenimax they had no plans whatsoever to support it in any updates. From my interpretation it even further limited an already limited customer base since all Macs supported OpenGL. And the obvious was the amount of work required to code for it.

I love to see a game update for it just to see the improvements.

I know Blizzard is going to add Metal starting with WoW (Likely the new expansion that's going to beta later this year will be written for it.) And I'm sure their new game Over Watch will as well since that is soon to be in beta as well. Since they are a supporter of it, I'm sure they will include it in Starcraft and Diablo III eventually.
 
OpenGL performance in C4D
Bildschirmfoto 2015-10-03 um 07.01.41.png


CPU Performance from 10.8 to 10.11
Bildschirmfoto 2015-10-03 um 06.58.45.png


10.10 was the slowest. Hanging Finder, when open folders with many many files.

10.11 is much snappier in the Finder.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.