Newest Mac OS X 10.4.8 with Faster OpenGL? [Updated]

[MacRumors]

http://www.macrumors.com/images/macrumorsthreadlogo.gif

Apple just released a Mac OS X 10.4.8 developer seed marked as build 8L2112 (Intel) and 8L112 (PPC). This latest build was released to developers with little fan fare.

An unconfirmed post claimed the latest 10.4.8 build did incorporate enhanced OpenGL performance which led to the following Cinebench that does not show significant improvements:

Update: Apologies, it appears the benchmarks are not significantly better. Updated numbers seen. Improvements may have been gradual over 10.4.x updates or simply variability in the benchamrks

[arn]

can someone with a Macbook Pro 2.16, 2GB RAM, Mac OS X 10.4.7 run cinebench and post their results?

arn

[Eidorian]

I'm waiting for 10.4.8 Apple....

[EricNau]

I'm surprised Apple didn't just wait for Leopard.

...But I'm glad they didn't!

[nitynate]

C'mon apple!


I need more rosetta support.


As it is, it sucks.


Woot for Multithreaded OpenGL...

[arn]

What does this mean for PPC users such as myself?

It means if you have a multi-core or multi-processor Mac, then you will see similar speed benefits.

Remember that multi-core/multi-processor PowerPC macs where limited to PowerMacs, however.

arn

[mkjellman]

while these are impressive speed increases, i am not sure how impressive it would be in real life. does any part of the native os x gui or core elements use OpenGL?

the real question is would this increase the "snappyness"' of the next release?

[Eidorian]

Originally Posted by mkjellman

while these are impressive speed increases, i am not sure how impressive it would be in real life. does any part of the native os x gui or core elements use OpenGL?

the real question is would this increase the "snappyness"' of the next release?

Let me try not to die from the laughter.

http://www.apple.com/macosx/features/opengl/

Oh and yes you're going to see a performance increase on integrated graphics systems with multiple-cores.

[skoker]

Originally Posted by arn

It means if you have a multi-core or multi-processor Mac, then you will see similar speed benefits.

Remember that multi-core/multi-processor PowerPC macs where limited to PowerMacs, however.

arn

Well, guess my iBook G4 and in-the-mail Clamshell, PPC mini, and 12" PB G4 are left out of this party.

[xPismo]

Am I reading that correctly to say its a 4x to 7x improvement?

Thats staggering.

[arn]

Originally Posted by xPismo

Am I reading that correctly to say its a 4x to 7x improvement?

Thats staggering.

No, it's a 2x improvement if anything. I still need someone to run it on a latest 10.4.7 Mac Book Pro with 2.16GHz CPU and 2GB of ram if possible to verify results.

edit: MacBook Pro

arn

[skoker]

Originally Posted by arn

No, it's a 2x improvement if anything. I still need someone to run it on a latest 10.4.7 Mac Book with 2.16GHz CPU and 2GB of ram if possible to verify results.

arn

Are you looking for a MacBook or a MacBook Pro? I don't think there is a 2.16 GHz MacBook

[Animaniac]

Would this result in a Quartz Extreme speed up?

[Sol]

I really hope this update will double the graphics performance of my dual G4 PowerMac. Ideally games will not have to be updated to take advantage of 10.4.8's Open GL improvements.

[OwlBoy]

Originally Posted by Sol

I really hope this update will double the graphics performance of my dual G4 PowerMac. Ideally games will not have to be updated to take advantage of 10.4.8's Open GL improvements.

The way blizzard is talking, yes they do have to be updated.

-Owl

[sluthy]

Certainly looks like it's there, almost exactly 2x everything. Very cool, since OpenGL is used just about everywhere. Things are looking up for my hopefully imminent MB/MBP purchase in Jan/Feb maybe. Will Vista have multithreaded OpenGL or anything similar (is DirectX multithreaded)?

[arn]

I got some strange results from another MacBook Pro benchmark

2.0GHz, 2GB RAM, 10.4.7:

Rendering (Single CPU): 300 CB-CPU
Rendering (Multiple CPU): 569 CB-CPU

Multiprocessor Speedup: 1.90

Shading (CINEMA 4D) : 342 CB-GFX
Shading (OpenGL Software Lighting) : 1288 CB-GFX
Shading (OpenGL Hardware Lighting) : 2339 CB-GFX

OpenGL Speedup: 6.84

Still not as fast as the 10.4.8 #s, but not that bad. Until I see some more benchs, I pulled the story off the front page. Need more benchmarks...

arn

[OwlBoy]

Originally Posted by sluthy

Certainly looks like it's there, almost exactly 2x everything. Very cool, since OpenGL is used just about everywhere. Things are looking up for my hopefully imminent MB/MBP purchase in Jan/Feb maybe. Will Vista have multithreaded OpenGL or anything similar (is DirectX multithreaded)?

DirectX has had it for awhile.

-Owl

[sluthy]

Originally Posted by Sol

Ideally games will not have to be updated to take advantage of 10.4.8's Open GL improvements.

Originally Posted by Owlboy

The way blizzard is talking, yes they do have to be updated.

Originally Posted by Macrumors

Some code modification is reportedly required to see the full benefits of the multi-threaded OpenGL.

Hopefully it will just be a simple small download patch rather than a complete rewrite. Hopefully this OpenGL speedup will encourage more native Mac game development.

[sluthy]

Originally Posted by OwlBoy

DirectX has had it for awhile.

Bugger.

So hang on, why do gamers insist there's not much improvement to be had by dual cores yet if DX is multithread aware? How long have the developers had to rewrite their code?

[OwlBoy]

Originally Posted by sluthy

Hopefully it will just be a simple small download patch rather than a complete rewrite. Hopefully this OpenGL speedup will encourage more native Mac game development.

Yeah, from what I can tell World of Warcraft already has some of the stuff put in, if not all. And is just sitting idle untill OpenGL is patched.

Blizzard pushing on Apple for help to make WoW faster is part of the reason these large OpenGL changes are coming along.

-Owl

[melchior]

Just for fun here are my Cinebench frm 10.4.7

CINEBENCH 9.5
****************************************************

Tester : censored

Processor : Hackintosh
MHz : 2130 @ 3280
Number of CPUs : 2
Operating System : 10.4.7

Graphics Card : X1800 XT
Resolution : 1600 x 1200
Color Depth : 24 bit

****************************************************

Rendering (Single CPU): 540 CB-CPU
Rendering (Multiple CPU): 1013 CB-CPU

Multiprocessor Speedup: 1.88

Shading (CINEMA 4D) : 641 CB-GFX
Shading (OpenGL Software Lighting) : 2497 CB-GFX
Shading (OpenGL Hardware Lighting) : 5177 CB-GFX

OpenGL Speedup: 8.08

****************************************************

[marcosc]

Hi there,

My first post, with results.

Cinebench ran on MacBook Pro 15" 2.16GHz, 2GB Ram, X1600 256mb card, with OS X 10.4.7:

CPU (1 CPU): 329CB
CPU (2 CPU): 626CB
Multiprocessor Speedup: 1.90x

C4D Shading: 385CB
OpenGL SW-L: 1476 CB
OpenGL HW-L: 2828 CB
OpenGL Speedup: 7.35x

Double-checked results, they are accurate. It would appear that your results for 10.4.7 are pretty inaccurate. The real gain is practically negligible. Also, note that I'm running the MacBook on clamshell mode connected to a 30 ACD, so I'm pretty sure the number of pixels powered influences the results a bit.

Later,
M.

[nertzy]

CINEBENCH 9.5
****************************************************

Tester : Grant Hutchins

Processor : Intel Core Duo
MHz : 2.16 GHz
Number of CPUs : 2
Operating System : Mac OS X 10.4.7

Graphics Card : ATI Radeon X1600
Resolution : 1680 x 1050
Color Depth : 32-bit Color

****************************************************

Rendering (Single CPU): 328 CB-CPU
Rendering (Multiple CPU): 559 CB-CPU

Multiprocessor Speedup: 1.70

Shading (CINEMA 4D) : 381 CB-GFX
Shading (OpenGL Software Lighting) : 1503 CB-GFX
Shading (OpenGL Hardware Lighting) : 2947 CB-GFX

OpenGL Speedup: 7.74

****************************************************

[arn]

ok...

I suppose there have been speed increases along the way from the first 10.4.x results that where posted.

So, no substantial increase in speed I suppose. Also, I verified that the Multicore OpenGL speedups do require code modification, so it's unlikely to be the cause of any significant speed up on these benchmarks.

thanks. will update to a Page 2 story. Sorry for the false alarm.

arn