Jump to content


3D Game benchmarks for the MacBook Pro 2.6 "2008"


  • Please log in to reply
86 replies to this topic

#21 teflon

teflon

    Bastard of the Popeye Analogy

  • Members
  • PipPipPipPipPipPipPipPipPipPipPipPip
  • 9589 posts
  • Location:London, UK

Posted 12 March 2008 - 07:21 AM

seeing as its a UT2004 flyby, theres no game logic being run, which means theres a pretty minimal load on the CPU.. so yes, the GPU is clocked slightly higher, but itll only really be matching up to nV's maximum 8600M GT clocks, cos the ones in the 2007 MBPs were underclocked to start off with. I doubt theyll even be matching it yet though...
Polytetrafluoroethylene to my friends.

Macbook Pro - C2D 2.4Ghz / 4GB RAM / Samsung 830 256GB SSD / Geforce 8600M GT 256Mb / 15.4"
Cube - G4 1.7Ghz 7448 / 1.5GB RAM / Samsung Spinpoint 250GB / Geforce 6200 256Mb
Self-built PC - C2Q Q8300 2.5Ghz / 4GB RAM / Samsung 830 256GB SSD / Radeon 7850 OC 1GB / W7 x64
and a beautiful HP LP2475w 24" H-IPS monitor

#22 Quicksilver

Quicksilver

    Verbal Windbag

  • IMG Writers
  • 4227 posts
  • Location:Chicago Illinois
  • Pro Member:Yes

Posted 12 March 2008 - 09:10 AM

Define "no game logic."  There's something for the CPU to do while the flyby is happening, I assure you. :P
Former Senior Hardware Editor
InsideMacGames.com

#23 teflon

teflon

    Bastard of the Popeye Analogy

  • Members
  • PipPipPipPipPipPipPipPipPipPipPipPip
  • 9589 posts
  • Location:London, UK

Posted 12 March 2008 - 12:26 PM

well, no physics, no bot AI, etc. etc. etc. essentially just camera movement and rendering. Thats what I meant, which in turn means that itd be plenty under the amount of CPU usage for the CPU to not really affect the result... as evidenced by both 15" 2008 models getting the exact same result.
Polytetrafluoroethylene to my friends.

Macbook Pro - C2D 2.4Ghz / 4GB RAM / Samsung 830 256GB SSD / Geforce 8600M GT 256Mb / 15.4"
Cube - G4 1.7Ghz 7448 / 1.5GB RAM / Samsung Spinpoint 250GB / Geforce 6200 256Mb
Self-built PC - C2Q Q8300 2.5Ghz / 4GB RAM / Samsung 830 256GB SSD / Radeon 7850 OC 1GB / W7 x64
and a beautiful HP LP2475w 24" H-IPS monitor

#24 rob_ART

rob_ART

    Bare Feats

  • Members
  • PipPipPipPipPip
  • 792 posts
  • Location:Portland, OR

Posted 14 March 2008 - 08:43 PM

I agree that UT2004 is notoriously CPU bound during actual play or during a Botmatch benchmark. But the FlyBy benchmark is almost pure GPU.

As for specification on the 8600M GT, I'm not saying that Apple has overclocked the 2008 model. But the 8600M runs at below spec clock speeds except when 3D OpenGL apps are running. And when they do, the core and memory clock speeds ramp up depending on various factors including temperature. I've confirmed this with engineers who have worked on the MacBook Pro and I have measured the variance using tools under Windows XP Pro.

As for testing at 1440x900 with FSAA using Call of Duty 4 (or Crysis) under Windows XP or Vista, I'm all for that. If you and I and others on this forum continue to work together on this, we can all figure this all out.
rob-ART morgan
mad scientist
BareFeats.com

#25 rob_ART

rob_ART

    Bare Feats

  • Members
  • PipPipPipPipPip
  • 792 posts
  • Location:Portland, OR

Posted 15 March 2008 - 07:07 AM

FYI, when I run Cinebench 10 benchmark on the 2.6GHz 15" MacBook Pro "2008" compared to the 2.4GHz 15" MacBook Pro "2008," the 2.6 was 30% faster even though the core clock speed is only 8% faster.

Do you guys think that's the bigger L2 cache (6MB vs 3MB) kicking in?
rob-ART morgan
mad scientist
BareFeats.com

#26 Quicksilver

Quicksilver

    Verbal Windbag

  • IMG Writers
  • 4227 posts
  • Location:Chicago Illinois
  • Pro Member:Yes

Posted 15 March 2008 - 04:14 PM

View Postrob_ART, on March 15th 2008, 08:07 AM, said:

Do you guys think that's the bigger L2 cache (6MB vs 3MB) kicking in?

Yes, definitely.  Here's more results to back up your (correct) theory:

Posted Image

Don't forget about SSE4, which hasn't been implemented into either Mac OS X or Windows (or any apps) yet.  To see what kind of acceleration we'll eventually get, scroll down to the bottom of this ArsTechnica article.
Former Senior Hardware Editor
InsideMacGames.com

#27 rob_ART

rob_ART

    Bare Feats

  • Members
  • PipPipPipPipPip
  • 792 posts
  • Location:Portland, OR

Posted 15 March 2008 - 05:20 PM

So all you with MacBook Pros, per Quicksilver's suggestion, I propose we all run Prey again under Mac OS X at 1440 x 900 High or Ultra with 4X FSAA to stress the GPU even further than the first series of tests.

Then, we run Prey at that same setting under Windows XP Pro or Vista Ultimate (Boot Camp partition). I also have Crysis and Unreal Tournament III if you prefer those as a test case. If more of you have COD4 than those three games, I can get a copy of that. Let me know your ideas and preferences.
rob-ART morgan
mad scientist
BareFeats.com

#28 rob_ART

rob_ART

    Bare Feats

  • Members
  • PipPipPipPipPip
  • 792 posts
  • Location:Portland, OR

Posted 15 March 2008 - 05:27 PM

View Postteflon, on March 12th 2008, 11:26 AM, said:

well, no physics, no bot AI, etc. etc. etc. essentially just camera movement and rendering. Thats what I meant, which in turn means that itd be plenty under the amount of CPU usage for the CPU to not really affect the result... as evidenced by both 15" 2008 models getting the exact same result.

Back on UT2004 Flyby, I could use Hardware Monitor to compare CPU and GPU load during a Botmatch vs Flyby to document the relative loads.
rob-ART morgan
mad scientist
BareFeats.com

#29 rob_ART

rob_ART

    Bare Feats

  • Members
  • PipPipPipPipPip
  • 792 posts
  • Location:Portland, OR

Posted 15 March 2008 - 07:27 PM

I've been running benchmarks on Unreal Tournament 3 under Vista 64 on the Mac Pro. It has many flyby scenarios which would provide the GPU stress we want. Anandtech chose the Suspense Flyby because it was the most taxing of the three Flybys on the DEMO. Though they are now testing with the full Retail version, they continue to use the Suspense Flyby as their UT3 GPU test.

Even if you don't have the retail version of Unreal Tournament 3, you could download the DEMO and we could all run the Suspense Flyby at 1440x900. If you have an external LCD that does 1920x1200, we could run that, too.

What think ye?
rob-ART morgan
mad scientist
BareFeats.com

#30 teflon

teflon

    Bastard of the Popeye Analogy

  • Members
  • PipPipPipPipPipPipPipPipPipPipPipPip
  • 9589 posts
  • Location:London, UK

Posted 16 March 2008 - 01:58 AM

im up for that...
i dont have Prey on PC, just mac (its not the kind of game many would own on both)
Polytetrafluoroethylene to my friends.

Macbook Pro - C2D 2.4Ghz / 4GB RAM / Samsung 830 256GB SSD / Geforce 8600M GT 256Mb / 15.4"
Cube - G4 1.7Ghz 7448 / 1.5GB RAM / Samsung Spinpoint 250GB / Geforce 6200 256Mb
Self-built PC - C2Q Q8300 2.5Ghz / 4GB RAM / Samsung 830 256GB SSD / Radeon 7850 OC 1GB / W7 x64
and a beautiful HP LP2475w 24" H-IPS monitor

#31 Quicksilver

Quicksilver

    Verbal Windbag

  • IMG Writers
  • 4227 posts
  • Location:Chicago Illinois
  • Pro Member:Yes

Posted 16 March 2008 - 03:01 AM

View Postrob_ART, on March 15th 2008, 06:20 PM, said:

So all you with MacBook Pros, per Quicksilver's suggestion, I propose we all run Prey again under Mac OS X at 1440 x 900 High or Ultra with 4X FSAA to stress the GPU even further than the first series of tests.

Then, we run Prey at that same setting under Windows XP Pro or Vista Ultimate (Boot Camp partition). I also have Crysis and Unreal Tournament III if you prefer those as a test case. If more of you have COD4 than those three games, I can get a copy of that. Let me know your ideas and preferences.

I think that Call of Duty 4 (and Windows XP) is probably the best option.  OpenGL isn't a great way to test GPUs, since OpenGL drivers are nowhere near as good (or consistent) as their DirectX counterparts.  Furthermore, a game like Crysis is so demanding that its timedemo, which displays results down to the hundredth place, will probably not be precise enough to make a valid conclusion for this MacBook Pro question (the MBP will probably return low single-digit results on "high" at 1440x900).  While you can crank down the video quality, nobody in their right mind will actually play Crysis like that, so . . . to repeat myself: Call of Duty 4 is just about right.

A (probably unnecessary) note, and then a final word of warning: Remember what you're testing.  If you run Call of Duty 4 with everything maxed with 4X FSAA at 1440x900 resolution, then you're mostly evaluating the effect of doubling the VRAM on a card with a 128-bit memory bus.  If you don't see large gains in a test like that, don't be surprised.   Finally, I have the Steam version of Call of Duty 4, and it refuses to run every timedemo I throw at it.  While I would normally recommend the Steam version, if all you're looking to do is use the game as a benchmark, then a) bribe me to figure out what the problem is with the Steam version (I like donuts), or b) buy the standalone version.
Former Senior Hardware Editor
InsideMacGames.com

#32 maccer

maccer

    Notorious

  • Members
  • PipPipPip
  • 222 posts
  • Location:Sweden

Posted 16 March 2008 - 03:30 AM

Have you got any results for the new 2.4 GHz model yet? I might be ordering one early next week, but it would be nice to see how it compares with the middle model first :)

#33 rob_ART

rob_ART

    Bare Feats

  • Members
  • PipPipPipPipPip
  • 792 posts
  • Location:Portland, OR

Posted 16 March 2008 - 05:16 AM

View Postmaccer, on March 16th 2008, 02:30 AM, said:

Have you got any results for the new 2.4 GHz model yet? I might be ordering one early next week, but it would be nice to see how it compares with the middle model first :)

We have the 2008 2.4 model in our lab along with the 2.6 model. We don't have the 2.5, though.

What kind of comparisons did you have in mind?
rob-ART morgan
mad scientist
BareFeats.com

#34 rob_ART

rob_ART

    Bare Feats

  • Members
  • PipPipPipPipPip
  • 792 posts
  • Location:Portland, OR

Posted 16 March 2008 - 05:28 AM

View PostQuicksilver, on March 16th 2008, 02:01 AM, said:

I think that Call of Duty 4 (and Windows XP) is probably the best option.

I've ordered a copy of CoD4 so I can test it on our two MacBook Pros. I'd still like to see Unreal Tournament III results, too. It should be easy enough since a DEMO version exists for those who don't own it and a benchmarking tool is also available.
rob-ART morgan
mad scientist
BareFeats.com

#35 maccer

maccer

    Notorious

  • Members
  • PipPipPip
  • 222 posts
  • Location:Sweden

Posted 16 March 2008 - 06:20 AM

View Postrob_ART, on March 16th 2008, 12:16 PM, said:

We have the 2008 2.4 model in our lab along with the 2.6 model. We don't have the 2.5, though.
That's great, the 2.4 is enough for me.

View Postrob_ART, on March 16th 2008, 12:16 PM, said:

What kind of comparisons did you have in mind?
The usual game (Mac + Bootcamp/Windows) and productivity app benchmarks. Both CPU and GPU-intensive tests; After Effects CS3, Photoshop, Cinebench, Geekbench, UT2004, Prey, BF2142. It would be interesting to see if there's any difference when running Windows XP in Parallels Desktop 3.0, but I don't know what kind of benchmarks should be used in that caes. I'd like to see a comparison between the 2006 2,33 GHz model and the 2008 2,4 / 2,6 models. I don't want to buy a computer that's slower (I'm mainly thinking of the difference in cache size).

Regarding game benchmarks on MBP:s they should be run at 1440x900 if possible. If the framerates are so low that the game is close to being unplayable, the tests could be run at a lower resolution as well (such as 1280x800) to see if that's enough to make the game playable.

I'd definitely like to see UT 3 in Windows, since that's a bit too slow on my current MBP.

#36 teflon

teflon

    Bastard of the Popeye Analogy

  • Members
  • PipPipPipPipPipPipPipPipPipPipPipPip
  • 9589 posts
  • Location:London, UK

Posted 16 March 2008 - 07:27 AM

if were going to test UT3, then it should always be the demo. otherwise, people with the retail version will be wanting to update to v1.2, which, I believe, has some performance enhancements. Not to mention that the demo only goes up to level 3 textures...

and I dont personally fancy buying CoD4 till it manages to reach bargain bin prices... something it wont do for a while yet.
Polytetrafluoroethylene to my friends.

Macbook Pro - C2D 2.4Ghz / 4GB RAM / Samsung 830 256GB SSD / Geforce 8600M GT 256Mb / 15.4"
Cube - G4 1.7Ghz 7448 / 1.5GB RAM / Samsung Spinpoint 250GB / Geforce 6200 256Mb
Self-built PC - C2Q Q8300 2.5Ghz / 4GB RAM / Samsung 830 256GB SSD / Radeon 7850 OC 1GB / W7 x64
and a beautiful HP LP2475w 24" H-IPS monitor

#37 rob_ART

rob_ART

    Bare Feats

  • Members
  • PipPipPipPipPip
  • 792 posts
  • Location:Portland, OR

Posted 16 March 2008 - 09:26 AM

View Postteflon, on March 16th 2008, 06:27 AM, said:

if were going to test UT3, then it should always be the demo. otherwise, people with the retail version will be wanting to update to v1.2, which, I believe, has some performance enhancements. Not to mention that the demo only goes up to level 3 textures...

and I dont personally fancy buying CoD4 till it manages to reach bargain bin prices... something it wont do for a while yet.

Okay. I'll download the UT3 DEMO and install Windows XP 32 bit in my MBP's Boot Camp partition. That will give us a common denominator. After running the Suspense Flyby at 1440x900, we can compare our results to see if that teaches us anything about the relative performance of the different MacBook Pro models with different amounts of VRAM and different CPU types/speeds.
rob-ART morgan
mad scientist
BareFeats.com

#38 Quicksilver

Quicksilver

    Verbal Windbag

  • IMG Writers
  • 4227 posts
  • Location:Chicago Illinois
  • Pro Member:Yes

Posted 16 March 2008 - 10:50 AM

Try this for UT3.  I'll do it very soon; I'm on the verge of sending old MacBook Pro back.
Former Senior Hardware Editor
InsideMacGames.com

#39 teflon

teflon

    Bastard of the Popeye Analogy

  • Members
  • PipPipPipPipPipPipPipPipPipPipPipPip
  • 9589 posts
  • Location:London, UK

Posted 16 March 2008 - 11:12 AM

oh, and another thing, we should probably make sure we use the same Forceware drivers to get the best performance available. im not on 171.16 yet, but ill update to that later on, and get the UT3 demo and try it all again and shizzle.
Polytetrafluoroethylene to my friends.

Macbook Pro - C2D 2.4Ghz / 4GB RAM / Samsung 830 256GB SSD / Geforce 8600M GT 256Mb / 15.4"
Cube - G4 1.7Ghz 7448 / 1.5GB RAM / Samsung Spinpoint 250GB / Geforce 6200 256Mb
Self-built PC - C2Q Q8300 2.5Ghz / 4GB RAM / Samsung 830 256GB SSD / Radeon 7850 OC 1GB / W7 x64
and a beautiful HP LP2475w 24" H-IPS monitor

#40 rob_ART

rob_ART

    Bare Feats

  • Members
  • PipPipPipPipPip
  • 792 posts
  • Location:Portland, OR

Posted 16 March 2008 - 11:24 AM

View Postteflon, on March 16th 2008, 10:12 AM, said:

oh, and another thing, we should probably make sure we use the same Forceware drivers to get the best performance available. im not on 171.16 yet, but ill update to that later on, and get the UT3 demo and try it all again and shizzle.

Where is that listed on nVidia's site? It's not coming up on this page:
http://www.nvidia.co...aspx?lang=en-us
rob-ART morgan
mad scientist
BareFeats.com