Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Macbook Airs (11 and 13) both smoking the 13" 2.4Ghz Macbook Pro in gaming benchmarks

D

drvelocity

macrumors regular
Original poster
Oct 20, 2008
119
88
By far the most interesting benchmark trend coming out of the latest Macbook Air tests is that of the 320M GPU - is this thing somehow clocked differently than in the Macbook/Macbook Pro?

From:

http://www.macworld.com/article/155224/2010/10/macbookairbto_benchmarks.html

The latest Macbook Pro 13" 2.4 Ghz gets 33 FPS in Call of Duty 4, whereas the Macbook Air 13" (using the same 320M GPU) gets 40 FPS. Even the 1.4Ghz 11" gets 37! So obviously we're not talking about a CPU limited game - the only explanation then is that the GPU in the Macbook Air is clocked differently than in the 13" Macbook Pro, no?

From:

http://barefeats.com/mbpp30.html

Again the Macbook Air clocked at 2.13 ghz is beating the 2.4 Ghz Macbook Pro in World of Warcraft and Portal! And in WOW the 11" 1.4ghz still manages to beat the 2.4 Ghz 13" Macbook Pro.

Anyone have any additional insight into this? Anand did a terrible job of testing these for gaming performance, unfortunately, so he may not have even noticed this trend.
 
1BadMac

1BadMac

macrumors 6502
Jan 27, 2010
318
3
Would be curious to know if the 13" tested had the graphics update patch applied. The standard 2010 build did not include the optimum drivers that were released during the summer and are undoubtedly included in the new Airs by default.

Perhaps something to do with the SSD? I know my gaming performance increased once I went from HDD to SSD.
 
D

drvelocity

macrumors regular
Original poster
Oct 20, 2008
119
88
Ahh, right of course, I forgot about the graphics driver update. That could very well be the culprit. An SSD shouldn't make a difference in actual game FPS performance, unless your ram is low to the point where the game has to swap out to the disk all the time.
 
1BadMac

1BadMac

macrumors 6502
Jan 27, 2010
318
3
There is always "some" swapping occurring regardless of RAM consumption, albeit not extreme. However, any write to an ssd is much faster then hdd, so i do think ssd can have some oversell impact on gaming / fps performance. If the 13" had separate memory for video, I would not say that, but since it shares the same memory / bus as everything else, it could have a slight impact. Will definitely be to see additional tests and data.
 
D

drvelocity

macrumors regular
Original poster
Oct 20, 2008
119
88
1BadMac said:
Would be curious to know if the 13" tested had the graphics update patch applied. The standard 2010 build did not include the optimum drivers that were released during the summer and are undoubtedly included in the new Airs by default.

Perhaps something to do with the SSD? I know my gaming performance increased once I went from HDD to SSD.
Click to expand...

Well I just got an email reply from the owner of the site:

"Yes. I double checked and even ran the SL graphics update again. Then re-ran the tests. Same numbers. One theory is that because the MBP 2.4's fans only spin up to 3500 rpm, the GPU is downclocked if that doesn't keep it cool enough. I've seen this in earlier MBPs but had to boot Windows to prove it.

In the case of the MBA, the fans spin up to 6500 rpm when it gets hot. That allows it to run at higher clock speeds even when warm. But it also makes it noisier."

Very interesting!
 
M

Maven1975

macrumors 6502a
Aug 24, 2008
985
222
I can report that under bootcamp (Win 7 64), the 11" 1.6 can play COD MW2 at native resolution (no AA or shaders) without issue.

However, Dolphin EMU needs more juice than the 1.6, those with the 13" 2.13 will be fine in that department.
 
robeddie

robeddie

Suspended
Jul 21, 2003
1,777
1,731
Atlanta
Skaterbasist said:
Well, you would hope any newer versions were better in any situation.
Click to expand...

Yes. But at the same time, these news MBA's seem to be chipping away at so many of these people's biases. The bias that CPU power is everything, the bias that says such a small laptop can only be used for surfing the web and word processing.

These new Airs are poking holes in theories that many have held for a couple decades now - that only the systems with the highest cpu clock speeds are worth a damn.
 
P

Perdification

macrumors regular
Sep 22, 2010
202
0
I don't see why apple would intentionally throttle down the GPU of the MBP due to heat, when the MBP is bigger than the MBA and has better heat dissipation capabilities. If the fan in the MBP couldn't cope and only runs at 3500 RPM, then why didn't they use a 6500 RPM fan instead, but throttled down the GPU? They wanted to give us killer graphics, didn't they? Then why put a 6500 RPM fan in a teeny MBA when you can do it in a MBP? It really puzzles me.
 
1BadMac

1BadMac

macrumors 6502
Jan 27, 2010
318
3
I'm not sure the fan speed has as much to do with cooling as we think. It's been a while since I've studied air flow, but a smaller fan can run faster but still not "move" as much air (CFM?). So due to design and how thin the fan had to be for the Air, it's quite possible they have to rev it up to 6500 RPM just to flow enough Air to cool the processor. When, in a MBP, 3500 RPM with a larger fan may move more air and thus be more effective at cooling. I could be wrong, but this is my theory anyways.

That still doesn't explain the throttling of the 320m. The 320m should be 48 cuda cores clocked at 450 mhz. I wonder if the 320m in the Air's are slightly higher? I wish OS X had something like Riva Tuner where we could find the core speed easily.
 
P

Perdification

macrumors regular
Sep 22, 2010
202
0
1BadMac said:
I'm not sure the fan speed has as much to do with cooling as we think. It's been a while since I've studied air flow, but a smaller fan can run faster but still not "move" as much air (CFM?). So due to design and how thin the fan had to be for the Air, it's quite possible they have to rev it up to 6500 RPM just to flow enough Air to cool the processor. When, in a MBP, 3500 RPM with a larger fan may move more air and thus be more effective at cooling. I could be wrong, but this is my theory anyways.

That still doesn't explain the throttling of the 320m. The 320m should be 48 cuda cores clocked at 450 mhz. I wonder if the 320m in the Air's are slightly higher? I wish OS X had something like Riva Tuner where we could find the core speed easily.
Click to expand...

Yeah, true. The amount of CFM of the fan matters more than RPM. We'll need to locate more sources of information before we can come to any conclusions.
 
barefeats

barefeats

macrumors 65816
Jul 6, 2000
1,058
19
1BadMac said:
Would be curious to know if the 13" tested had the graphics update patch applied. The standard 2010 build did not include the optimum drivers that were released during the summer and are undoubtedly included in the new Airs by default.

Perhaps something to do with the SSD? I know my gaming performance increased once I went from HDD to SSD.
Click to expand...

Hi from Bare Feats. The SL graphics patch was applied to the 13" MacBook Pro 2.4. I doubt the SSD would make any difference since there were no virtual memory pageouts during gaming. On the other hand, there is some logging going on. On the other hand, the 15" MacBook Pro had a regular HDD, too, and was not handicapped.

I think it has to do with different "clock down when it gets hot" algorithms.
 
1BadMac

1BadMac

macrumors 6502
Jan 27, 2010
318
3
Hey Barefeats,

Can you monitor the clock speed of the GPU when you are doing your tests?

I know I've used some terminal commands in the past that allow you to hook into the graphics driver to pull the clock speed settings. There is a GUI packaged up as well, just not sure if it works with the Air.

http://www.groths.org/?p=207

Would be nice to prove / disprove the theory of the GPU clockdown. Of all devices, I would suspect the Air 320m to clockdown before the MBP.
 
R

relativist

macrumors regular
Jan 13, 2009
179
0
With so many factors affecting graphics it's difficult to compare on a level playing field. Integrated graphics share system ram, so small things like cas latency settings can make a big difference. Also the MBA cpu is a different chip with 6mb cache. Another difference is the screen. Lastly, of course, the HDD is different. Would it be possible to benchmark a MBP with a SSD of comparable speed?

In regards to the video settings, I'm thinking people will lower setting before they lower the native resolution. As for WoW in particular, I cap my fps at 30, so my Macbook does not overheat, and have settings lowered quite a bit, lag is rarely an issue for me.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom