Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Were you using the latest intel driver? Others have said that had increased performance. Thanks for the results. I've been following this forum.

I have an Alienware m11x R2 that just benched 7083 on 3dmark06. It has an Nvidia GT 335m. It just seems like the score for the 2013 MBA should be a bit more to me. I could be dreaming too because I want the new Air so bad.

Thanks again to you and others for testing these out.

I agree! Maybe notebookcheck mistakenly rated the HD 5000 just above the 540M which was better than the 335M. I think the 540M got around 8000 in 3Dmark06. Is it possible the hd5000 is underclocked?
 
the Intel HD4600 gets 7700-8800 in 3dmark06.
http://www.notebookcheck.net/Intel-HD-Graphics-4600.86106.0.html

So the hd5000 should be higher. Id firstly make sure latest drivers from intel. And secondly make sure the power profile is set to max performance and not max battery/balanced etc.

Im eager to see the results. Pretty much depending on this thread weather I pickup a new MBA or stick with my 2010 MBA for another year.
 
For 3DMark06, Intel expects the HD 5000 to be 20% faster than the HD 4000, Core i7 to Core i7. That gets to 50% in 3DMark11.

By the way, the Macbook Air 2012 13-inch gets 6000 points in 3DMark06 according to Notebookcheck.
 
I agree! Maybe notebookcheck mistakenly rated the HD 5000 just above the 540M which was better than the 335M. I think the 540M got around 8000 in 3Dmark06. Is it possible the hd5000 is underclocked?

The TDP of the chips in the macbook air is very low, so don't expect miracles from it. I highly doubt that it would get anywhere near the 540M in benchmarks and especially real games.

BTW, macworld tested the new airs on Portal 2 and it was less than 10% faster than the 2012 models. Most CPU benchmarks were lower. It did redeem itself in the storage speed benchmarks, thanks to the PCIe SSDs, so overall the macworld speedmark was the same.

http://www.macworld.com/article/204...efits-from-faster-graphics-flash-storage.html
 
Last edited:
Just ran 3dmark06 but the bloody thing only shows you results online and I can't find a driver for the network adapter, has anyone had this problem in Windows 8?

If someone can provide a serial then I can rerun it.

The Ice Storm test ran at 60fps the whole way through, the Cloud one ran at 30fps the whole way through, and lol.. The Fire Strike test was a step too far, slideshow mode, about 5fps all the way through.

I did notice some CPU throttling throughout.
 
Just ran 3dmark06 but the bloody thing only shows you results online and I can't find a driver for the network adapter, has anyone had this problem in Windows 8?

If someone can provide a serial then I can rerun it.

The Ice Storm test ran at 60fps the whole way through, the Cloud one ran at 30fps the whole way through, and lol.. The Fire Strike test was a step too far, slideshow mode, about 5fps all the way through.

I did notice some CPU throttling throughout.

thanks for trying to run the program. 3dmark06 should be free ( or at least the version to give you the benchmarks ). can't give you a windows 8 serial though. sorry
 
thanks for trying to run the program. 3dmark06 should be free ( or at least the version to give you the benchmarks ). can't give you a windows 8 serial though. sorry

No no, 3dmark06 basic edition is free but it shows you the results via. it's website, but I haven't got a working driver for my network adapter. With a serial for the 3dmark06 advanced edition I can view the results offline.

If someone is going to try 3dnark06 again, please install HD 5000 drivers direct from Intel, turn all CPU and GPU settings to "macimum performance" (guide in the OP), disable Aero and allocate at least 1GB of SSD to virtual memory.
 
Last edited:
The TDP of the chips in the macbook air is very low, so don't expect miracles from it. I highly doubt that it would get anywhere near the 540M in benchmarks and especially real games.

BTW, macworld tested the new airs on Portal 2 and it was less than 10% faster than the 2012 models.

The macbook air haswell still has a full blown hd5000 running at 1000-1100mhz with 40 EU's. i suspect these early benchmarks aren't pushing the hd5000 to its full potential.

The portal 2 benchmarks linked
Don't even mention settings :mad: i suspect the results are cpu bound?

I wish anand would hurry up!
 
The macbook air haswell still has a full blown hd5000 running at 1000-1100mhz with 40 EU's. i suspect these early benchmarks aren't pushing the hd5000 to its full potential.

The portal 2 benchmarks linked
Don't even mention settings :mad: i suspect the results are cpu bound?

I wish anand would hurry up!

You can get some description of speedmark 8 here. i.e. it is run at 1280x800, so I doubt it is really CPU bound.

http://www.macworld.com/article/2013458/measure-mac-performance-with-speedmark-8.html

And yes, they do have 40EU, but they are not running at 1000-1100mhz. The i5-4250U is running at 200-1000Mhz. i.e. when thermal temperatures allow, it can go up to 1000Mhz, but if it doesn't allow, it can be much less than 1000Mhz.

http://ark.intel.com/products/75028/
 
Slightly gutted about that! I'm really fond of 7. Thanks for the heads up, I'll definitely go with a stripped-down version of 8.

Honestly don't even bother with Windows 7, as far as I can tell its not compatible with the new MBAs, I wasted so many hours on it..

Windows 8 is the way to go, just delete Metro and all it's apps and it runs very smooth.

Windows 7 is supposed to work. It sounds like a driver issue.

http://support.apple.com/kb/TS4517?viewlocale=en_US&locale=en_US

Note that I had a similar problem when the 2012 MacBook Airs came out. I had to use a workaround to get Boot Camp working with the internal keyboard and trackpad.
 
You can get some description of speedmark 8 here. i.e. it is run at 1280x800, so I doubt it is really CPU bound.

http://www.macworld.com/article/2013458/measure-mac-performance-with-speedmark-8.html

And yes, they do have 40EU, but they are not running at 1000-1100mhz. The i5-4250U is running at 200-1000Mhz. i.e. when thermal temperatures allow, it can go up to 1000Mhz, but if it doesn't allow, it can be much less than 1000Mhz.

http://ark.intel.com/products/75028/

Speedmark isnt a good a good indicator as it just lumps together all the sythetic tests of the system. More interested in exposing the hd5000 to see truley how much better it is to the hd4000. So far i have seen some games with doule the fps. Others with %10 better (portal 2).

What i ment with the hd5000 running at 1000-1100mhz was the i5 running it at 1000mhz, and the i7 running it at 1100mhz. Poorly conveyed i know :(-
Surely the macbook air cooling system is well specced to handle the 15watt tdp and should allow the gpu to run at full speed? If not this would explain some of the results like a measily %10 gain etc.
 
Updated OP with new results, still waiting reports on games like Crysis 1, 2 and 3 and Battlefield 3.
 
Just ran Minecraft under OS X at 1400x900, all video settings set to max (furthest draw distance, advanced open gl, "fancy" graphics, all particules etc. etc. everything max basically) and got 30-50fps. That includes in mines, in villages, on the tops of hills looking out over the world and during combat.

Not FPS drops visible to the human eye.

This is fullscreen and with Safari running in the background. 13" 4GB i5.

Awesome, thanks so much for the response! Looks like it can handle some light gaming really well!
 
Surely the macbook air cooling system is well specced to handle the 15watt tdp and should allow the gpu to run at full speed? If not this would explain some of the results like a measily %10 gain etc.

It has nothing to do with the macbook air cooling specs. The 15W TDP of the chip in the macbook air does not in most circumstances handle the GPU running at its top speed.

That's why Intel introduced the 28W Iris 5100 chips ULV Haswell chips. Exactly the same specs for the GPU (40EU), but has a much higher TDP (almost double) to allow the GPU to run close to its maximum speed. But this chip is of course too hot for the macbook air.
 
It has nothing to do with the macbook air cooling specs. The 15W TDP of the chip in the macbook air does not in most circumstances handle the GPU running at its top speed.

That's why Intel introduced the 28W Iris 5100 chips ULV Haswell chips. Exactly the same specs for the GPU (40EU), but has a much higher TDP (almost double) to allow the GPU to run close to its maximum speed. But this chip is of course too hot for the macbook air.

After an hour of gaming in Farcry 3 the fans are running but the computer itself is pretty cool, perhaps it has some overclock potential.
 
so the fact that the new zenbook infinity has the HD 5100 is because it is clocked higher, for laptops that can handle more heat dissipation?

how much better is the 5100 compared to the 5000 on the air anyway.
 
so the fact that the new zenbook infinity has the HD 5100 is because it is clocked higher, for laptops that can handle more heat dissipation?

how much better is the 5100 compared to the 5000 on the air anyway.

In numbers, the 5100 just has a 100-200MHz higher turboboost.
 
I obviously can't say without trying it out but I used to game on a MacBook Pro with a 9600M GT and even without overclocking that thing would pretty much burn a hole through the table, when on my lap I would seriously worry about my fertility lol, and I could overclock that to the tune of about 800Mhz...

This Air doesn't get to 1/3rd of that temperature while gaming so possibly it has the potential to be clocked up to 5100/5200 levels.

That said, I heard that there was virtually no benefit to overclocking the HD 4000 series, with gains of only 2-3 fps even at +400Mhz, so maybe we have nothing to gain.. and similarly, nothing to feel bad about, not getting those higher clocked cards.
 
Officially it should be like that:

iris_ultrabook_large.jpg


e: It's HD4000 vs HD5000 vs HD5100
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.