I'm a heavy number cruncher. Should I even consider the Air?

[Scott6666]

I ran this on my week 3 MBA 1.8 SSD for an hour. Highest temp I saw was CPU of 82C, most of the time it was about 80C. Case was slightly warm on the bottom back. No core shutdowns or problems, fans were at 6200 rpm most of the time but slowed after a couple of minutes.

I also ran it on a new MBP 2.6GHz. Highest CPU I saw was 67C. The dual fans stayed at their default of 2000 rpm and never went up. Case temperature about the same as on the MBA.

I would have expected the MBA to be a lot closer to MBP performance, especially a chip with only 20W TDP vs 35W. I have come to the conclusion that I am wrong. The MBA's design is such that it needs a fair amount of fan to dissipate the heat of the CPU or GPU given it's heatsink design. I could have preferred different but I think this is what is so.

I firmly believe there isn't any general heat problem on the MBA but can be broken down into 3 categories...

You can add a fourth category: people with incorrect expectations.

Many of those in category #1 are actually in category #4.

 

[bcaslis]

That's almost identical to my results w/o an external monitor.

Did you have an external monitor plugged in?

That was without an external monitor. Sorry but I don't have one to test with.

 

[bcaslis]

I would have expected the MBA to be a lot closer to MBP performance, especially a chip with only 20W TDP vs 35W. I have come to the conclusion that I am wrong. The MBA's design is such that it needs a fair amount of fan to dissipate the heat of the CPU or GPU given it's heatsink design. I could have preferred different but I think this is what is so.


You can add a fourth category: people with incorrect expectations.

Many of those in category #1 are actually in category #4.

I do think the MBA needs a higher fan speed. But it also has only one fan while the MBP has two fans so it's probably getting better circulation at lower speeds. Finally, the MBP I tested was the new revision with the cooler chip. I don't have my previous MBP anymore but I can guarantee that the fans would have been screaming on it. Considering the MBA has the same generation chip as that, I'm not surprised or disappointed in the MBA performance. I will say I'm very impressed by the new 45nm chip in the new MBP, it definitely runs cooler with better battery performance than the previous generation. The MBA would definitely benefit this chip but I don't think it's available in the smaller packaging the MBA uses yet.

 

[Adam Ierymenko]

You can add a fourth category: people with incorrect expectations.

Many of those in category #1 are actually in category #4.

Sorry, but I am of the expectation that any computer should be able to reach and sustain its advertised computational throughput under normal operating conditions: a room temperature environment and adequate ventilation.

My Air fulfills this as long as no external monitor is connected. The Air isn't designed to be a desktop and I have another machine that I use as such, so this isn't a big deal.

Anyone who has an Air that can't sustain full throughput under normal conditions (and without an external monitor) almost certainly has a defective one. Mine does not overheat as easily as some people on here seem to be saying theirs do.

 

[bcaslis]

Sorry, but I am of the expectation that any computer should be able to reach and sustain its advertised computational throughput under normal operating conditions: a room temperature environment and adequate ventilation.

My Air fulfills this as long as no external monitor is connected. The Air isn't designed to be a desktop and I have another machine that I use as such, so this isn't a big deal.

Anyone who has an Air that can't sustain full throughput under normal conditions (and without an external monitor) almost certainly has a defective one. Mine does not overheat as easily as some people on here seem to be saying theirs do.

I basically agree with you but I think there are people who do have incorrect expectations. There was a thread on the Apple forums about people complaining about the overheating. One of those then said he had found something interesting, that when placed on a desk the MBA worked correctly. Although he never stated it I can guarantee you he was running it on a soft surface like a bed or couch. If you do that the vents are blocked and it won't cool adequately. In my view that's an incorrect expectation (actually I have a different word in mind, it's just not nice to say it ).

 

[Phil A.]

Wirelessly posted (iPhone 16GB: Mozilla/5.0 (iPhone; U; CPU like Mac OS X; en) AppleWebKit/420.1 (KHTML, like Gecko) Version/3.0 Mobile/4A102 Safari/419.3)

I ran this on my week 3 MBA 1.8 SSD for an hour. Highest temp I saw was CPU of 82C, most of the time it was about 80C. Case was slightly warm on the bottom back. No core shutdowns or problems, fans were at 6200 rpm most of the time but slowed after a couple of minutes.

That's almost identical to my results w/o an external monitor.

Did you have an external monitor plugged in?

mine functions identically with and without my 23" ACD plugged in. I always leave the lid open though to use the screen: did you have the lid open or closed?

 

[Adam Ierymenko]

Wirelessly posted (iPhone 16GB: Mozilla/5.0 (iPhone; U; CPU like Mac OS X; en) AppleWebKit/420.1 (KHTML, like Gecko) Version/3.0 Mobile/4A102 Safari/419.3)



mine functions identically with and without my 23" ACD plugged in. I always leave the lid open though to use the screen: did you have the lid open or closed?

Open.

Are you using VGA or DVI?

 

[hotdamn]

Open.

Are you using VGA or DVI?

with a 23" ACD?
Seriously?
DVI of course, I dont think you can use a VGA with that.

 

[pv2b]

Sorry for the late response to this thread, but there is one obvious solution you're all missing.

The original poster says he needs a machine to number crunch, and that he already has a fast desktop computer to do it on. So -- when you need to crunch numbers, why don't you connect to your desktop computer over the Internet to perform number crunching? That way you can run a simulation for 8 hours or so, and you won't even have to have your Air powered on while you're waiting for your results. Sounds pretty convenient to me.

As for remote control -- I guess that depends on how your applications are written. If they're simple command line apps, you can use something like ssh (using the "screen" application to allow them to run in the background even after disconecting). If they're GUI apps you can use a remote desktop solution -- like VNC (free but slightly fiddly) or Apple's Back To My Mac (dead easy, but requires $99/year .Mac subscription).

I guess that's well within the spirit of the Air being a "wireless machine".