How Long Can a MBA Sustain Turbo Boost?

[echosystm]

Hi

Under typical room temperature conditions, how long could an MBA sustain Turbo Boost if both cores were being hit 100%? I can't seem to find test results for this anywhere.

I'm trying to gauge the performance of MBAs, but GeekBench scores are virtually useless; the tests only run for about a minute, which is not long enough to see how the machine performs over time.

Thanks

 

[theSeb]

I have done some testing on this on a 2011 MBA and turbo boost throttles back down once it reaches around 90 degrees C on the CPU core. Then it moves up a little bit and moves down in a repeating pattern

 

[echosystm]

I have done some testing on this on a 2011 MBA and turbo boost throttles back down once it reaches around 90 degrees C on the CPU core. Then it moves up a little bit and moves down in a repeating pattern

Thanks. This is the kind of thing I was hoping to hear. How long did it take to hit 90 though? Were you definitely hitting both cores at 100% and how were you doing so?

 

[theSeb]

Thanks. This is the kind of thing I was hoping to hear. How long did it take to hit 90 though? Were you definitely hitting both cores at 100% and how were you doing so?

Both cores were hitting 100% - load was sustained using both Handbrake encoding a BR mkv rip and using the command line

yes > /dev/null

(obviously you need to run at least two of these to hit the load on both cores)

It hits 90 pretty quickly, but I cannot tell you how long because I did this a while ago and I honestly don't remember. Perhaps a minute of sustained load?

I also monitored the turbo boost during a Geekbench test and you're right that it is useless. Even running it over and over temperatures never went up and Turbo boost never throttled. I remember taking screenshots of all of this, but I am not sure where they are. I'll check on the MBA when I get home tomorrow. I actually did all of this to write an article, but I could only monitor the turbo boost on the MBA and I was trying to do it on the rMBP. Unfortunately using the same techniques resulted in kernel panics on the rMBP so I binned the article.