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

TheBearman

macrumors 6502
Original poster
May 23, 2008
445
87
Cary, NC
Yea, I know another MBA temp thread, sorry. My problem with all the other threads is lack of repeatable user test to do an apples to apples comparison.

I wanted to do a test, report results so others could do the same and see if they found a problem with their Air. I will say that I can and have pushed the Air to complete shutdown, but the activity was something that I didn't originally buy the Air to perform. The reason I picked this thing up was to have a light portable system for travel. This it has done very well and I am quite happy with it.

So the test. The first thing I did was place the Air on a flat wood surface and made it a point to not wrap the power cord around the back and over the cooling vents. I started off with the Air cool and let it idle for 10 minutes with nothing running. The room temp was 72 F. I did verify the active processes using terminal. I found 46 active, all of which I would expect to be normal on the AIR. I am running 10.5.3, I have Coolbook but disabled it for the first set of test. The first column of numbers are those recorded from iStat Pro after the 10 minute warmup. I then ran a YouTube video which was in 5 parts. I'm sure any video would work but to keep this repeatable here is the link.

Dr Who Blink
(interesting if you haven't watched this before)

I also ran this at full screen to push it as much as possible. I used Safari since this is something everyone has available. After this completed I let it sit for 10 minutes to cool and the last column has that cool down temp.

.......Start....1....2....3....4....5..Post
CPU.......43...68...64...67...64...62....45
HS........39...59...53...56...56...59....43
Base 1....31...38...38...39...40...41....37
Base 2....31...38...38...39...40...41....37
Base 3....29...32...33...34...33...35....33
Airport...31...35...35...37...38...37....35
Power.....40...59...57...58...59...62....44
HD........29...29...31...31...32...32....32
Fan.....2496.4970.6188.6200.6196.6196..2500
Time...10:00.9:21.8:50.7:33.7:31.9:39.10:00

I then repeated the test running just the first part and allowing the temps to drop to the "Post" level between runs. The additional test included running in Firefox and then running in Safari with Coolbook enabled. The settings I had for Coolbook are as follows:

Freq Default Coolbook
.800 0.9000..0.9000
1200 0.9750..0.9125
1400 1.0625..0.9250
1600 1.1500..0.9500

The results were somewhat interesting and I was concerned that even with lower CPU voltages the Coolbook run seemed to run a little hotter. I did verify the process was active in my active pids, so not sure why I'm seeing this.

.......Firefox Coolbook
CPU.........65.......69
HS..........53.......58
Base 1......41.......41
Base 2......41.......41
Base 3......35.......34
Airport.....38.......37
Power.......56.......61
HD..........35.......33
Fan.......6196.....6200

As a final test I wanted to see if Coolbook was helping or not on the core temps. I ran the CPU test with and without the above Coolbook settings. The test I ran was Huge, Repetitions 10, Instances 2. I didn't let it complete as the temps all seemed to stabilize after 5 minutes. I did let it run for 6:30 seconds, but again the temps didn't change from the readings at 5 minutes. The iStat showed both CPU cores running at 95% or greater during the test.

.......With CB Without CB
CPU.........82.........78
HS..........63.........63
Base 1......42.........42
Base 2......42.........42
Base 3......36.........36
Airport.....39.........38
Power.......64.........64
HD..........33.........35
Fan.......6200.......6200

No, it's not wrong, that's what was reported during this test so if any of the Coolbook users can tell me what I am doing wrong here I would love to hear it. For now I'm not going to use it.

The Air is a 1.6 with the 80gb drive. I did replace the thermal paste and prior to that the above video would lead to core shutdowns in about the 3rd part. As for which week it was built, no idea how you determine that.

Sorry for the long post, but with so many post talking about this I wanted to give folks something to run and compare to see if they do have a problem or not. As for me, I'm quite happy with how my system now performs.
 

ayeying

macrumors 601
Dec 5, 2007
4,547
13
Yay Area, CA
I noticed coolbook does make the system run slightly hotter. But I havne't been monitoring like you have to verify my suspicions. Right now, I just enable coolbook for the throttling, not so much for the voltage settings.
 

jont-fu

macrumors regular
Mar 13, 2008
153
56
Thank you, THEBEARMAN, for your extensive tests! The results are interesting, indeed.

One thing is missing from your test setup: monitoring of CPU clock frequency. I have the feeling that in 10.5.3 update Apple has "improved" their throttling function, so that the system more easily throttles down to 800MHz, resulting in all the parts of the system cooling down more. With CoolBook enabled my system is able to run much more in full CPU clock, 1.6GHz. To my understanding that causes all the other components in the system to run hotter too. That's why you're probably getting higher temps in the CB test.

Interesting note about re-applying the thermal paste. I have one of the early production Air's, and without CoolBook enabled could run it only for 5 seconds in full 1600MHz before throttling to 1200MHz. Seems like a thermal paste issue? I'm still a bit scared to open up the machine...

After the 10.5.3 update I did notice core shutdowns while watching Youtube equivalent video for 20mins. That's when I launched coolbookcontroller to monitor the clock speed and it was fixed to 800MHz (also did I notice that the OS update did indeed disable coolbook).
 

TheBearman

macrumors 6502
Original poster
May 23, 2008
445
87
Cary, NC
Followup

I think you're right about the more aggressive throttling after 10.3.5. I don't think if you really push the load on the system the temps will be different, but in normal operation I think it now does all it can to stay at 800Mhz. I did do a quick test with Coolbook up to monitor temps and running the video it was running 1200 most of the time. It did spike to 1600 once in a while but that speed didn't stay constant. I see that this will match with the findings in the Macbook Air overheating results in 1.2 GHz CPU thread. The difference I see is that given the option the Air would rather stay at 800 now rather than waiting for the temps to drive the throttling.

As for opening the Air and reapplying the thermal past, understand your reservations. I have built several PC systems over the years and feel someone comfortable doing this. Even so, it took me 3 attempts to get it right. The very thin heat sink I believe is the reason for the overuse of paste in the first place.

If your system does overheat that quickly and you don't feel comfortable opening the system yourself I would take it to a Mac store and have them take a look. Watching videos shouldn't cause any system to overheat to the point of core shutdowns.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.