Automated tool to reveal throttling and overheating - GitHub

[NazgulRR]

Out of curiosity, anyone run these on the new 12" Macbooks?

See my early posts wrt the boost. In short, it's normal not to boost to 2.60 GHz since this stress test uses all cores (2.60 is only reached when only one core is used). The throttling when using the GPU is indeed quite bad ...

Thanks for getting back on this. To my horror, after just re-running the test half a year later, it seems that my MBA has deteriorated a bit as well. The CPU throttles even more under the GPU load than before. Will investigate/re-run the tests later. I wonder at which point it becomes a case for Apple Care...

 

[qnxor]

Thanks for getting back on this. To my horror, after just re-running the test half a year later, it seems that my MBA has deteriorated a bit as well. The CPU throttles even more under the GPU load than before. Will investigate/re-run the tests later. I wonder at which point it becomes a case for Apple Care...

Post here the "before" and "after" results if you have them.

I recently witnessed how a modded rMBP with a liquid metal thermal paste (instead of the usual non-metal pastes) has had the liquid metal solidify over the period of a few months, and then slightly detach from the chip core, leaving very poor heat transfer between the chip and the heatsink ... the poor thing was throttling like crazy.

If the temperatures deteriorate visibly over time, then I'd say it's a sign that either the thermal paste has begun to dry up or, perhaps due to some physical shock, the heatsink/chip was slightly displaced from the thermal paste leaving poorer contact (personally I think the latter scenario is much less likely).

 

[NazgulRR]

Post here the "before" and "after" results if you have them.

I recently witnessed how a modded rMBP with a liquid metal thermal paste (instead of the usual non-metal pastes) has had the liquid metal solidify over the period of a few months, and then slightly detach from the chip core, leaving very poor heat transfer between the chip and the heatsink ... the poor thing was throttling like crazy.

If the temperatures deteriorate visibly over time, then I'd say it's a sign that either the thermal paste has begun to dry up or, perhaps due to some physical shock, the heatsink/chip was slightly displaced from the thermal paste leaving poorer contact (personally I think the latter scenario is much less likely).

Here:

December 31 top and July 31 below. The drop in average fps was about 0.40.

The x264 was very similar - with a 0.50fps drop on average.

In retrospect, it doesn't seems like it deteriorated as much as I thought. I will definitely keep doing these tests periodically.

 

[qnxor]

December 31 top and July 31 below. The drop in average fps was about 0.40.

The x264 was very similar - with a 0.50fps drop on average.

In retrospect, it doesn't seems like it deteriorated as much as I thought. I will definitely keep doing these tests periodically.

First, I don't see much difference to be honest. Average and peaks are very close. I don't see the fps result in the bottom one but I'd think they are close (assuming you used the exact same settings in the GPUTest test). 0.5 fps drop in x264 is not significant, mine goes up and down by that much when running multiple tests consecutively.

Second, the idle temperatures are very different (look at the start): one is 45C the other is 60C ... obviously you either were in a hotter room, or had more tasks running in the background. It's actually a pretty good result considering!

In general, these stress test are best run with nothing else running in the background, i.e. Activity Monitor should show you 99% idle cpu on average. I always close all running or resident apps (in dock or tray).

 

[NazgulRR]

First, I don't see much difference to be honest.

Definitely agree (as admitted already in the post). Good point about the idle temperature too. I can think of two reasons 1) Winter-Summer change or 2) running the test too soon after the x264 one.

 

[yashag]

Heres my 2015 rMBP 13 inch OSX 10.11.6 Intel Iris 6100
ambient temp 22˚C

 

[Schranke]

First (....)

Hi @qnxor so I found this thread through a link posted another site on the forum.
I do not seems to be able to get it to run, when ever I try to run Prime95 it wants to download intel power gadget, which fails (I also have the newest version). how would one get around "cannot create /user/xxxx/macoh/tmp/intel/ and so on Illegal byte sequence"?

 

[BGS]

Hi @qnxor so I found this thread through a link posted another site on the forum.
I do not seems to be able to get it to run, when ever I try to run Prime95 it wants to download intel power gadget, which fails (I also have the newest version). how would one get around "cannot create /user/xxxx/macoh/tmp/intel/ and so on Illegal byte sequence"?

Bit late in the day for this, but just in case anyone else has the same question, I found a solution in the github comments:

Comment this line in macoh.sh
#unzip -q -o $tmp/ipg.zip -d $tmp
and change these lines as well
wget $tmp/ipg.dmg "$url_ipg" -#
mnt $tmp/ipg*.dmg

and add change the url to
url_ipg="https://software.intel.com/file/613985/download"

then it will work



Alternatively, use the free Tunabelly Stress Test App, which seems fairly similar but is up to date:

‎Endurance: CPU Stress Test

‎Endurance is a free tool to benchmark battery life as well as run CPU stress tests. To test battery life, it's as simple as clicking the start button and waiting until the battery has depleted to zero percent. Once that occurs, the total time that the Mac lasted will be available along with...

apps.apple.com