Automated tool to reveal throttling and overheating - GitHub

[qnxor]

Oops, missed all the posts since last year, sorry. Some bulk replies below.

My 2014 2.2 base model. Is this good or bad?

Typical newer generation Haswell and MBP: should have been better. A few of us discussed it in earlier posts. In short, Intel and Apple allowed higher max frequency which overheats quicker and then throttles more heavily than the previous generation. In other words, your laptop can't sustain the high frequency for too long. Looking at your graph, you got 3.3 Ghz for only about 15 sec, then it dropped to ~2.8 GHz. My previous gen MBP has a lower max freq but doesn't throttle and maintains a steady 3.2 GHz. I do long simulations so I'm happy. If you don't do heavy stuff using *all* cores, then you're ok.

Thanks ^, Got it working.

Can someone interpret this chart? 2012 Macbook Pro, Yosemite 10.10

Looks excellent.

Hi guys. Here are my results on a macbook pro late 2013 16GB RAM. Regarding ambient temp I don't see any reading on the istat menus 5.01 (568).

What are your thoughts?

Similar explanation as @grame, see my comments above (yours is less severe, dropped only to 3.0 GHz).

2013 Macbook Air 13" with Intel® Core™ i5-4250U Processor (3M Cache, up to 2.60 GHz) here.

How come the CPU never raises above 2.30GHz? Shouldn't it boost to 2.60GHz??

Also, that is some heavy throttling on the CPU when GPU is in use, no?

I re-ran the two tests twice with 12hours in between them, fresh boot, ambient T around 22C.

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 this great "app"

Here are my results from early 2011 macbook pro 13" with some dust, and the factory thermal grease. And then some cleaning, and applying Arctic Silver MX-4 instead the 4 yrs old factory stuff...

You're welcome, and nicely done on the new tim.

First off, thank you very much for coding this. I have been suspecting my MBP has been throttling a lot when using it in clamshell mode and I think the test results show that. The longer an intensive process, the more it seems to throttle I'm hoping the more informed MacRumors community can confirm by interpreting my results for me.

Run the x264 tests (and the GPU tests if you have an Nvidia GPU) and then report back. Don't trust the Prime95 tests too much. I was specific in my early posts that Prime95 stresses the CPU way too much and that virtually any laptop will throttle. I've only included Prime95 for extreme testing and for knowledgeable overclockers. Still, yours drops all the way from 3.3 GHz to 2.2 Ghz, which is severe.

For your CPU which is older generation (heats more by default), a good thermal design should not drop below 2.8-3.0 GHz in the x264 test.

Also, I'm not sure what "clamshell" means but if it means "lid closed" then it's expected to heat more (at least on the newer models which have vents pushing air up the screen). You should run the tests with the lid open normally.

Cheers.

p.s. I may just remove the Prime95 tests from the MacOH tool. Causes confusion for people with insufficient understanding.

 

[matty1551]

Run the x264 tests (and the GPU tests if you have an Nvidia GPU) and then report back. Don't trust the Prime95 tests too much. I was specific in my early posts that Prime95 stresses the CPU way too much and that virtually any laptop will throttle. I've only included Prime95 for extreme testing and for knowledgeable overclockers. Still, yours drops all the way from 3.3 GHz to 2.2 Ghz, which is severe.

For your CPU which is older generation (heats more by default), a good thermal design should not drop below 2.8-3.0 GHz in the x264 test.

Also, I'm not sure what "clamshell" means but if it means "lid closed" then it's expected to heat more (at least on the newer models which have vents pushing air up the screen). You should run the tests with the lid open normally.

Cheers.

p.s. I may just remove the Prime95 tests from the MacOH tool. Causes confusion for people with insufficient understanding.

I've tried running the x264 test but get this every time. "hdiutil: attach failed - not recognized"

I'd venture to guess that I will get heavy throttling because most times when I do a handbrake encode it will throttle down well below the advertised 2.2GHz.

 

[qnxor]

I've tried running the x264 test but get this every time. "hdiutil: attach failed - not recognized"

I'd venture to guess that I will get heavy throttling because most times when I do a handbrake encode it will throttle down well below the advertised 2.2GHz.

I see. Sounds like a different version of MacOSX. Which one is it? I may be able to fix it, but I can't test it (I was running 10.9.3 when I wrote it).

Throttling below 2.2 GHz in Handbrake for your CPU is not good. REMEMBER that 2.2 GHz is only the base frequency. Your CPU should do 3.3 GHz when using 1 core, and at least 3.0 GHz when using 4 cores. The idle temperature (close to 60C) is high to begin with, but still ...

I was going to suggest that you can still manually install Intel Power Gadget and run Handbrake with the settings in the macoh.sh script to check. YOu seem to have done that already. A graph would look better (helped a few people in this thread to convince Apple to fix their throttling problems), but Apple may agree to fix it regardless - is it still under warranty though?

 

[matty1551]

I was going to suggest that you can still manually install Intel Power Gadget and run Handbrake with the settings in the macoh.sh script to check. YOu seem to have done that already. A graph would look better (helped a few people in this thread to convince Apple to fix their throttling problems), but Apple may agree to fix it regardless - is it still under warranty though?

I was using the latest version of 10.8. I partitioned my drive and did a clean install of mavericks and then ran all updates. I still get the same error when trying to perform the x264 encode. It throttles heavily in the prime test and monitoring a handbrake encode, it throttles to an unacceptable level as well.

My depot warranty is up in a week. Would you recommend I make an appointment?

Also, attached is a video of the throttling if you have time to watch it.

Prime 95 Clamshell Mode (Laptop Closed) 10.9.5

Handbrake Encode Throttling Video
https://www.youtube.com/watch?v=ff6TbW0LgDU

 

[qnxor]

I was using the latest version of 10.8. I partitioned my drive and did a clean install of mavericks and then ran all updates. I still get the same error when trying to perform the x264 encode. It throttles heavily in the prime test and monitoring a handbrake encode, it throttles to an unacceptable level as well.

My depot warranty is up in a week. Would you recommend I make an appointment?

Also, attached is a video of the throttling if you have time to watch it.

Prime 95 Clamshell Mode (Laptop Closed) 10.9.5
Image

Handbrake Encode Throttling Video
https://www.youtube.com/watch?v=ff6TbW0LgDU

Definitely make an appointment now. Apple will run their own tests regardless of what you show them, but it helps if you have a strong case which is documented thoroughly.

The video you made of the Handbrake encode is good but you should try to show the full handbrake window as well as the Intel Power Gadget window. I would also take a screenshot of the IPG window's graph to show the decrease from 3 GHz down to 1.6 GHz throttling -- that is very serious.

Before you start encoding in Handbrake, Choose Logging -> Log to file from the IPG window, leave it idle for 30 seconds, then start the Handbrake encode, leave it encoding for 5 minutes, stop the encoding, leave it idle for 30 seconds, then choose Logging > Log to file again to stop logging. Send me the log file afterwards or put it on pastebin and give a link here. I'll try to turn it into a graph before you go to see Apple.

The clamshell thing is a bit worthless though. It's expected to heat up more (and you're using Prime95 again too). The throttling looks horrible, I agree, but it won't help your case with Apple. You want to convince Apple that it throttles badly in normal usage conditions.

 

[matty1551]

The clamshell thing is a bit worthless though. It's expected to heat up more (and you're using Prime95 again too). The throttling looks horrible, I agree, but it won't help your case with Apple. You want to convince Apple that it throttles badly in normal usage conditions.

Why would the clamshell thing be worthless? If it's an advertised feature of the machine, then aren't these normal usage conditions? Shouldn't I reasonably expect to be able to perform the same tasks at similar speeds?

Not trying to be contradictory as you are clearly a very intelligent person. I just don't understand how a feature of the machine wouldn't be considered normal use.

I will run a handbrake tonight with power gadget logging it. I'll send it to you. If you have time to create a graph for it that'd be awesome. My genius bar appointment is this sunday.

 

[qnxor]

Why would the clamshell thing be worthless? If it's an advertised feature of the machine, then aren't these normal usage conditions? Shouldn't I reasonably expect to be able to perform the same tasks at similar speeds?

Not trying to be contradictory as you are clearly a very intelligent person. I just don't understand how a feature of the machine wouldn't be considered normal use.

I will run a handbrake tonight with power gadget logging it. I'll send it to you. If you have time to create a graph for it that'd be awesome. My genius bar appointment is this sunday.

I should have been clearer. I meant it's "worthless" from the point of view of testing and capturing throttling conditions. It doesn't help me much to determine the actual conditions which cause yours to throttle, I have virtually no comparison base, and I also expect that it will throttle (much) more than with the lid open (on my MBP Late 2013, the air vents look more obstructed when the lid is closed, and the heat also builds up quicker due to the lid -- with it open, the vents look less obstructed and the top aluminum side also helps dissipate more heat).

I offered to help by modifying the macoh.sh script so you can use it, or by producing a graph which you can directly compare against the results posted until now and against the ones I know. For that, I don't want to use clamshell mode. Apple may also be more convinced if it throttles heavily in "normal" conditions.

Otherwise, yes, it is supposed to work with the lid closed too. You should just expect it to throttle and heat up more, that's all.

Oh and I'm not intelligent. I just spent enough time in academia to become a professional bull****ter.

Cheers.

 

[yjchua95]

Oh and I'm not intelligent. I just spent enough time in academia to become a professional bull****ter.

Cheers.

Says the professor who's at the University of Cambridge's Department of Applied Maths and Theoretical Physics

 

[qnxor]

Oh and I'm not intelligent. I just spent enough time in academia to become a professional bull****ter.

Says the professor who's at the University of Cambridge's Department of Applied Maths and Theoretical Physics

See how easy that was? It only proves my point

( I'm not a professor though. Those guys are on a whole different level of bull****ing. )

 

[johnnnw]

Can you clarify for me a little how you interpret that graph?

 

[johnnnw]

 

[matty1551]

I should have been clearer. I meant it's "worthless" from the point of view of testing and capturing throttling conditions. It doesn't help me much to determine the actual conditions which cause yours to throttle, I have virtually no comparison base, and I also expect that it will throttle (much) more than with the lid open (on my MBP Late 2013, the air vents look more obstructed when the lid is closed, and the heat also builds up quicker due to the lid -- with it open, the vents look less obstructed and the top aluminum side also helps dissipate more heat).

I offered to help by modifying the macoh.sh script so you can use it, or by producing a graph which you can directly compare against the results posted until now and against the ones I know. For that, I don't want to use clamshell mode. Apple may also be more convinced if it throttles heavily in "normal" conditions.

Otherwise, yes, it is supposed to work with the lid closed too. You should just expect it to throttle and heat up more, that's all.

Oh and I'm not intelligent. I just spent enough time in academia to become a professional bull****ter.

Cheers.

I dropped the laptop off tonight. The genius I spoke to really didn't seem very knowledgable and I ended up asking him to check it in for further evaluation. He even ran the GPU test. Don't see how that's relevant.

I explained that I was not getting the advertised "2.2 Ghz" and showed him the tests I ran with graphs.

One thing I figured out was that for some reason Macs Fan Control was causing the throttling to SOME degree. I had the fans set based on the cpu sensors. The fan threshold was much lower than Apple's so I don't understand why this would cause throttling.However, any time I set it back to Automatic, the computer would get much hotter before the fans kicked in but the throttling would happen to a lesser degree.

It now throttles to 1.8GHz using external displays.

My question is, what in your opinion defines an acceptable amount of throttling? I asked the Apple guy the same and he had no answer. If they call me and tell me it's fine do you feel it's worth forcing the issue here?

 

[qnxor]

I explained that I was not getting the advertised "2.2 Ghz" and showed him the tests I ran with graphs.

As I explained, 2.2 GHz is just the base frequency; your CPU boosts via turbo to 3.4 GHz. Your CPU should get 3.0 GHz when all 4 cores are active, and 3.3 GHz when only one core is active! Look at your own graphs and observe the frequency at the beginning and then the fall (then look at my graphs in the 1st post of this thread, mine is advertised as max 3.5 GHz (one core) and 3.3 GHz (4 cores)).

One thing I figured out was that for some reason Macs Fan Control was causing the throttling to SOME degree. I had the fans set based on the cpu sensors. The fan threshold was much lower than Apple's so I don't understand why this would cause throttling.However, any time I set it back to Automatic, the computer would get much hotter before the fans kicked in but the throttling would happen to a lesser degree.

You should set the fans to 100% speed and then check for throttling. I have a hunch it will still throttle heavily.

My question is, what in your opinion defines an acceptable amount of throttling? I asked the Apple guy the same and he had no answer. If they call me and tell me it's fine do you feel it's worth forcing the issue here?

For the throttling you are experiencing, I for one would ram it down their throats ... in other words, yes, it's worth forcing the issue. Like I said above, you should get 3.0 GHz on that CPU when all 4 cores are active. Anything less than 2.5 GHz when you are stress-testing using Handbrake and I'd be at Apple's throat, but that's me.

You haven't provided me with Handbrake data that I asked so I can't give a more accurate answer.

My MBP Late 2013 is advertised as "up to 3.5 GHz (one core active) and up to 3.3 GHz (4 cores active)". If on average I was getting half that after paying 2000 GBP, then I'd either demand my money back or be onto them until they fixed it. If you are in the UK, recall that there is the Trading Standards Institute and the option of settling the matter in a small claims court. However, Apple has played nice in many cases until now (including in this thread as a result of this MacOH tool I coded), so have hope, I'd say!

p.s. You really don't want to hear my opinion about the technical competence or knowledge of the average Apple Genius.

 

[Queen6]

I have followed this thread with some interest, Apple`s drive for "thin & quiet" has often brought less than perfect quality and performance. It`s admirable what Apple is striving for, equally it`s not acceptable to commercially launch systems that don't live up to advertised specifications, nor systems that fail prematurely due to "thermal shock" as is frequently the case with the MacBook Pro with dGPU.

Same sentiment as the OP, if any of my Mac`s throttle excessively I will simply return the Notebook and ask for a replacement, a small depreciation is acceptable given the thermal envelope of the MBP. Handbrake is definitely a representative test of the MBP, combined with Intel Power Gadget any throttling is very clear to all.

Q-6

 

[matty1551]

You haven't provided me with Handbrake data that I asked so I can't give a more accurate answer.

Yeah I apologize about that. I dropped the ball with this. It is "Ready" so when I go pick it up tomorrow and they don't have any sort of resolution for me then I will run the handbrake encode and send you the log file.

Again, thanks so much for explaining all of this in so much detail. Helps a ton.

 

[qnxor]

Yeah I apologize about that. I dropped the ball with this. It is "Ready" so when I go pick it up tomorrow and they don't have any sort of resolution for me then I will run the handbrake encode and send you the log file.

Again, thanks so much for explaining all of this in so much detail. Helps a ton.

No worries, my pleasure. Fingers crossed with the Apple replacement/repair.

 

[someoneoutthere]

I've tried running the x264 test but get this every time. "hdiutil: attach failed - not recognized"

Same problem here. I initiated the x264 transcode test on terminal, it started fetching and installing [Intel Power Gadget --> Graphic Layout Engine --> ImageMagick --> Handbrake CLI (%100)] --> gave the error "hdiutil: attach failed - image not recognized"

I am running Yosemite; 10.10.2

 

[qnxor]

@someoneoutthere I reverted to Mavericks because of the graphics switching bug in Yosemite (freezes my MBP regularly) so I can't test it myself.

Can you please copy + paste here the entire output that you see on the screen starting with the first line until it errors out? I can't tell which module fails.

EDIT: Don't bother, I can reproduce it on Mavericks. Will post again when fixed.

 

[johnnnw]

If you could OP can you comment on my results pictures above? I'm curious.

Thank you.

 

[qnxor]

If you could OP can you comment on my results pictures above? I'm curious.

Thank you.

I believe I did, see post #176.

 

[someoneoutthere]

EDIT: Don't bother, I can reproduce it on Mavericks. Will post again when fixed.

Thank you! I was able to run the Prime95 test though! This is how it ended up:
Edit: Ambient temp of 65F.

 

[qnxor]

Fixed. The version of Handbrake used was no longer available at the former URL.

Download the latest release from the releases page on GitHub and retry:

https://github.com/qnxor/macoh/releases

p.s. That's a very good result for Prime95, though your ambient temp is very low which helps a lot (your idle cpu temp is less than 35C, mine is 50C).

 

[johnnnw]

I believe I did, see post #176.

Oh okay thank you! I redownloaded it and posted new results because that old photo disappeared.

 

[qnxor]

Oh okay thank you! I redownloaded it and posted new results because that old photo disappeared.

Can't see any new results, but your old one was looking great anyway.

 

[whg]

Macbook Pro 15" Retina mid 2012

Hi Bogdan, I just discovered this thread and your great diagnostic tool. I had to test it on my now 2.5 years old mid 2012 rMBP. I was a little shocked by the high temperatures reached, but throttling seems to be no issue?
Regards, Werner