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

Starfyre

macrumors 68030
Original poster
Nov 7, 2010
2,905
1,136
If you guys and gals have the time, please consider installing the following free tool:

http://www.apple.com/downloads/dashboard/status/istatpro.html

After running your machine for around 20 minutes, please get a screenshot of your dashboard or post your reported temperatures and/or battery remaining so we can get an idea what the heat and battery life are like. Don't forget to state your processor (i.e. 2.0/2.3/2.6) and what your workload is (idle,light,moderate,heavy) . If you own a NON-HASWELL RMBPs are welcome too!

Hopefully with this information, we can all get an idea how each of our temperatures are. Thank you!
 
Last edited:
Generally, none of my temps go above 35 degrees under moderate load.
Haven't pushed the computer past 60 yet.
15" - 2.3 750M
 
bout 95-100 while gaming / rendering.


just been to a LAN this whole weekend, not a single crash.
 
Early 2013
Mavericks
Hooked up to Thunderbolt display
Moderate load - skype, outlook, streaming music, firefox 5 tabs, no flash.

48-50C average.
 
Don't know if I'm the only one that's noticed this, but iStat pro doesn't seem to be recording the CPU diode temperature; for me at least.

I also have SMC fan control installed. Often, SMC fan Control will say the CPU diode temp is 45-50C, yet iStat will only display the CPU heatsink temperature, which is about 10C lower and not an accurate representation.

FanControls temperature readings match pretty well what a multitude of other tools running under windows are reportiing. So I'm more inclined to believe FanControl over iStat pro.


Generally, none of my temps go above 35 degrees under moderate load.
Haven't pushed the computer past 60 yet.
15" - 2.3 750M

This could have to do with Haswell, but I still kind of don't believe temps like this could be accurate. There is no way your core CPU temp stays under 40C all the time -- just no. That's got to be the temperature at the heatsink or something.
 
After 20 minutes of having it closed and hooked up to a monitor, nothing is warmer than 105˚F; most are about 89-98˚F. This is just moderate web browsing, Aperture, and Word.

*Edit*Forgot to mention I have a Haswell 2.6Ghz with 16GB RAM and 512GB SSD.
 
Heres mine:

47 Celsius @
rMBP Late 2013 15" 2.3/ 16Gb Ram / 1Tb Pcie SSD/ 750m (but i always use Gfxcardstatus to only use the integrated card Iris Pro)


nbc47a.jpg
 
Don't know if I'm the only one that's noticed this, but iStat pro doesn't seem to be recording the CPU diode temperature; for me at least.

I also have SMC fan control installed. Often, SMC fan Control will say the CPU diode temp is 45-50C, yet iStat will only display the CPU heatsink temperature, which is about 10C lower and not an accurate representation.

FanControls temperature readings match pretty well what a multitude of other tools running under windows are reportiing. So I'm more inclined to believe FanControl over iStat pro.




This could have to do with Haswell, but I still kind of don't believe temps like this could be accurate. There is no way your core CPU temp stays under 40C all the time -- just no. That's got to be the temperature at the heatsink or something.
Well either way thats what iStatPro is showing.
I couldn't believe it myself - that a laptop could run as cool as my gaming rig at idle...
 
Here are my iStat temps. Normal usage using Firefox.
 

Attachments

  • Screen Shot 2013-11-05 at 00.03.48.png
    Screen Shot 2013-11-05 at 00.03.48.png
    1.7 MB · Views: 233
Guys, idle/low usage temps mean almost nothing.

Please post your temps while stressing your CPU (gaming or synthetic tests such as geekbench3)
 
bout 95-100 while gaming / rendering.


just been to a LAN this whole weekend, not a single crash.

I'm guessing you have the 750m 15" rmbp? I'm curious how it performs in general. I play Starcraft II, and apparently I can't go any higher than medium graphics settings at 1680x1050, while my three year old 460 GTX can manage ultra settings.

----------

Guys, idle/low usage temps mean almost nothing.

Please post your temps while stressing your CPU (gaming or synthetic tests such as geekbench3)

I've hit around 90-100C as a max. I believe that's the max the cpu will let you go before it throttles.
 
How in the world are you managing those temps?

I don't know, it might be because I live in a very cold area and it's freaking 50 degrees fahrenheit in my bedroom right now with the heat on. I even have a small space heater under my desk to keep my feet warm.
 
bout 95-100 while gaming / rendering.


just been to a LAN this whole weekend, not a single crash.

that's extremely high. i think its too much if its over 90. 95-100 really seems like it will wear down the graphics processor in the long run.



I assume this is related to no proper EFI windows drivers in bootcamp? the fans simply can appropriately cool the GPU in windows?
 
I think this would be more helpful if we could come up with a standardized test, since the point is mostly to compare the thermal performance between the three CPUs, right? That's something I'd also like to know about that even Anandtech's review won't cover (they only have the 2.3).

Maybe a specific youtube video with nothing else running? Except then we'd need to make sure it's in the same browser, and if in Chrome the dGPU would need to be disabled with gfxCardStatus or the numbers will be way off between machines. What we don't want it something that will go to full load; it should be a light/medium work load that proceeds at a set rate regardless of CPU speed.
 
I think this would be more helpful if we could come up with a standardized test, since the point is mostly to compare the thermal performance between the three CPUs, right? That's something I'd also like to know about that even Anandtech's review won't cover (they only have the 2.3).

Maybe a specific youtube video with nothing else running? Except then we'd need to make sure it's in the same browser, and if in Chrome the dGPU would need to be disabled with gfxCardStatus or the numbers will be way off between machines. What we don't want it something that will go to full load; it should be a light/medium work load that proceeds at a set rate regardless of CPU speed.

You are right. Best test, IMO would be for everyone to run Geekbench3 32-bit, or something similar. It is free (64bit costs money). See Link: http://www.primatelabs.com/geekbench/download/ (Make sure other programs are closed when running this test)

I just ran Geekbench3 (32bit). Here are my max temp results (using iStat Menu). (15'' 2.3, 16gb, 512gb, 750m):
CPU Core 1: 84c
CPU Core 2: 84c
CPU Core 3: 81c
CPU Core 4:79c
CPU Die: 86c
CPU Proximity: 56c

GPU Die: 62c
GPU Proximity: 53c

Other observations:
Idle temps: ~35c (CPU Die)
 
Last edited:
You are right. Best test, IMO would be for everyone to run Geekbench3 32-bit, or something similar. It is free (64bit costs money).

It's worth doing, but this may only end up tests peak thermals, which are going to be exactly the same (because all 3 chips will likely hit peak temps, throttle, and produce no viable results while running an all-cores 100% usage test).

We need a benchmark that doesn't max the CPU and try to complete as quickly as possible, which is why I suggested video playback (regardless of your CPU speed video is played back at its actual frame rate, so the different CPUs will sit at different usage % during the test and we'll see temperatures at the same work load over time).
 
It's worth doing, but this may only end up tests peak thermals, which are going to be exactly the same (because all 3 chips will likely hit peak temps, throttle, and produce no viable results while running an all-cores 100% usage test).

We need a benchmark that doesn't max the CPU and try to complete as quickly as possible, which is why I suggested video playback (regardless of your CPU speed video is played back at its actual frame rate, so the different CPUs will sit at different usage % during the test and we'll see temperatures at the same work load over time).

Well, see my above results. It did not peak thermals, or even close to it, I would say. So I think this may be a good measurement tool. Test only takes ~1 min 30 seconds to finish.
 
Well, see my above results. It did not peak thermals, or even close to it, I would say. So I think this may be a good measurement tool. Test only takes ~1 min 30 seconds to finish.
Can you post step by step instructions?
e.g. close out all applications, shutdown machine. Log in, wait 10 minutes? Can you include the link to the video or other application that we should run.

Also include small details like whether bluetooth must be off, brightness?
 
I don't know, it might be because I live in a very cold area and it's freaking 50 degrees fahrenheit in my bedroom right now with the heat on. I even have a small space heater under my desk to keep my feet warm.

According to Google that's around 10 C. I start getting a cold when it's 20 C if I have to sit in one place for too long, so how on earth can you survive at those low temperatures?
 
Can you post step by step instructions?
e.g. close out all applications, shutdown machine. Log in, wait 10 minutes? Can you include the link to the video or other application that we should run.

Also include small details like whether bluetooth must be off, brightness?

See Link: http://www.primatelabs.com/geekbench/download/

Just click the link and download geekbench3. Open the program, but before clicking "Run Benchmarks," make sure you are not running other programs (because running other programs in the background may increase temperatures).

Make sure you have a temperature monitoring program. I recommend iStat Menu: http://bjango.com/mac/istatmenus/ (there is a free 14-day trial)

That is pretty much it. Post your results (in Celsius). It would also be helpful to post your ambient temperature if you know it.
 
Well, see my above results. It did not peak thermals, or even close to it, I would say. So I think this may be a good measurement tool. Test only takes ~1 min 30 seconds to finish.

Did you notice what the fan speeds peaked to?
 
The hardest test I've thought of so far is compiling a large amount of packages from macports. I can hit 90-100C and stay there for 20-30 min if I'm compiling something big. As for peak temps, I believe that unless you hit 90-100C the CPU goes into turbo mode, so 100C might be the hard maximum that we have to live with.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.