iMac Pro iMac Pro Fan Noise Definitive Measurement?

Discussion in 'iMac' started by Uncle Freddie, May 10, 2018.

Tags:
  1. Uncle Freddie macrumors newbie

    Joined:
    Jul 13, 2015
    #1
    Hi,

    I am this >< close to buying an iMac Pro (10 core). However, I am an audio engineer who needs as close as possible to silence in my studio. Over the last 20 years, I have gone to great lengths and untold expense to achieve silence.

    I have seen articles and YouTube videos for months describing the iMac Pro as "silent" but they are, by and large, the opinion of video professionals. Folks who generally need a lesser calibration of silence than audio pros do.

    I have heard that the 10-core model is louder than the 8-core, and vice versa. And that the 18-core model is the loudest, except for when it is the quietest. Nobody seems to have their facts straight. Maybe the biggest drawback of all: I cannot just go to my Apple Store and try it myself, since the store is far too noisy.

    So I wrote a test script to saturate an iMac Pro's CPU to generate fan noise.

    The script is here: http://substrate.com/saturate.txt

    It doesn't touch the GPU, which is fine, since audio apps like Logic Pro X, etc. don't either.

    Would anyone like to run it and report back?

    Please take a look and confirm for yourself that it will do absolutely no harm to your system. Save it, rename it to saturate.sh, chmod 755 it, and run it from the Terminal.

    My thinking is that a generous person out there could run it and report back how loud their iMac Pro gets.

    Ideal: measure the noise with an SPL meter.

    Report your results here, or private message me with them. I would greatly appreciate it.

    Thanks,
    Fred
     
  2. joema2 macrumors 68000

    joema2

    Joined:
    Sep 3, 2013
    #2
    I'll do this but -- A or C weighted? What measuring position? If several people do that it's better to use the same procedure.

    They should also take a baseline reading before running the test and post that.

    I'd suggest not using "at your listening position" because that varies too much. Maybe 1 foot in front of the Apple logo at the same height and not resting on the table (else it could pick up noise by conduction).

    If it's an iPhone app like SPL Pro (http://www.studiosixdigital.com/audiotools-modules-2/spl-modules/spl/), then everyone should also state the model of iPhone and hold it in the same orientation. Maybe parallel to the monitor face with the phone bottom at the height of the Apple logo and 1 ft in front?

    Let me know your preferences and I'll measure my 10-core iMac Pro and 2017 i7 iMac.
     
  3. Uncle Freddie thread starter macrumors newbie

    Joined:
    Jul 13, 2015
    #3

    Thanks joema2.

    I think A-weighted measurements would be best.

    I agree on taking a baseline reading first.

    I agree with you on listening position. 1 foot away, in some kind of shock mount, sounds right.

    I have made a minor update to http://substrate.com/saturate.txt

    Now it will invoke the load more elegantly, and runs entirely in the foreground, so a simple control-C will stop everything.

    Also now, instead of finishing after 10 or so minutes, the test will keep running until you control-C it.


    Thanks,
    Fred
     
  4. joema2 macrumors 68000

    joema2

    Joined:
    Sep 3, 2013
    #4
    Sorry about the delay, I've been busy on a film project, will do this tomorrow.
     
  5. joema2 macrumors 68000

    joema2

    Joined:
    Sep 3, 2013
    #5
    OK I ran it on my 2017 i7 iMac 27 and 10-core Vega 64 iMac Pro. Measured at above-listed location using iPhone 7 running SPL Pro, A-weighted, slow averaging:

    iMac 27 baseline: 38.2 dB - fan speed 1200 rpm
    iMac Pro baseline: 36.7 dB - fan speed 1100 rpm

    iMac 27 after 5 min load: 43.7 dB - fan speed 2700 rpm
    iMac Pro after 5 min load: 38.0 dB - fan speed 1500 rpm

    The iMac ran speed ramped up almost immediately. The iMac Pro fans didn't budge off idle for several minutes, then slowly increased to a peak of about 1500 rpm.

    Besides the dB numbers, the subjective difference was huge. The iMac was pretty loud -- anyone who walked in the room would hear it. Nobody would notice the iMac Pro, I only heard it because I was listening for it.

    So the iMac Pro is a lot quieter under this particular load. However it is not silent and when doing sustained heavy video rendering involving both CPU and GPU I have occasionally heard it. There have been a few infrequent cases when it was very noticeable, but it doesn't happen often or last long.
     
  6. Uncle Freddie thread starter macrumors newbie

    Joined:
    Jul 13, 2015
    #6
    Cool! Thank you very much joema2! I really appreciate it.


    So the baseline is 36.7 dBA, and the under load number is 38.0 dBA. That's a 1.3 dBA increase, and seems to prove that the iMac Pro will remain quiet under the loads I plan to give it during my audio recordings.

    In the meantime, I scoured the internet and after much, much searching, actually found some specs direct from Apple (see attachment), at least for the 8-core/Vega56 model: https://support.apple.com/kb/SP771?locale=en_US

    That page says that the idle noise of an 8-core/Vega56 is 13 dBA. I am guessing that is of course measured in an anechoic chamber.

    So if I subtract 23.7 dBA from your readings to compensate for the noise floor of your environment: fridge, cars going by, furnace/ac, etc., I am guessing the ramped up dBA for your 10-core iMac Pro, in an anechoic chamber, would be 13 + 1.3 = 14.3 dBA

    Can you critique my calculations?

    Thanks,
    Fred
     

    Attached Files:

  7. joema2 macrumors 68000

    joema2

    Joined:
    Sep 3, 2013
    #7
    I think that's probably fairly close. Be advised that very infrequently my 10-core iMac Pro is sometimes considerably louder than my 38.0 dB using your test. But if that test is completely representative of your production workload it's probably OK.

    I'd also recommend you buy the machine from a retailer that allows returns with no restocking fee. B&H and Adorama do not accept returns on computers, including Macs.
     
  8. tired-cummander macrumors newbie

    Joined:
    Nov 10, 2014
    #8
    Dear Sir Mr. Uncle Freddie,

    i came here for exactly the same reasons you describe: i am considering to upgrade to the iMac Pro 2018 but i am unsure as to whether it will cause a lot of fan noise and thus create disturbances while mixing and doing other audio tasks.

    Could you perhaps shortly update me/us on how you decided in the end, whether you got the iMac Pro and whether you are happy with its level of noise/silence?

    That would be of great help! Thank you!
     
  9. Uncle Freddie thread starter macrumors newbie

    Joined:
    Jul 13, 2015
    #9

    Sure. I ended up getting the 3GHz 10-core, with 64GB ram, 2TB disk, and Vega64 graphics. Audio tends to be single-threaded, so I wanted the cpu with the highest turbo boost.

    So far, so good. I don't hear the fans under normal circumstances. I don't hear them over the ambient noise of the room. Do they ever ramp up? For what I do, no.

    I use Logic Pro X and other audio applications. I don't do any video work, but I opted for the Vega64 anyway so that all components -- cpu, disk, ram and gpu -- got an across the board +1 bump.

    If I run the saturate script I linked to, I can get the fans to spin up to an audible level. But to be fair, that script saturates all 10 cores; 20 if you count hyperthreading.

    So to summarize: an artificial experiment can cause the fans to be audible. But working daily on my music does not. Not even close, honestly.
     

    Attached Files:

Share This Page