Boosta & Boostb

Discussion in 'Mac Pro' started by HugeHungarian, Sep 30, 2017.

  1. HugeHungarian macrumors member

    Joined:
    Sep 29, 2017
    Location:
    Mount Pleasant, MI
    #1
    Straight forward question: Where are boosta and boostb?

    Looking at the list, I assume PCI is the one in the center, PS is Power Supply, Intake is the lower front bigger fan in front of the tray, Exhaust is exact opposite, but I had no idea there were more than that. When I crank them up to 5200rpm, I hear them, but relative to the others, they are very quiet.

    I couldn't find a diagram online either. Every time I search for them on google images, I just keep seeing the main aforementioned fans. Could a fella have some help?

    boostfan.PNG
     
  2. Lucas Godfrey macrumors 6502

    Lucas Godfrey

    Joined:
    Jul 12, 2010
    Location:
    Somewhere between Here and There
  3. HugeHungarian, Sep 30, 2017
    Last edited: Sep 30, 2017

    HugeHungarian thread starter macrumors member

    Joined:
    Sep 29, 2017
    Location:
    Mount Pleasant, MI
    #3
    That explains why I didn't see them when cleaning it out and checking. Thank you for the quick reply!

    I was trying to figure all of this out, because I'm trying to get this machine tuned for heavy use, without thermal throttling. Whether gaming, or working. Most games are heavily gpu reliant, I know, but the card fans just exhaust into the case and things get warmer than I'd like.

    Also, trying to setup positive case pressure if possible.
     
  4. h9826790 macrumors 604

    h9826790

    Joined:
    Apr 3, 2014
    Location:
    Hong Kong
    #4
    If you didn’t intentionally break the SMC logic, you can only have positive pressure. Every time you increase the exhaust fan speed, the intake and boost fan will also spin up accordingly.

    If you mean want to avoid GPU thermal throttling. You may need to fine tune the graphic card’s BIOS to have a more aggressive onboard fan profile. Spin up the PCIe fan won’t help much.

    If you mean CPU thermal throttling. You don’t need to worry about it. Unless something like the thermal paste application is very wrong. The CPU should never be throttled on your cMP.
     
  5. HugeHungarian thread starter macrumors member

    Joined:
    Sep 29, 2017
    Location:
    Mount Pleasant, MI
    #5
    I didn't realize that about the SMC logic. Good to know. I'll have to test it.

    I was talking about CPU throttling, I should've clarified. I will have to find the thread, but I feel like I read somewhere that these CPUs started being throttled at 67c, which seemed low to me. I was seeing temps of low to mid 60s for general use loads.

    I just figured Apple programmed the fan logic to favor low fan noise, over cooling performance. I personally don't mind reasonable noise, I always prefer performance and stability.

    I was manually overriding the fans, so I will test the system balance with exhaust change only.
     
  6. h9826790 macrumors 604

    h9826790

    Joined:
    Apr 3, 2014
    Location:
    Hong Kong
    #6
    No, the CPU may start throttling at 67C T-case, but not T-diode. The CPU temperature that those software shows you are the CPU T-diode, I did some extensive tests on it, and no thermal throttling even up to 85C T-diode.
     
  7. HugeHungarian thread starter macrumors member

    Joined:
    Sep 29, 2017
    Location:
    Mount Pleasant, MI
    #7
    Thank you, this is very helpful info! It will make monitoring the system much easier. Not that I thought all the R&D done by Apple was useless, just trying to put the picture together for myself.
     
  8. h9826790 macrumors 604

    h9826790

    Joined:
    Apr 3, 2014
    Location:
    Hong Kong
    #8
    More info for you

    https://forums.macrumors.com/threads/x5690-temperature.2070707/#post-25106645

    I also prefer just a little bit more noise than warmer CPU. I now have my own fan profile, so the CPU never go above 80C. Anyway, as per the screen capture. My W3690 works fine above 67.5C T-diode (W3690's intel official limit is 67.5 T-case max), no thermal warning, CPU speed limit 100 (means no thermal throttling). In fact, no thermal throttling with native fan profile which allows the CPU warm up to (and stay at) around 85C.
    Screen Shot 2017-10-02 at 10.16.57.jpg

    Your E5520's T-case max should be 72C, even higher than my W3690. Pretty sure your CPU need at least 90C T-diode before you can reach thermal throttling.
     
  9. HugeHungarian thread starter macrumors member

    Joined:
    Sep 29, 2017
    Location:
    Mount Pleasant, MI
    #9
    Wow, that's a lot of good info. So T-case is the temp of the IHS? The IHS is the "lid" on the CPU, correct? I thought that the cMP CPUs shipped without an IHS. Is this correct? If so, I'm guessing that's why you mentioned in the thread you linked that there's no way to monitor it on cMP?

    If all of that is true, then what are the different sensors listed in the OP screenshot? I know that they are labeled, and I'm assuming the CPU Diode sensors are what you have been describing, but what about the others: "CPU A core from PCECI" and "CPU A HeatSink"

    I know what the heat sink is, but I'm wondering if this is the closest to T-case that I will get, as a point of reference? This assumption made on the fact that T-case represents IHS temp, and also assuming IHS and heat sink sensors taking readings from the pretty much the same area. I'm sure as long as I keep diode temps within the limits you've previously mentioned, all should be fine; I'm simply just trying to understand these sensors.
     
  10. h9826790 macrumors 604

    h9826790

    Joined:
    Apr 3, 2014
    Location:
    Hong Kong
    #10
    Correct, that's the temperature on the IHS (or the "lid").

    Most cMP shipped with lidded CPU indeed, but even with the IHS, there is no practical accurate way to measure the IHS temperature (IHS temperature is NOT the cMP's heatsink temperature).

    "CPU A core from PCECI" is how far the CPU AWAY from the predetermined max temperature. In other words, the higher the better, and any number greater than zero is fine.

    "CPU A HeatSink" is the cMP's heatsink temperature.

    The cMP's heatsink temperature may be very close to the IHS temperature. However, really have no idea what's the heat transfer efficiency between the IHS and cMP's heatsink. And no idea where the temperature measure from.
     
  11. HugeHungarian thread starter macrumors member

    Joined:
    Sep 29, 2017
    Location:
    Mount Pleasant, MI
    #11
    Well if mine came with IHS, that should make swapping CPUs easier, as I'll have a point of reference for how tight to have the heat sinks without crushing the socket. Cool!

    Now regarding the "CPU A core from PCECI" being a measurement of how far away you are from the limit, then why do the temps fall with increased fan speed? Wouldn't a fall in this number indicate a rise in the system temps?
     
  12. h9826790 macrumors 604

    h9826790

    Joined:
    Apr 3, 2014
    Location:
    Hong Kong
    #12
    Sorry, yours CPU are delidded, but really only your model, all other real 5,1 or single processor 4,1 are with normal lidded CPU.

    Any screens captures about fan speed increase, and the CPU A core from PCECI fall?

    From memory, MacsFanContorl is not very good at measuring that 'Temp to processor hot" parameter, but I am really not sure about if my memory is right on this one. From memory, that core from PCECI can go up and down without any definitive logic. I always ignore that number in MacsFanControl.
     
  13. HugeHungarian thread starter macrumors member

    Joined:
    Sep 29, 2017
    Location:
    Mount Pleasant, MI
    #13
    I will need to gather them. I'm not going to speak in an absolute without proof, but I am pretty sure that's the change that I noticed. It is completely possible that I read the wrong line. You're obviously a reliable source of info here, so I'm not questioning you as much as I'm questioning what I saw, or what my crazy system printed.

    Stand by.
     
  14. h9826790 macrumors 604

    h9826790

    Joined:
    Apr 3, 2014
    Location:
    Hong Kong
    #14
    No worries, I am more than happy to discuss any Mac technically stuff at here. And of course I often make mistakes. And I am always welcome if anyone points out my mistake (my memory isn't that reliable indeed :D).
     
  15. HugeHungarian thread starter macrumors member

    Joined:
    Sep 29, 2017
    Location:
    Mount Pleasant, MI
    #15
    Ok, so I found another post on this forum from 2013 that backs what you say about the PCECI measurement.

    https://forums.macrumors.com/threads/mp-5-1-what-are-your-cpu-temps.1601231/#post-17770715

    Prepare for curve ball in the screenshots below. I made sure to capture the time as well, so you could see the difference.

    PCECI_1.PNG PCECI_2.PNG

    Also, I tried to adjust exhaust fan only, leaving the others on AUTO, and nothing else adjusted. Had to change all manually for a fan increase.
     
  16. h9826790, Oct 5, 2017
    Last edited: Oct 5, 2017

    h9826790 macrumors 604

    h9826790

    Joined:
    Apr 3, 2014
    Location:
    Hong Kong
    #16
    I see, did you try that exhaust fan logic in MacOS? I really never test that in Windows indeed. And I just tested it on my cMP, the logic still there in High Sierra.

    Also, I am not 100% sure if this logic can work with MacsFanControl. Because this is the only software that hard tune the fan speed by altering the Target Fan Speed. All other software are by modify the Minimum Fan Speed. May be the Exhaust Fan (must be the lowest) logic only works by comparing the minimum fan speed. If the software actually control the fan by the Target Fan Speed, since the minimum exhaust fan speed still the lowest, therefore other fans don’t react.

    Besides, did you try stress the CPU (e.g. by running Luxmark CPU stress) and observe that temp from PCECI?
     
  17. HugeHungarian thread starter macrumors member

    Joined:
    Sep 29, 2017
    Location:
    Mount Pleasant, MI
    #17
    The only tests I've done are with MacsFanControl, in Windows. So it very well may be the issue, which I kind of wondered. Even if the logic doesn't end up working in Windows, this fan control allows for quite a bit of custom control. I can assign any fan to any sensor, and adjust custom fan speeds or curves based on that sensor. So it should work out well.

    Just need to research this PCECI sensor more. I will do some more testing later, and post my findings :)

    If anyone else has input, please jump in!
     

Share This Page