Announcement

Collapse
No announcement yet.

Gigabyte F2A88XM-D3H AMD A88X

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • #31
    This is an addition to my previous post, which I can't edit right now because it has to be approved by a forum-moderator first.
    I tried again: I rebooted and sensors gives me no sensors found. Then I ran sudo modprobe it87 force_id=0x8728 . Immediately after that I ran sensors again and I got the data from the IT-chip. How can I put the sudo modprobe it87 force_id=0x8728 thing into the autostart on boot and where/how can I persistently get the measurements from the AMD CPU sensor?

    Comment


    • #32
      Sorry for tripple posting!
      I added it87 force_id=0x8728 to /etc/modules and now the sensors from the IT-chip show up even after rebooting. I'm still questioning wether the AMD CPU's sensor is recognized. sensors shows temp1, temp2 and temp3, but to me it looks like those are from the IT-Chip.

      Comment


      • #33
        sensors should show something like this before the it87 sensor:
        Code:
        radeon-pci-0008
        Adapter: PCI adapter
        temp1:        +26.0C  (crit = +120.0C, hyst = +90.0C)
        
        k10temp-pci-00c3
        Adapter: PCI adapter
        temp1:        +25.5C  (high = +70.0C)
                               (crit = +70.0C, hyst = +69.0C)
        The first one is the GPU sensor, provided by the radeon driver, if you use catalyst it probably has a different sensor. The second one is the CPU sensor. If you don't see that one try a `modprobe k10temp`.

        Comment


        • #34
          Thanks for your reply! I found out via the lm-sensor website ,where they have a device-list, that AMD Kaveri CPU's sensor is only supported from Kernel 3.14 onwards. I therefor upgraded to the backported 3.16 kernel and now it shows the k10temp-pci-00c3 you mentioned as well, but in my case current CPU temperature is shown as 0.0 C, but surely the Thermalright AXP-200 isn't that good ;-)
          It shows the right data for the high, crit and hyst temperatures though.
          Any suggestiones why the right CPU temp isn't shown?

          Comment


          • #35
            One thing to consider is that the AMD sensors are known to be highly inaccurate at low temperature even in BIOS or Windows. Just ran sensors again and this is the result:
            Code:
            sensors
            radeon-pci-0008
            Adapter: PCI adapter
            temp1:         +8.0C  (crit = +120.0C, hyst = +90.0C)
            
            k10temp-pci-00c3
            Adapter: PCI adapter
            temp1:         +8.2C  (high = +70.0C)
                                   (crit = +70.0C, hyst = +69.0C)
            Room temperature in here is over 20C so the values shown are clearly wrong in this case, but it's supposed to be accurate when the temperature gets close to critical.

            Comment


            • #36
              I moved a step forward :-) By putting all cores under maximum stress I figured out, that k10temp indeed is working, but at idle it will not move above 0 C which equates to about 31 C in the BIOS. I'll now figure out how high the real temperatur rises during a longer period of maximum stress and set fancontrol accordingly. Thanks guys!

              Comment


              • #37
                Originally posted by villeneuve View Post
                I moved a step forward :-) By putting all cores under maximum stress I figured out, that k10temp indeed is working, but at idle it will not move above 0 C which equates to about 31 C in the BIOS. I'll now figure out how high the real temperatur rises during a longer period of maximum stress and set fancontrol accordingly. Thanks guys!
                That sounds like sensor calibration is off. I know lm_sensors can used to calibrate the sensor data, but off the top of my head I'm not sure how to do it.

                Comment


                • #38
                  Thanks for your answer, however I forgot to add here, that in the meantime I found out, that k10temp-readings, at least with relatively current AMD-CPUs isn't a real temperature reading at all, but a direct result of a calculation based purely on current CPU-frequency and workload. You can also see that very clearly, that the temperatur k10temp gives drops instantly by several degrees when you stop a CPU-stresstest and immediately jumps by up when you re-start that stresstest. Windows software takes those data to very roughly calculate CPU-temperatures. The way the calculations are done is sadly only known to the manufacturers of the mainboards and/or system health chips. I've also not found any info about a way to do the same with lm_sensors. So if you should rediscover something about that please tell us here.
                  In my case, with Gigabyte F2A88XM-D3H, A8-7600 Kaveri and the fan of the CPU-cooler Thermalright AXP-200, the fan would allow the fancontrol software to go below 700 rpm anyway and at that speed the cooling seems to be very good anyway, even after very long stress tests, so I was confident enough so that I've now set a very high value for k10temp to reach before the fanspeeds start climbing.
                  What I don't like about fancontrol and/or lm-sensors (IDK which is to blame here) is that it's seems to be unable to take smartmontools' temperature measurements of harddisks into account.

                  Comment

                  Working...
                  X