Re: [REGRESSION] 3.15: Seems to turbo mode Intel Sandybridge Dual Core without need, overheating CPU

From: Martin Steigerwald
Date: Tue Jun 10 2014 - 05:34:00 EST


Am Montag, 9. Juni 2014, 15:44:02 schrieb Dirk Brandewie:
> On 06/09/2014 03:02 PM, Martin Steigerwald wrote:
> > Am Montag, 9. Juni 2014, 23:41:40 schrieb Martin Steigerwald:
> >> Am Montag, 9. Juni 2014, 23:33:43 schrieb Martin Steigerwald:
> >>> Hi!
> >>>
> >>> Added linux-pm to Cc. Also reboots seems to fix up the condition:
> >>>
> >>> merkaba:~> grep . /sys/devices/system/cpu/cpu[0-3]/cpufreq/cpuinfo_cur_freq
> >>> /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq:830957
> >>> /sys/devices/system/cpu/cpu1/cpufreq/cpuinfo_cur_freq:819628
> >>> /sys/devices/system/cpu/cpu2/cpufreq/cpuinfo_cur_freq:800000
> >>> /sys/devices/system/cpu/cpu3/cpufreq/cpuinfo_cur_freq:813476
> >>> merkaba:~> sensors
> >>> acpitz-virtual-0
> >>> Adapter: Virtual device
> >>> temp1: +71.0°C (crit = +98.0°C)
> >>>
> >>> coretemp-isa-0000
> >>> Adapter: ISA adapter
> >>> Physical id 0: +71.0°C (high = +86.0°C, crit = +100.0°C)
> >>> Core 0: +70.0°C (high = +86.0°C, crit = +100.0°C)
> >>> Core 1: +71.0°C (high = +86.0°C, crit = +100.0°C)
> >>>
> >>> thinkpad-isa-0000
> >>> Adapter: ISA adapter
> >>> fan1: 3137 R
> >>>
> >>>
> >>> Still hot in here and after reboot and login into KDE session there is quite
> >>> some CPU activity for a while.
> >>>
> >>> But way better than before.
> >>>
> >>> I can test whether this also happens with ACPI cpufreq driver.
> >>>
> >>> I think I didn´t see this with 3.14.
> >>
> >> Its not just me:
> >>
> >> Please change intel_pstate default to disable
> >> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1188647
> >
> > Way better with ACPI cpufreq driver and ondemand governor:
> >
> > merkaba:~> grep . /sys/devices/system/cpu/cpu[0-3]/cpufreq/cpuinfo_cur_freq
> > /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq:2501000
> > /sys/devices/system/cpu/cpu1/cpufreq/cpuinfo_cur_freq:800000
> > /sys/devices/system/cpu/cpu2/cpufreq/cpuinfo_cur_freq:800000
> > /sys/devices/system/cpu/cpu3/cpufreq/cpuinfo_cur_freq:800000
> > merkaba:~> grep . /sys/devices/system/cpu/cpu[0-3]/cpufreq/cpuinfo_cur_freq
> > /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq:1600000
> > /sys/devices/system/cpu/cpu1/cpufreq/cpuinfo_cur_freq:2500000
> > /sys/devices/system/cpu/cpu2/cpufreq/cpuinfo_cur_freq:800000
> > /sys/devices/system/cpu/cpu3/cpufreq/cpuinfo_cur_freq:1600000
> > merkaba:~> grep . /sys/devices/system/cpu/cpu[0-3]/cpufreq/cpuinfo_cur_freq
> > /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq:1800000
> > /sys/devices/system/cpu/cpu1/cpufreq/cpuinfo_cur_freq:2501000
> > /sys/devices/system/cpu/cpu2/cpufreq/cpuinfo_cur_freq:2500000
> > /sys/devices/system/cpu/cpu3/cpufreq/cpuinfo_cur_freq:800000
> > merkaba:~> grep . /sys/devices/system/cpu/cpu[0-3]/cpufreq/cpuinfo_cur_freq
> > /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq:800000
> > /sys/devices/system/cpu/cpu1/cpufreq/cpuinfo_cur_freq:2501000
> > /sys/devices/system/cpu/cpu2/cpufreq/cpuinfo_cur_freq:800000
> > /sys/devices/system/cpu/cpu3/cpufreq/cpuinfo_cur_freq:1800000
> > merkaba:~> grep . /sys/devices/system/cpu/cpu[0-3]/cpufreq/cpuinfo_cur_freq
> > /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq:1200000
> > /sys/devices/system/cpu/cpu1/cpufreq/cpuinfo_cur_freq:2501000
> > /sys/devices/system/cpu/cpu2/cpufreq/cpuinfo_cur_freq:800000
> > /sys/devices/system/cpu/cpu3/cpufreq/cpuinfo_cur_freq:800000
>
> This is ondemand/acpi_cpufreq telling a well documented lie. They tell you
> what P state was requested not the frequency the core is running at.
>
> /sys/devices/system/cpu/cpu1/cpufreq/cpuinfo_cur_freq:2501000
> is code for select the highest turbo P state in ACPI frequency table
> terms. With the HW coordination on the chip all the cores will actually be
> running at the frequency of the highest requested P state.
>
> intel_pstate returns the measured/actual frequency the core ran at during
> the most recent sample that the driver took.
>
> Something is convincing intel_pstate and ondemand that one(or more) cores
> is very busy.

Thing is: With intel_pstate after running PlaneShift *all* cores are in turbo mode.

With ondemand acpi only some. At least due to what it requests.

I expect one or maybe two cores busy while playing PS. One for PS, and one
for other stuff.

I will try to make the tests you requested. Quite busy here now. So will take
some time.

Thanks,
Martin

> > merkaba:~> sensors
> > acpitz-virtual-0
> > Adapter: Virtual device
> > temp1: +83.0°C (crit = +98.0°C)
> >
> > coretemp-isa-0000
> > Adapter: ISA adapter
> > Physical id 0: +84.0°C (high = +86.0°C, crit = +100.0°C)
> > Core 0: +83.0°C (high = +86.0°C, crit = +100.0°C)
> > Core 1: +84.0°C (high = +86.0°C, crit = +100.0°C)
> >
> > thinkpad-isa-0000
> > Adapter: ISA adapter
> > fan1: 3586 RPM
> >
> >
> >
> > Its still hot in this room, but this definately looks saner.
> >
> > Thanks,
> > Martin
> >
> >>>
> >>>
> >>>
> >>> Am Montag, 9. Juni 2014, 23:24:54 schrieb Martin Steigerwald:
> >>>> Hi!
> >>>>
> >>>> I get:
> >>>>
> >>>> Jun 9 22:41:32 merkaba kernel: [39978.006479] CPU0: Package temperature/speed normal
> >>>> Jun 9 22:41:32 merkaba kernel: [39978.006481] CPU3: Package temperature/speed normal
> >>>> Jun 9 22:41:32 merkaba kernel: [39978.006482] CPU2: Package temperature/speed normal
> >>>> Jun 9 22:41:32 merkaba kernel: [39978.006487] CPU1: Package temperature/speed normal
> >>>> Jun 9 22:44:02 merkaba kernel: [40127.673372] CPU2: Core temperature above threshold, cpu clock throttled (total events = 56554)
> >>>> Jun 9 22:44:02 merkaba kernel: [40127.673383] CPU3: Core temperature above threshold, cpu clock throttled (total events = 56554)
> >>>> Jun 9 22:44:02 merkaba kernel: [40127.674313] CPU3: Core temperature/speed normal
> >>>> Jun 9 22:44:02 merkaba kernel: [40127.674352] CPU2: Core temperature/speed normal
> >>>> Jun 9 22:45:21 merkaba kernel: [40207.302287] mce: [Hardware Error]: Machine check events logged
> >>>> Jun 9 22:46:32 merkaba kernel: [40278.054568] CPU0: Package temperature/speed normal
> >>>> Jun 9 22:46:32 merkaba kernel: [40278.054572] CPU3: Package temperature/speed normal
> >>>> Jun 9 22:46:32 merkaba kernel: [40278.054574] CPU2: Package temperature/speed normal
> >>>> Jun 9 22:46:32 merkaba kernel: [40278.054578] CPU1: Package temperature/speed normal
> >>>> Jun 9 22:48:06 merkaba kernel: [40371.570654] perf interrupt took too long (19348 > 17857), lowering kernel.perf_event_max_sample_rate to 7000
> >>>> Jun 9 22:51:32 merkaba kernel: [40578.103629] CPU3: Package temperature/speed normal
> >>>> Jun 9 22:51:32 merkaba kernel: [40578.103633] CPU0: Package temperature/speed normal
> >>>> Jun 9 22:51:32 merkaba kernel: [40578.103638] CPU2: Package temperature/speed normal
> >>>> Jun 9 22:51:32 merkaba kernel: [40578.103639] CPU1: Package temperature/speed normal
> >>>> Jun 9 22:56:32 merkaba kernel: [40878.174734] CPU1: Package temperature above threshold, cpu clock throttled (total events = 152620)
> >>>> Jun 9 22:56:32 merkaba kernel: [40878.174737] CPU0: Package temperature above threshold, cpu clock throttled (total events = 152620)
> >>>> Jun 9 22:56:32 merkaba kernel: [40878.174742] CPU3: Package temperature above threshold, cpu clock throttled (total events = 152620)
> >>>> Jun 9 22:56:32 merkaba kernel: [40878.174744] CPU2: Package temperature above threshold, cpu clock throttled (total events = 152620)
> >>>> Jun 9 22:56:32 merkaba kernel: [40878.176744] CPU3: Package temperature/speed normal
> >>>> Jun 9 22:56:32 merkaba kernel: [40878.176746] CPU2: Package temperature/speed normal
> >>>> Jun 9 22:56:32 merkaba kernel: [40878.176748] CPU1: Package temperature/speed normal
> >>>> Jun 9 22:56:32 merkaba kernel: [40878.176749] CPU0: Package temperature/speed normal
> >>>> Jun 9 22:59:11 merkaba kernel: [41037.278705] CPU3: Core temperature/speed normal
> >>>> Jun 9 22:59:11 merkaba kernel: [41037.278707] CPU2: Core temperature/speed normal
> >>>> Jun 9 23:01:32 merkaba kernel: [41178.225837] CPU2: Package temperature above threshold, cpu clock throttled (total events = 177343)
> >>>> Jun 9 23:01:32 merkaba kernel: [41178.225841] CPU0: Package temperature above threshold, cpu clock throttled (total events = 177343)
> >>>> Jun 9 23:01:32 merkaba kernel: [41178.225843] CPU3: Package temperature above threshold, cpu clock throttled (total events = 177343)
> >>>> Jun 9 23:01:32 merkaba kernel: [41178.225845] CPU1: Package temperature above threshold, cpu clock throttled (total events = 177343)
> >>>> Jun 9 23:01:32 merkaba kernel: [41178.237850] CPU1: Package temperature/speed normal
> >>>> Jun 9 23:01:32 merkaba kernel: [41178.237853] CPU2: Package temperature/speed normal
> >>>> Jun 9 23:01:32 merkaba kernel: [41178.237855] CPU0: Package temperature/speed normal
> >>>> Jun 9 23:01:32 merkaba kernel: [41178.237856] CPU3: Package temperature/speed normal
> >>>> Jun 9 23:01:36 merkaba kernel: [41182.452403] mce: [Hardware Error]: Machine check events logged
> >>>> Jun 9 23:06:32 merkaba kernel: [41478.291923] CPU1: Package temperature above threshold, cpu clock throttled (total events = 204756)
> >>>> Jun 9 23:06:32 merkaba kernel: [41478.291926] CPU0: Package temperature above threshold, cpu clock throttled (total events = 204756)
> >>>> Jun 9 23:06:32 merkaba kernel: [41478.291946] CPU3: Package temperature above threshold, cpu clock throttled (total events = 204756)
> >>>> Jun 9 23:06:32 merkaba kernel: [41478.291950] CPU2: Package temperature above threshold, cpu clock throttled (total events = 204756)
> >>>> Jun 9 23:11:32 merkaba kernel: [41778.341992] CPU3: Package temperature/speed normal
> >>>> Jun 9 23:11:32 merkaba kernel: [41778.341995] CPU0: Package temperature/speed normal
> >>>> Jun 9 23:11:32 merkaba kernel: [41778.341996] CPU1: Package temperature/speed normal
> >>>> Jun 9 23:11:32 merkaba kernel: [41778.341997] CPU2: Package temperature/speed normal
> >>>>
> >>>>
> >>>> And this:
> >>>>
> >>>> merkaba:~> sensors
> >>>> acpitz-virtual-0
> >>>> Adapter: Virtual device
> >>>> temp1: +96.0°C (crit = +98.0°C)
> >>>>
> >>>> coretemp-isa-0000
> >>>> Adapter: ISA adapter
> >>>> Physical id 0: +98.0°C (high = +86.0°C, crit = +100.0°C)
> >>>> Core 0: +96.0°C (high = +86.0°C, crit = +100.0°C)
> >>>> Core 1: +96.0°C (high = +86.0°C, crit = +100.0°C)
> >>>>
> >>>> thinkpad-isa-0000
> >>>> Adapter: ISA adapter
> >>>> fan1: 3580 RPM
> >>>>
> >>>> merkaba:~> acpi -t
> >>>> Thermal 0: ok, 96.0 degrees C
> >>>>
> >>>>
> >>>> On
> >>>>
> >>>> martin@merkaba:~> phoronix-test-suite system-info
> >>>>
> >>>> Phoronix Test Suite v4.8.3
> >>>> System Information
> >>>>
> >>>> Hardware:
> >>>> Processor: Intel Core i5-2520M @ 3.20GHz (4 Cores), Motherboard: LENOVO 42433WG, Chipset: Intel 2nd Generation Core Family DRAM, Memory: 8192MB, Disk: 300GB INTEL SSDSA2CW30 + 480GB Crucial_CT480M50, Graphics: Intel HD 3000 (1300MHz), Audio: Intel 6 /C200, Network: Intel 82579LM Gigabit Connection + Intel Centrino Advanced-N 6205
> >>>>
> >>>> Software:
> >>>> OS: Debian unstable, Kernel: 3.15.0-tp520 (x86_64), Desktop: KDE 4.13.1, Display Server: X Server 1.15.1, Display Driver: intel 2.21.15, OpenGL: 3.1 Mesa 10.1.4, Compiler: GCC 4.8, File-System: btrfs, Screen Resolution: 1920x1080
> >>>>
> >>>>
> >>>> during playing PlaneShift with Intel gfx.
> >>>>
> >>>>
> >>>> I get the impression that Intel P-State driver turbo modes all of the cores
> >>>> needlessly *and* permanently once this condition is triggered. Usually
> >>>> temperature is just about 55-60 degree celsius. But not at maximum.
> >>>>
> >>>>
> >>>> I see this:
> >>>>
> >>>> merkaba:/sys/devices/system/cpu> grep . cpu[0-3]/cpufreq/cpuinfo_cur_freq
> >>>> cpu0/cpufreq/cpuinfo_cur_freq:3015917
> >>>> cpu1/cpufreq/cpuinfo_cur_freq:3008984
> >>>> cpu2/cpufreq/cpuinfo_cur_freq:3000000
> >>>> cpu3/cpufreq/cpuinfo_cur_freq:3000000
> >>>>
> >>>>
> >>>> Without that much work to do:
> >>>>
> >>>> merkaba:~> mpstat -P ALL 10
> >>>> Linux 3.15.0-tp520 (merkaba) 09.06.2014 _x86_64_ (4 CPU)
> >>>>
> >>>> 23:21:42 CPU %usr %nice %sys %iowait %irq %soft %steal %guest %gnice %idle
> >>>> 23:21:52 all 15,61 0,00 2,68 10,33 0,15 0,00 0,00 0,00 0,00 71,23
> >>>> 23:21:52 0 25,70 0,00 4,22 22,59 0,30 0,00 0,00 0,00 0,00 47,19
> >>>> 23:21:52 1 20,02 0,00 1,80 14,71 0,10 0,00 0,00 0,00 0,00 63,36
> >>>> 23:21:52 2 10,14 0,00 2,61 3,01 0,10 0,00 0,00 0,00 0,00 84,14
> >>>> 23:21:52 3 6,63 0,00 2,11 0,80 0,10 0,00 0,00 0,00 0,00 90,36
> >>>>
> >>>> 23:21:52 CPU %usr %nice %sys %iowait %irq %soft %steal %guest %gnice %idle
> >>>> 23:22:02 all 14,19 0,00 2,26 10,93 0,13 0,00 0,00 0,00 0,00 72,50
> >>>> 23:22:02 0 32,26 0,00 4,01 31,86 0,20 0,00 0,00 0,00 0,00 31,66
> >>>> 23:22:02 1 12,11 0,00 1,30 8,51 0,10 0,00 0,00 0,00 0,00 77,98
> >>>> 23:22:02 2 8,12 0,00 2,21 2,91 0,10 0,00 0,00 0,00 0,00 86,66
> >>>> 23:22:02 3 4,31 0,00 1,60 0,40 0,10 0,00 0,00 0,00 0,00 93,59
> >>>>
> >>>> 23:22:02 CPU %usr %nice %sys %iowait %irq %soft %steal %guest %gnice %idle
> >>>> 23:22:12 all 25,33 0,03 2,63 9,00 0,15 0,00 0,00 0,00 0,00 62,87
> >>>> 23:22:12 0 35,71 0,00 4,01 19,56 0,20 0,00 0,00 0,00 0,00 40,52
> >>>> 23:22:12 1 27,43 0,00 1,90 10,61 0,20 0,00 0,00 0,00 0,00 59,86
> >>>> 23:22:12 2 22,14 0,10 2,40 4,41 0,10 0,00 0,00 0,00 0,00 70,84
> >>>> 23:22:12 3 15,96 0,10 2,21 1,41 0,10 0,00 0,00 0,00 0,00 80,22
> >>>>
> >>>> 23:22:12 CPU %usr %nice %sys %iowait %irq %soft %steal %guest %gnice %idle
> >>>> 23:22:22 all 27,90 0,03 3,94 8,87 0,33 0,00 0,00 0,00 0,00 58,94
> >>>> 23:22:22 0 36,77 0,00 5,81 20,74 0,90 0,00 0,00 0,00 0,00 35,77
> >>>> 23:22:22 1 28,99 0,00 2,41 10,13 0,10 0,00 0,00 0,00 0,00 58,38
> >>>> 23:22:22 2 25,48 0,00 3,81 3,21 0,20 0,00 0,00 0,00 0,00 67,30
> >>>> 23:22:22 3 20,34 0,00 3,71 1,40 0,20 0,00 0,00 0,00 0,00 74,35
> >>>>
> >>>>
> >>>>
> >>>> Will reboot now to see whether it resets that condition. Didn´t happen during the whole
> >>>> day.
> >>>>
> >>>> But seems to happen after a while of playing PlaneShift.
> >>>>
> >>>> Any hints?
> >>>
> >>> Thanks,
> >>>
> >>
> >>
> >
>

--
Martin 'Helios' Steigerwald - http://www.Lichtvoll.de
GPG: 03B0 0D6C 0040 0710 4AFA B82F 991B EAAC A599 84C7
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/