[BUG] ThinkPad T520 overheating with P-State driver

[Martin Steigerwald]

Hello Kristen, hello,

Laptop overheats with Intel P-State driver like follows:

[ 6743.833543] CPU0: Package temperature above threshold, cpu clock 
throttled (total events = 1)
[ 6743.833545] CPU1: Package temperature above threshold, cpu clock 
throttled (total events = 1)
[ 6743.833567] CPU2: Package temperature above threshold, cpu clock 
throttled (total events = 1)
[ 6743.833568] CPU3: Package temperature above threshold, cpu clock 
throttled (total events = 1)
[ 6743.834580] CPU0: Package temperature/speed normal
[ 6743.834581] CPU1: Package temperature/speed normal
[ 6743.834607] CPU3: Package temperature/speed normal
[ 6743.834608] CPU2: Package temperature/speed normal

This happens on high cpu load and is easily triggerable by playing 
PlaneShift or OpenMW.

Also reports a MCE machine check error sometimes.


This may partly be due to an aging hardware.


Yet, after I switched from Intel P-State driver to acpi-cpufreq driver the 
issue got *much* better. I switched after I found that Intel P-State driver 
doesn´t respect "noturbo" setting at all on this ThinkPad:

[Bug 97261] New: Intel P-State driver does not honor no_turbo
https://bugzilla.kernel.org/show_bug.cgi?id=97261

(I wanted to limit maximum performance in order to prevent the overheating)


I get frequencies like:

3080566
3068945
3009082
2999902 

despite no_turbo setting. So it basically switches the complete dual core 
hypertreading CPU into turbo mode. Despite not even all cores being used by 
processes. PlaneShift basically runs single-threaded. Then there are some 
other processes active in background from time to time, but but they do not 
use the other core completely usually.


Yet with acpi-cpufreq without limiting maximum performance at all, I get the 
following with the *same* workload:

Every 5,0s: cat cpu?/cpufreq/scaling_cur_freq ; sensors                                         
Wed May  6 22:12:41 2015

2501000
2501000
1000000
1000000
acpitz-virtual-0
Adapter: Virtual device
temp1:        +95.0°C  (crit = +98.0°C)

coretemp-isa-0000
Adapter: ISA adapter
Physical id 0:  +96.0°C  (high = +86.0°C, crit = +100.0°C)
Core 0:         +88.0°C  (high = +86.0°C, crit = +100.0°C)
Core 1:         +90.0°C  (high = +86.0°C, crit = +100.0°C)

thinkpad-isa-0000
Adapter: ISA adapter
fan1:        3578 RPM



Every 5,0s: cat cpu?/cpufreq/scaling_cur_freq ; sensors                                         
Wed May  6 22:22:22 2015

2501000
1000000
800000
1600000
acpitz-virtual-0
Adapter: Virtual device
temp1:        +96.0°C  (crit = +98.0°C)

coretemp-isa-0000
Adapter: ISA adapter
Physical id 0:  +97.0°C  (high = +86.0°C, crit = +100.0°C)
Core 0:         +89.0°C  (high = +86.0°C, crit = +100.0°C)
Core 1:         +90.0°C  (high = +86.0°C, crit = +100.0°C)

thinkpad-isa-0000
Adapter: ISA adapter
fan1:        3600 RPM


I barely see the Core temps go up to more than 92 degrees while with P-State 
they were consistently hitting 96-98 degrees.

The performance is better and its overheating way less. It hit throttling 
just twice so far, despite warmer room temperature today, while it basically 
hits throttling to the extent PlaneShift or OpenMW become basically 
unplayable with Intel P-State.

Its not perfect as I think it shouldn´t hit overheating at all, but well, 
maybe thats aging hardware.


It is often said Intel P-State is technically better, but now I see that 
acpi-cpufreq runs way better on my machine here.


martin@merkaba:~> phoronix-test-suite system-info

Phoronix Test Suite v5.2.1
System Information

Hardware:
Processor: Intel Core i5-2520M @ 2.50GHz (4 Cores), Motherboard: LENOVO 
42433WG, Chipset: Intel 2nd Generation Core Family DRAM, Memory: 16384MB, 
Disk: 300GB INTEL SSDSA2CW30 + 480GB Crucial_CT480M50, Graphics: Intel HD 
3000 (1300MHz), Audio: Intel 6 /C200, Monitor: P24T-7 LED, Network: Intel 
82579LM Gigabit Connection + Intel Centrino Advanced-N 6205

Software:
OS: Debian unstable, Kernel: 4.0.1-tp520-btrfs-trim-norace+ (x86_64), 
Desktop: KDE 4.14.2, Display Server: X Server 1.16.4, Display Driver: intel 
2.21.15, OpenGL: 3.3 Mesa 10.4.2, Compiler: GCC 4.9.2, File-System: btrfs, 
Screen Resolution: 3840x1080


martin@merkaba:~> LANG=C lscpu
Architecture:          x86_64
CPU op-mode(s):        32-bit, 64-bit
Byte Order:            Little Endian
CPU(s):                4
On-line CPU(s) list:   0-3
Thread(s) per core:    2
Core(s) per socket:    2
Socket(s):             1
NUMA node(s):          1
Vendor ID:             GenuineIntel
CPU family:            6
Model:                 42
Model name:            Intel(R) Core(TM) i5-2520M CPU @ 2.50GHz
Stepping:              7
CPU MHz:               1200.000
CPU max MHz:           2501.0000
CPU min MHz:           800.0000
BogoMIPS:              4983.83
Virtualization:        VT-x
L1d cache:             32K
L1i cache:             32K
L2 cache:              256K
L3 cache:              3072K
NUMA node0 CPU(s):     0-3



There is also some other bug report about this:

Please change intel_pstate default to disable 
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1188647

appears to be quite old, but still seems unresolved.

Ciao,
-- 
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/