Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
Haswell laptop owerheats under load with intel_pstate
View unanswered posts
View posts from last 24 hours

 
Reply to topic    Gentoo Forums Forum Index Kernel & Hardware
View previous topic :: View next topic  
Author Message
Barbieken
n00b
n00b


Joined: 22 Mar 2014
Posts: 62

PostPosted: Sat Mar 22, 2014 11:17 am    Post subject: Haswell laptop owerheats under load with intel_pstate Reply with quote

Lenovo Thinkpad T540p i7 4700 (8 cores) overheats under load (for example compiling chromium).
I use the latest kernel from testing

intel_pstate works. when I do
Code:

cd /sys/devices/system/cpu/intel_pstate/
echo 50 > max_perf_pct

it does not overheats no more, frequency stays at 1700Mhz. But when max_perf_freq is 100 (default value) the laptop switches to turbo mode (3400Mhz) under load and does not throttle even when temperature goes well above 80C. The fan kicks in when temperature is high but it still can't cope with overload. And it looks like the fan kicks in too late. Under windows the fan works with short but very powerful bursts even when temperature is low. With linux it seems like it prefers to run more slowly but constantly.

Is there anything I can do to control the fan and temperature with intel_pstate, short of writing manually values to max_perf_pci and to the fan device (forgot its' name)? I want the laptop to stay under 70C even under heaviest load.

Code:

# cpupower frequency-info
analyzing CPU 0:
  driver: intel_pstate
  CPUs which run at the same hardware frequency: 0
  CPUs which need to have their frequency coordinated by software: 0
  maximum transition latency: 0.97 ms.
  hardware limits: 800 MHz - 3.40 GHz
  available cpufreq governors: performance, powersave
  current policy: frequency should be within 800 MHz and 3.40 GHz.
                  The governor "powersave" may decide which speed to use
                  within this range.
  current CPU frequency is 1.70 GHz (asserted by call to hardware).
  boost state support:
    Supported: yes
    Active: yes
Back to top
View user's profile Send private message
Barbieken
n00b
n00b


Joined: 22 Mar 2014
Posts: 62

PostPosted: Sat Mar 22, 2014 11:51 am    Post subject: Reply with quote

i7z output:
Code:

Linux's inbuilt cpu_khz code emulated now
True Frequency (without accounting Turbo) 2394 MHz
  CPU Multiplier 24x || Bus clock frequency (BCLK) 99.75 MHz

Socket [0] - [physical cores=4, logical cores=8, max online cores ever=4]
  TURBO ENABLED on 4 Cores, Hyper Threading ON
  Max Frequency without considering Turbo 2493.75 MHz (99.75 x [25])
  Max TURBO Multiplier (if Enabled) with 1/2/3/4 Cores is  34x/33x/32x/32x
  Real Current Frequency 3334.28 MHz [99.75 x 33.43] (Max of below)
        Core [core-id]  :Actual Freq (Mult.)      C0%   Halt(C1)%  C3 %   C6 %   C7 %  Temp      VCore
        Core 1 [0]:       3257.97 (32.66x)      1.82    8.78       1       0    87.7    56      1.0049
        Core 2 [2]:       3334.28 (33.43x)      1.18    98.4       0       0       0    60      1.0048
        Core 3 [4]:       3281.68 (32.90x)      2.61    6.75       1    1.63      87    54      1.0023
        Core 4 [6]:       3266.11 (32.74x)      2.11    0.325      1       1    94.8    53      1.0049

Back to top
View user's profile Send private message
Gusar
Advocate
Advocate


Joined: 09 Apr 2005
Posts: 2664
Location: Slovenia

PostPosted: Sat Mar 22, 2014 12:08 pm    Post subject: Reply with quote

Install thermald. It's not in portage yet, but you can find an ebuild here: http://gpo.zugaina.org/sys-power/intel-thermald
Back to top
View user's profile Send private message
albright
Advocate
Advocate


Joined: 16 Nov 2003
Posts: 2541
Location: Near Toronto

PostPosted: Sat Mar 22, 2014 12:17 pm    Post subject: Reply with quote

or install thinkfan, which will let you define temperature
levels for various fan levels ...

Code:
* app-laptop/thinkfan
     Available versions:  0.8.1-r1 0.9_beta2 {atasmart}
     Homepage:            http://thinkfan.sourceforge.net
     Description:         simple fan control program for thinkpads

_________________
.... there is nothing - absolutely nothing - half so much worth
doing as simply messing about with Linux ...
(apologies to Kenneth Graeme)
Back to top
View user's profile Send private message
Barbieken
n00b
n00b


Joined: 22 Mar 2014
Posts: 62

PostPosted: Sat Mar 22, 2014 1:56 pm    Post subject: Reply with quote

thermald takes 10-15% of CPU and just makes things worse. Either its' defaults are unusable or it just does not work with the latest Haswell chips, there are hint's on it in the thermald issues page on github

Thinkfan works, it speeds up the fan earlier, according to its' settings, still it is not exactly what I want. It makes the laptop very noisy and the fan can't cope with heavy load anyway when turbo_boost is on.. I'd prefer to throttle cores somehow instead of cooling them. Any way to do this beside disabling intel_pstate in the kernel or disabling turbo_boost by writing '1' to intel_pstate/no_turbo?

That's what I get when run sensors:
Code:

 # sensors
acpitz-virtual-0
Adapter: Virtual device
temp1:        +55.0°C  (crit = +200.0°C)

coretemp-isa-0000
Adapter: ISA adapter
Physical id 0:  +55.0°C  (high = +84.0°C, crit = +100.0°C)
Core 0:         +55.0°C  (high = +84.0°C, crit = +100.0°C)
Core 1:         +50.0°C  (high = +84.0°C, crit = +100.0°C)
Core 2:         +49.0°C  (high = +84.0°C, crit = +100.0°C)
Core 3:         +50.0°C  (high = +84.0°C, crit = +100.0°C)


What do 'high' and 'crit' mean? Am I right that 'high' is a temperature when throttling starts and 'crit' when laptop should shut down? If I am right, how can I lower high to, say, 70C?


ATM I've just disabled turbo_boost, it makes temperature more sane even without thinkfan
Back to top
View user's profile Send private message
creaker
l33t
l33t


Joined: 14 Jul 2012
Posts: 651

PostPosted: Sat Mar 22, 2014 6:38 pm    Post subject: Reply with quote

Overheating is a well known Haswell problem. Intel decided to save on the cost of thermal paste. Strange decision for $350 CPU.
CPU scalping and thermal paste replacement reduces a temperature a lot.
http://www.youtube.com/watch?v=XXs0I5kuoX4
Back to top
View user's profile Send private message
hephooey
n00b
n00b


Joined: 24 Feb 2003
Posts: 74

PostPosted: Sun Mar 23, 2014 5:45 pm    Post subject: Reply with quote

creaker wrote:
Overheating is a well known Haswell problem. Intel decided to save on the cost of thermal paste. Strange decision for $350 CPU.
CPU scalping and thermal paste replacement reduces a temperature a lot.
http://www.youtube.com/watch?v=XXs0I5kuoX4

I think only the desktop Haswell/Ivy Bridge cpus have the thermal paste problem. The laptop version is still a bare die. So if it is the thermal paste to blame then it is Lenovo's fault. Anyway, I think 80C is nowhere near overheating during tasking jobs like emerge. I have a RMBP and the cpu temp goes up to 100C frequently and cpu will not throttle before it hit that temperature.
Back to top
View user's profile Send private message
pa1983
Tux's lil' helper
Tux's lil' helper


Joined: 09 Jan 2004
Posts: 100

PostPosted: Sun Mar 23, 2014 8:49 pm    Post subject: Reply with quote

I dont understand how you define overheating? Overheating should mean it runs to hot then whats allowed and also negative side effects as a result.

My old Via C3 overheated after 7 years. Thermal past was so dry it idled at 60-70C and hit 80C+ just like that during compiling that it segafaulted.

I7 processors or any modern intel CPU can take at least 95C most can take 100C and some 105C. 80C and below are usually safe recommended temperatures.

80C for an i3, i5 or i7 especially stock cooler or laptop is well no problem during load.

I built a i7 3770 ITX system and it uses a Noctua low profile cooler, more or less comparable with stock in terms of cooling and that system hits 79C but the 3770 can and will hold 3.7Ghz turbo on all cores none stop with out missing a beat under prime95 or intel linpack so thats pretty good when its base clock is 3.4Ghz.

I use to run my old i7 920 @ 3.7Ghz whit realy low RPM fans on a Noctua NH-U12P and I could hit 90C+ during two instances of Intel linpack and was rock solid. I used Intel C-STATE and C1E to allow it to run at any clock up to 3.7Ghz and shut its cores down to save power when going in to C1 state. Had that system for 3 years with no problems running 24/7.

The howl idea of turbo boost is to get as much performance out of the CPU with in the TDP and temperature and voltages allowed and for laptops not exiding TDP is important for the limited power supply on the motherboard. Running Intel linpack and with Hyper treading on, yeas its 4 cores 8 threads not 8 cores. That would allow you to test how high turbo you can maintain and how high temps you can get. i7z Will show that for you you like you know.

But 80C for a laptop is not unusual. They are built whit no extra margins on the cooling and the CPU should protect it self. My frinds ASUS laptop with an i5 is dam hot to touch even at idle. 50-60C. But its not a real problem because it never runs to hot during load then what intel specify any way. Its just uncomfortable to hold :P

As you can see your CPU as Max allowed temperature of a 100C. usually they dont throttle until they hit that.
http://ark.intel.com/products/75117

80C and lower is realy safe temperatures.
_________________
WS: i7 3930K@4Ghz, 32Gb ram, 256Gb NVME & 128Gb sata SSD, GTX780 3Gb & RX 460 2Gb
NAS: i3 4360 3.7Ghz, 20Gb ram, 256Gb SSD, 42Tb HDD, NIC: Intel 2x1Gbit
ROUTER: J1900 2Ghz, 8Gb ram, 128Gb SSD, NIC: 2x1Gbit, WIFI: Atheros AR9462 and AR5005G
Back to top
View user's profile Send private message
Navar
Guru
Guru


Joined: 20 Aug 2012
Posts: 353

PostPosted: Sun Mar 23, 2014 9:51 pm    Post subject: Confusing thermal protection features as 'normal' Reply with quote

Current mobile processors are never in the power consumption range of their desktop versions.

If your thermal paste is actually dried out from age or excessive heat strain, it's more an insulator than a conductor, along with additional air gaps due to contraction. It's usually one of the worst things to allow to happen, short of running a die bare on air.

Intel removed solder to the IHS (integrated heat spreader) replacing with a paste compound for TIM (thermal interface material) layer as part of cost cutting and profit padding. Unfortunately due to that, it doesn't transfer as well and the process seems to have some QA issues (large temp variances across the die->IHS). I haven't purchased a laptop since 2006, but the TDP and cooling issues remain the same in low profile space. That dual core laptop does not hit 60C under load, but I allow the fans to properly vent off the heat pipes.

Most won't care on the desktop side unless they're overclocking. My i5-3570K peaks around 55C at high load. Most of the time it is well below that. But I'm not overclocking it. IB->Haswell+ due to the IHS change tend to be poor overclockers to keep reasonably cool.

Please explain how anywhere near 100C is defined as normal and acceptable by Intel engineering on a unit with TDP ranging from 17 to 40ish in the worst of cases?

There's no magic or miracles here. That's a small TDP range on mobile with small power supplies. Which translates into little incoming energy to make excessive heat. Either you're allowing heat to ventilate out properly, or you're not, to be getting anywhere near those temps. If what was said is true about no IHS on current mobile chips from Intel, then there's even less reason other than mentioned next...

Either you've got TIM quality issues, a bad heat pipe (or design, e.g. coupled poorly with a hot GPU), dead fans, excessive insulation (dust), or a broken BIOS with very poor fan control.

I can go to a store to units active for hours on display, idle or not, and generally not feel excessive heat spots on the casing or the vent exhaust. They're clearly not running near 100C.

Edit: price correction
_________________
Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.


Last edited by Navar on Tue Mar 25, 2014 5:49 pm; edited 1 time in total
Back to top
View user's profile Send private message
pa1983
Tux's lil' helper
Tux's lil' helper


Joined: 09 Jan 2004
Posts: 100

PostPosted: Mon Mar 24, 2014 4:35 am    Post subject: Reply with quote

The heat flux from modern processors per square area is Enormous. This means that the temperature difference is big between silicon and CPU cooler but also the air cooling it. Simple Thermal dynamics. Silicon conducts heat worse the aluminum actually. My i7 3930K at 4Ghz under intel linpack can hit over 70C and close to 80C with my NH-D14 and I will measure about 37C exhaust air temperature from the CPU cooler. So Delta T is pretty high on high end processors. Its obviously lower on a Mobile CPU since Delta T is dependence on the thermal conductivity of the materials, quare area and energy given of by the CPU. But the lower TDP advantage on mobile processors is used to reduce the size of the heatsink instead because it can simply run hotter due to lower Delta T between sink and CPU.

A modern processor can run fine at high loads at 80C+ and because of that idle temps are often held high to, 50-60C. Same can be seen on modern graphic cards. Just to reduce noise because the silicon dont realy take any significant damaged from those temperatures.

The math for a desktop 77W TDP Ivy Bridge with Tim under its IHS VS soldering like SB its about 10C higher.
I did the math a few years ago. I also know people that used liquid metals in room temperature specifically made for IC and copper cooler. They removed the IHS and replaced the tim with Liquid metal.

Any one is interested I can quote my self. There realy is no seacret or question about how good or bad Tim VS soldered IHS because its just high school math.

Quote:
If we assume that pasta has 0.5 to 10 W / (m.K) which means that if the pasta has 5W (m.K), it takes 5W added heat on one side of area 1m ² and a thickness of 1 m to achieve a difference of 1K or 1C. So very simple cube that measures 1x1x1m that we introduces 5W to one side, that side will be 1C warmer than the side opposite to it.

Artic Silver 5 is said to have 8.7W / (m.K), so for example, I planned to use 5W / (m.K) because it's an average between best / worst thermal pastas.

Artic silver is 0.003in or 0.08mm thick when it is under pressure between two flat surfaces so that I use as the thickness of the paste when you need it to calculate thermal resistance and the variable called k so k = 0.00008 when it is in meters .

Area needed to and it is 121mm² on IVY and the area indicated by the variable x as x = 0.000121m² for it shall be given in meters.

R = resistance we want.

R = x / A * k = 0.00008/0.000121 * 5 = 0.13K / W

So for every watt input, we get a difference in temperature between silicon and IHS at 0.13C.

So for a stock i7 at 77W that ends up to 10C.

An overclocked IVY in 4.5-5GHz range draws about 150W so then we get 150 * 0.13 = 19.5C Delta T. So thats how big the difference was between silicon and IHS with a normal layer off average thermal pasta.

With Artic Silver 5 I get 11.4C Delta T at 150W.

Aluminium has a value of 200W (m.K) so it had given 0.5C difference between silicon and IHS.

But as I said the difference is greater the more power the processor pumps out. Does not mean it's a problem for most, but the difference was larger than I believed.

So if the IHS was soldiered to the chipset on could expect 0.5C Delta T between the silicons top surface and the IHS bottom where they meet instead of 10C or so. And the difference gets huge when overclocking.



Any one how wants to know what intel Deems acceptable can just go to intels web site or google any of there products and get there. Then download and read the pdf dock with there specifications.
All there processors have extensive documents detailing everything from operating temperatures, voltages and current draws. Min, max, typical, what ever you can think off.

I have done that for 15 years.
_________________
WS: i7 3930K@4Ghz, 32Gb ram, 256Gb NVME & 128Gb sata SSD, GTX780 3Gb & RX 460 2Gb
NAS: i3 4360 3.7Ghz, 20Gb ram, 256Gb SSD, 42Tb HDD, NIC: Intel 2x1Gbit
ROUTER: J1900 2Ghz, 8Gb ram, 128Gb SSD, NIC: 2x1Gbit, WIFI: Atheros AR9462 and AR5005G
Back to top
View user's profile Send private message
creaker
l33t
l33t


Joined: 14 Jul 2012
Posts: 651

PostPosted: Mon Mar 24, 2014 1:52 pm    Post subject: Reply with quote

60C, 80C or even 100C (in some cases) is a safe temperature for CPU. CPU can work at such a conditions.
But laptop isn't a CPU only. There are a lot of parts that dislikes overheating. Hot CPU warms up all around. HDD that operates at 55-60C has a lifetime two times shorter than HDD that operates at 30-35C.
So, temperature reducing is a thing to deal with, even if CPU is in a safe range. I think it worth to check thermal paste and heatsink state.
Back to top
View user's profile Send private message
pa1983
Tux's lil' helper
Tux's lil' helper


Joined: 09 Jan 2004
Posts: 100

PostPosted: Mon Mar 24, 2014 9:05 pm    Post subject: Reply with quote

That I never disagreed with Either. Its good that hardware to day can mostly take high temps, allows for low noise. Disadvantage with laptops is that they tend to be very warm to the touch for the user. Some hardware might not like it but we do have higher rated capacitors to day for example then used historically. Placement of harddrive in a laptop can alleviate such problems to.

Googles study have shown that cooler hard drive breaks faster then hotter once. Optimal temperature google found after 100K hard drives where 42C or so. Hard drives under 30C broke more often then those running closer to 60C.

Laptop designers and OEM designers in general of PC's often make bad design decisions but thats just to cut cost and increase margins. Critical components like a CPU still takes priority, hard drives not so much.

But hard drive temp can easily be read with hddtemp so not a real problem to figure out if thats optimal or not. But to hot running hard drives is not a common problem I would say.
_________________
WS: i7 3930K@4Ghz, 32Gb ram, 256Gb NVME & 128Gb sata SSD, GTX780 3Gb & RX 460 2Gb
NAS: i3 4360 3.7Ghz, 20Gb ram, 256Gb SSD, 42Tb HDD, NIC: Intel 2x1Gbit
ROUTER: J1900 2Ghz, 8Gb ram, 128Gb SSD, NIC: 2x1Gbit, WIFI: Atheros AR9462 and AR5005G
Back to top
View user's profile Send private message
creaker
l33t
l33t


Joined: 14 Jul 2012
Posts: 651

PostPosted: Mon Mar 24, 2014 11:09 pm    Post subject: Reply with quote

pa1983 wrote:

Googles study have shown that cooler hard drive breaks faster then hotter once. Optimal temperature google found after 100K hard drives where 42C or so. Hard drives under 30C broke more often then those running closer to 60C.


Google knows more than a HDD manufacturers?
As for laptops, HDD usually doesn't operates with temp below 30C, so, a comparison for e.g. 20C and 60C is incorrect.
If your HDD operates at temp below 30C (10-15-20) it means some unusual conditions

Check out this Fujitsu pdf:
www.fujitsu.com/downloads/COMP/fcpa/hdd/sata-mobile-ext-duty_wp.pdf
Especially pages 3, 4, 5.
Back to top
View user's profile Send private message
Anon-E-moose
Advocate
Advocate


Joined: 23 May 2008
Posts: 3921
Location: Dallas area

PostPosted: Mon Mar 24, 2014 11:18 pm    Post subject: Reply with quote

Bottom line, if hardware is kept to a reasonable temp it will last longer, whether CPU or HD.

A CPU may have a max temp of 100C but if it runs at anywhere near that all the time
it will have a much shorter lifespan than if it ran at 50C most of the time.

The same for hard disks.

Room temp right now is 24, but my disks are 26, 27 and 30 for an external raid array in it's own case (all celsius )

The more moderate the temp the longer electronic components will last.
_________________
Asus m5a99fx, FX 8320 - nouveau, oss4, rx550 for qemu passthrough
Acer laptop E5-575, i3-7100u - i965, alsa
---both---
5.0.13 zen kernel, profile 17.0 (no-pie) amd64-no-multilib
gcc 8.2.0, eudev, openrc, openbox, palemoon
Back to top
View user's profile Send private message
pa1983
Tux's lil' helper
Tux's lil' helper


Joined: 09 Jan 2004
Posts: 100

PostPosted: Tue Mar 25, 2014 2:03 am    Post subject: Reply with quote

Your CPU wont run at 100C, it will run at 80C a few percent of its life time assuming its a PC for a normal user. The rest it will run in the 40-50C range. The CPU will outlast any other component most likley so life time is irrelevant. You dont plan to have that CPU for 10 year do you? When did your CPU die before any other component?

So yes its true that hardware last longer at lower temperatures but what kills hardware is electromigation and thats caused by current primarily. Higher load means higher current so even a well cooled system worrying at 100% 24/7 will die sooner then a similar system running hotter but spending less time drawing a lot of current.
Temperature to some extent increases electromigration but current still is the main cause.

Electromigration kills modern electronics before temperature assuming temperature and voltage is with in sad max specifications. So basically if it runs with in spec electromigration is the killer.

Hard drives are not only electrical devices but mechanical. Most mechanical hardware has a optimal operating temperature. Controller card running bit hotter or colder is not what I would worry about but rather how well the mechanical parts takes excessive cold or heat.

So even if your CPU ran a bit hot at idle, say 60C assuming its a modern CPU the cores will often be shut down completely so no current flow, no electromigration.
Other parts still powered runs at reduced voltage and that means reduced current at idle so once again the main cause for say a CPU to fail is reduced.

And google dont need to be a manufacturer. Discarding there statistical findings because they are a BIG user of hard drives is just a cheap shoot if you cant back your claim up.

Getting a hard drive down to under 30C so the 20C region is easy. Normal room temperature (22C) and any power efficient hard drive and you will get there if you just have a weak fan cooling it.
Most PC's also power down there hard drives so in those cases the hard drives rarely go above 30C due to inactivity. In laptops it all depends on the placement of the hard drive. Ideally if you want it cool with a hot idling CPU you would place it so the incoming air passes over the hard drive. That way the CPU cooler can keep it cold to.
_________________
WS: i7 3930K@4Ghz, 32Gb ram, 256Gb NVME & 128Gb sata SSD, GTX780 3Gb & RX 460 2Gb
NAS: i3 4360 3.7Ghz, 20Gb ram, 256Gb SSD, 42Tb HDD, NIC: Intel 2x1Gbit
ROUTER: J1900 2Ghz, 8Gb ram, 128Gb SSD, NIC: 2x1Gbit, WIFI: Atheros AR9462 and AR5005G
Back to top
View user's profile Send private message
666threesixes666
Veteran
Veteran


Joined: 31 May 2011
Posts: 1248
Location: 42.68n 85.41w

PostPosted: Tue Mar 25, 2014 5:20 am    Post subject: Reply with quote

ebay "cookie rack" cooler.

http://www.ebay.com/itm/3-Tier-Wire-Baking-Cooling-Rack-Cooking-Stackable-Cookie-Sheet-Nonstick-Metal-/271415710636?pt=Bakeware&hash=item3f31a313ac
Back to top
View user's profile Send private message
Barbieken
n00b
n00b


Joined: 22 Mar 2014
Posts: 62

PostPosted: Tue Mar 25, 2014 8:22 am    Post subject: Reply with quote

All right guys, this discussion is fascinating. I don't have HDD, replaced it with SSD immediatelly, but I really want to keep the laptop into sub-70C range. Performance is enough even without turboboost, so I just disabled it altogether. But I really would prefer it to run in ~35C when idle and in sub-70 when turbo-boosted, and tur off turbo-boost after temperature exceeds 70-75C. As it was said there are other components, thermal paste loses its' properties faster when temperature is high and imo laptop should not work like teapot and boil water anyway.
If it is impossible well, then it is impossible.


upd: the laptop is brand new so I cannot disassemble it and replace paste, this will woid warranty.
Back to top
View user's profile Send private message
Anon-E-moose
Advocate
Advocate


Joined: 23 May 2008
Posts: 3921
Location: Dallas area

PostPosted: Tue Mar 25, 2014 9:35 am    Post subject: Reply with quote

Barbieken wrote:
upd: the laptop is brand new so I cannot disassemble it and replace paste, this will woid warranty.


Given that you don't want to open the case (not sure why the warranty would disallow that) there's not much to do.

If you bought it from some local store, I suppose you could take it there and have them check it out.
Could be the fan isn't working properly or blocked airflow.

Good luck.
_________________
Asus m5a99fx, FX 8320 - nouveau, oss4, rx550 for qemu passthrough
Acer laptop E5-575, i3-7100u - i965, alsa
---both---
5.0.13 zen kernel, profile 17.0 (no-pie) amd64-no-multilib
gcc 8.2.0, eudev, openrc, openbox, palemoon
Back to top
View user's profile Send private message
Gusar
Advocate
Advocate


Joined: 09 Apr 2005
Posts: 2664
Location: Slovenia

PostPosted: Tue Mar 25, 2014 11:15 am    Post subject: Reply with quote

Barbieken wrote:
But I really would prefer it to run in ~35C when idle

I don't know any laptop which can achieve that. All idle at about 45-50C. My desktops barely do what you want, they idle around 30-32C.
Back to top
View user's profile Send private message
albright
Advocate
Advocate


Joined: 16 Nov 2003
Posts: 2541
Location: Near Toronto

PostPosted: Tue Mar 25, 2014 11:56 am    Post subject: Reply with quote

Quote:
I don't know any laptop which can achieve that. All idle at about 45-50C.


depends on the chip; my i5-4300U (ultra low voltage) idles at 35.
_________________
.... there is nothing - absolutely nothing - half so much worth
doing as simply messing about with Linux ...
(apologies to Kenneth Graeme)
Back to top
View user's profile Send private message
creaker
l33t
l33t


Joined: 14 Jul 2012
Posts: 651

PostPosted: Tue Mar 25, 2014 2:00 pm    Post subject: Reply with quote

Gusar wrote:
Barbieken wrote:
But I really would prefer it to run in ~35C when idle

I don't know any laptop which can achieve that. All idle at about 45-50C. My desktops barely do what you want, they idle around 30-32C.


Even AMD CPU can be cool at idle. Recently I repaired HP Pavilion G6 with AMD chip that was 55-60C at idle and over 80C under the load.
After thermal paste replacing idle temp was reduced down to 33C.
Back to top
View user's profile Send private message
Navar
Guru
Guru


Joined: 20 Aug 2012
Posts: 353

PostPosted: Tue Mar 25, 2014 6:51 pm    Post subject: Reply with quote

@OP

You wanted help, here are some ways I'd go about it.

https://bugzilla.redhat.com/show_bug.cgi?id=859597 might provide some clues on your quad core. Doesn't make me want to run linux on a Lenovo 'business' laptop though. Sounds like a good hand warmer in the winter.

Energy in -> heat. Start at the power supply adaptor. The intel_pstate driver may have issues. I haven't seen it be the end all be all and a google search seems to indicate problems. When I see on a search a number of systems, like yours, exhibiting substantially high temps in linux versus windows, something is suspect. I had these problems with my laptop years ago due to kernel bugs. Not the first time that has happened. If you have a dual boot setup with windows on this system, is it showing similar temps? How do the casing/power supply feel for heat?

Nowhere is Intel stating Tcase/Tjunction temps as being the expected area to run for normal use. I wish people would quit making these claims in recent years.

You can try disabling Intel's pstate driver and try the older powersave / userspace governors. With userspace, set frequency on cpu0-3 to your frequency minimum. Watch your temps, make sure if you have any fan control at all that you have some amount of active cooling going (otherwise temps will climb over time no matter what). At 35C+ off sensors to heatpipes, you should be able to feel warm exhaust air from the fan exhaust vents. At 60C+ it should feel hot.

In more extreme cases, I've went with undervolting. You may not have a way to do this, however.

Your processor seems to have a TDP rating of 47. That's high for a non-GPU in a confined space. The real question is how frequently it approaches that limit. Apparently Intel is going back to Preshot days (->Hasfail) instead of improving heat/power constraints. It still doesn't justify those temperatures with active cooling while idle. It can be cooled, but not if Lenovo is doing it on the cheap or poor design.

I'm using an older Intel T2500 2 core processor (TDP 31). Definitely not the coolest running CPU, nor does it have all the fancier lower power states of modern. With an ambient temp of 20C the sensor temps are showing under 30C. The power supply is barely warm. If I force the frequency to minimum with full load it gets warm, but not hot. If I set everything on high, I'll hit around 50ish C with heavy load and the 90W power supply will feel quite a bit warmer. With discrete GPU loaded as well, the case will have warm spots unless the fans are on high and the power supply will feel hot (none of this includes battery charging).

If Intel core temps are to be believed, mine vary on low load from 5-10C above ambient on a desktop setup with good (whisper quiet) active cooling. But we're not talking desktop setup here for you. Many current day manufactures seem to think 'modern' equates to burning the user on the plastic casing than allowing them to hear a fan run on low. Slower dust intake lack of cleaning failure that way, until the warranty phase has passed.
_________________
Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Kernel & Hardware All times are GMT
Page 1 of 1

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum