You are not logged in.

#1 2013-10-13 18:04:42

mid-kid
Member
Registered: 2013-05-07
Posts: 177

[Solved] Side effects of compiling on slower machines.

So, I had been doin' casual stuff on my big 'ol laptop (Quad-core I7), when it's bios randomly bricked and I was unable to boot. (I think it was the bios, since it crashed on the boot logo before dying.)

Anyway, I now have a silly lil' lappy (Dual-Core I3 fuq yeah!) which I actually bought for school.
Compiling linux-pf took ~5 hours, and wine-silverlight around the same time.

I wonder, since compiling tasks have to be split over only 2 cores (I use -j4 anyway tongue), and they have to work much harder to compile because they don't have as much power as I7, does compiling also affect the lifetime of the lappy much more than it did on my "big 'ol laptop"?
And, does it also have more side effects?

EDIT: I forgot to state that the lappy is a netbook.

Last edited by mid-kid (2013-10-23 07:26:28)


If it ain't broke, you haven't tweaked it enough.

Offline

#2 2013-10-13 18:09:39

ewaller
Administrator
From: Pasadena, CA
Registered: 2009-07-13
Posts: 12,265

Re: [Solved] Side effects of compiling on slower machines.

mid-kid wrote:

And, does it also have more side effects?

Well, it is autumn here, and some of us are starting to heat our homes.  The heat load from lappy can only help that effort tongue
But seriously, I would not worry about it unless you start running close to the critical temperatures.


Nothing is too wonderful to be true, if it be consistent with the laws of nature -- Michael Faraday
Like you, I have no idea what you are doing, but I am pretty sure it is wrong...Jasonwryan
----
How to Ask Questions the Smart Way

Offline

#3 2013-10-13 18:17:14

Xyne
Moderator/TU
Registered: 2008-08-03
Posts: 5,616
Website

Re: [Solved] Side effects of compiling on slower machines.

Longer compilations mean higher core temperatures for longer periods of time. Keep an eye on those. If they approach critical, open up the laptop, clean out the dust, and maybe even change the thermal paste. If the temperature stays below critical then I think the only long-term effect will be a shorter lifetime for the thermal paste, but the difference should be small unless you're compiling 24/7.

If you're tuning for generic architectures then I do not think there will be any other side effects, but I do not know.

Offline

#4 2013-10-13 22:07:44

mid-kid
Member
Registered: 2013-05-07
Posts: 177

Re: [Solved] Side effects of compiling on slower machines.

Xyne wrote:

Longer compilations mean higher core temperatures for longer periods of time. Keep an eye on those. If they approach critical, open up the laptop, clean out the dust, and maybe even change the thermal paste. If the temperature stays below critical then I think the only long-term effect will be a shorter lifetime for the thermal paste, but the difference should be small unless you're compiling 24/7.

If you're tuning for generic architectures then I do not think there will be any other side effects, but I do not know.

By "thermal paste", you mean the funny blue stuff I found on the CPU of my big 'ol laptop?
If I ever have to change that, I'll void my warranty... *shudder*

So, if what you're saying is true, I should not worry about compiling (as long as it doesn't take too long) and just keep an eye on the temperature.
Also, what temperature can be considered critical? Usually, my lappy runs ~65-70C.

Last edited by mid-kid (2013-10-13 22:08:56)


If it ain't broke, you haven't tweaked it enough.

Offline

#5 2013-10-13 23:38:15

cfr
Member
From: Cymru
Registered: 2011-11-27
Posts: 5,661

Re: [Solved] Side effects of compiling on slower machines.

sensors can tell you that (from lm_sensors). For example, my hardware says 86 is high 100 is critical but it varies with the machine.

65-70 sounds high for idle but not if that's when you are compiling.

EDIT: If the paste needs replacing while it is under warranty, you get them to do it.

Last edited by cfr (2013-10-13 23:38:56)


How To Ask Questions The Smart Way | Help Vampires

Arch Linux | x86_64 | GPT | EFI boot | grub2 | systemd | LVM2 on LUKS
Lenovo x121e | Intel(R) Core(TM) i3-2367M CPU @ 1.40GHz GenuineIntel | Intel Centrino Wireless-N 1000 | US keyboard with Euro | 320G 7200 RPM Seagate HDD

Offline

#6 2013-10-14 09:05:25

mid-kid
Member
Registered: 2013-05-07
Posts: 177

Re: [Solved] Side effects of compiling on slower machines.

cfr wrote:

65-70 sounds high for idle but not if that's when you are compiling.

That is for normal use. Currently running 67.5C. `sensors` reports +70C high  and +100C critical.
Maybe it's partially due to KDE or laptop-mode erroring on boot. (Have been too lazy to look into.)

cfr wrote:

EDIT: If the paste needs replacing while it is under warranty, you get them to do it.

Heh, I'll keep that in mind tongue

EDIT: Fixed the laptop-mode error, still ~65-70C with casual use. Hmm... Maybe the sensors aren't tuned? because the lappy doesn't feel warm on touch. And when freezing in my cold room for the night, when starting up it's 50C

EDIT2: When starting skype, 72C... This ain't right...

Last edited by mid-kid (2013-10-14 14:37:14)


If it ain't broke, you haven't tweaked it enough.

Offline

#7 2013-10-14 15:17:09

ewaller
Administrator
From: Pasadena, CA
Registered: 2009-07-13
Posts: 12,265

Re: [Solved] Side effects of compiling on slower machines.

Also, be aware that many modern machines use thermal interface technologies that were not developed in the mid-20th century. 

It is fairly common place to find silicon thermal pads, or, for really high performance, phase change material
(Link is to a pdf data sheet on a product selected at random from the internet.  I've no affiliation with the manufacturer, nor have I used their product)

Don't make the mistake of replacing bi-phase material or silicon thermal pads with thermal grease.  The thermal resistance will go way up (this is bad)


Nothing is too wonderful to be true, if it be consistent with the laws of nature -- Michael Faraday
Like you, I have no idea what you are doing, but I am pretty sure it is wrong...Jasonwryan
----
How to Ask Questions the Smart Way

Offline

#8 2013-10-14 20:56:13

cfr
Member
From: Cymru
Registered: 2011-11-27
Posts: 5,661

Re: [Solved] Side effects of compiling on slower machines.

Did you configure sensors?

Did you go through the configuration for laptop-mode-tools? I found I needed to tweak it a fair amount.

EDIT: Oh, and I'm using KDE. I'm currently getting about 65C - probably because pacman has been creating initramfs images and updating font caches.

Last edited by cfr (2013-10-14 20:58:54)


How To Ask Questions The Smart Way | Help Vampires

Arch Linux | x86_64 | GPT | EFI boot | grub2 | systemd | LVM2 on LUKS
Lenovo x121e | Intel(R) Core(TM) i3-2367M CPU @ 1.40GHz GenuineIntel | Intel Centrino Wireless-N 1000 | US keyboard with Euro | 320G 7200 RPM Seagate HDD

Offline

#9 2013-10-16 13:43:16

mid-kid
Member
Registered: 2013-05-07
Posts: 177

Re: [Solved] Side effects of compiling on slower machines.

cfr wrote:

Did you configure sensors?

I've done the setup as described in the arch wiki.

cfr wrote:

Did you go through the configuration for laptop-mode-tools? I found I needed to tweak it a fair amount.

I'm not that experienced when it comes to hardware, so, I don't know what to tweak. The arch wiki explains very little on the matter. I've only tweaked the CPU config a bit.

I'm gonna install windows and try to update the bios or firmware with acer's crappy installers. If that doesn't work, I'm sending the lappy to warranty.


Also, since I'm curious, another question: What about compiling on ARM platforms? I don't see any thermal paste on, for example, a raspberry pi. How so?

Last edited by mid-kid (2013-10-16 13:43:38)


If it ain't broke, you haven't tweaked it enough.

Offline

#10 2013-10-16 16:01:58

sitquietly
Member
From: Moscow, Tennessee
Registered: 2010-07-12
Posts: 216

Re: [Solved] Side effects of compiling on slower machines.

mid-kid wrote:
Xyne wrote:

Longer compilations mean higher core temperatures for longer periods of time. Keep an eye on those. If they approach critical.....

.....what temperature can be considered critical? Usually, my lappy runs ~65-70C.

I consider anything above 60 degrees to be too hot.  Much too hot.  It seems to me that if you're hitting 70 degrees you should take steps to bring the temp down.  I would run those compiles with MAKEFLAGS="-j1" and be patient and cool.  I realize that modern cpus/gpus can run at very high temps without shutting down but that doesn't mean that system longevity isn't affected.

As Xyne suggested, get rid of the dust.  Clear the vents. 

And I would drop the cpu load by using -j1 or -j2 (j2 will run a compile almost as fast as j5 with lower peak loads and significantly lower temps).

Offline

#11 2013-10-17 01:10:39

brenix
Member
From: California
Registered: 2008-03-05
Posts: 182

Re: [Solved] Side effects of compiling on slower machines.

If you have multiple computers , you should check out Distcc if you haven't already..  I can compile a kernel in 2/3 the time it use to take me. Therefore, my CPU doesn't stay hot for too long..

Offline

#12 2013-10-18 13:30:32

ronnylov
Member
Registered: 2010-12-21
Posts: 55

Re: [Solved] Side effects of compiling on slower machines.

mid-kid wrote:
Xyne wrote:

Longer compilations mean higher core temperatures for longer periods of time. Keep an eye on those. If they approach critical, open up the laptop, clean out the dust, and maybe even change the thermal paste. If the temperature stays below critical then I think the only long-term effect will be a shorter lifetime for the thermal paste, but the difference should be small unless you're compiling 24/7.

If you're tuning for generic architectures then I do not think there will be any other side effects, but I do not know.

By "thermal paste", you mean the funny blue stuff I found on the CPU of my big 'ol laptop?
If I ever have to change that, I'll void my warranty... *shudder*

So, if what you're saying is true, I should not worry about compiling (as long as it doesn't take too long) and just keep an eye on the temperature.
Also, what temperature can be considered critical? Usually, my lappy runs ~65-70C.

If it really is an old laptop, then I guess there is no warranty left anyway?

Offline

#13 2013-10-19 12:56:16

mid-kid
Member
Registered: 2013-05-07
Posts: 177

Re: [Solved] Side effects of compiling on slower machines.

ronnylov wrote:

If it really is an old laptop, then I guess there is no warranty left anyway?

You misunderstand. My old laptop broke, now I have a silly lappy I bought for school. It still has warranty.

brenix wrote:

If you have multiple computers , you should check out Distcc if you haven't already..

Thanks for the recommendation. I actually used distcc before my old laptop broke.

sitquietly wrote:

I consider anything above 60 degrees to be too hot.  Much too hot.

I think I already said those temperatures aren't when compiling, and I'm looking to fix that.
Anyway, thanks for the recommendation.


I guess I forgot to state that the lappy is a netbook.
My computer sience teacher tells me that it's normal that it runs on such temperatures (because it's a netbook), and that when the warranty is over, I should replace the microscopic fan it has.

Also, since I'm curious, another question: What about compiling on ARM platforms? I don't see any thermal paste on, for example, a raspberry pi. How so?


If it ain't broke, you haven't tweaked it enough.

Offline

#14 2013-10-19 14:07:25

ANOKNUSA
Member
Registered: 2010-10-22
Posts: 2,141

Re: [Solved] Side effects of compiling on slower machines.

The over-simplified answer is that the Raspberry Pi only has a 700 Mhz processor and doesn't typically do anything processor-intensive.  The way that circuit boards are arranged and constructed for each architecture plays a huge role, too, but I'm no engineer.  For the record, though, compiling anything substantial on a Raspberry Pi takes forever, which is why both the Arch ARM and Gentoo folks recommend DistCC.

Offline

#15 2013-10-19 15:49:51

nomorewindows
Member
Registered: 2010-04-03
Posts: 2,944

Re: [Solved] Side effects of compiling on slower machines.

If you have a program that compiles your program using processor specific benchmarks, then as long as the architectures are backwards compatible it will still work, but won't have the optimization of the faster unit, if it is used on the newer equipment after being compiled on older equipment.


I may have to CONSOLE you about your usage of ridiculously easy graphical interfaces...
Look ma, no mouse.

Offline

#16 2013-10-19 16:08:36

ewaller
Administrator
From: Pasadena, CA
Registered: 2009-07-13
Posts: 12,265

Re: [Solved] Side effects of compiling on slower machines.

Anoknusa is on the right track.   There is actually an analogy to electrical circuits when looking at heat flow.   Essentially, heat is dissipated in the junctions of the transistors in the die.  the temperature of the die is a function of the heat capacity of silicon, and the amount of heat present.  If the die is insulated thermally, then heat cannot flow from the device and the temperature will increase without bounds (or until something melts)

If there is a thermal sink (Think the heat sink), and a thermal source (the die), and a thermal path between them, then heat will flow from the source to the sink.  The rate of heat flow is determined by the delta (difference) in temperature between the temperature of the source and the sink, and the thermal resistance between them.  Thermal resistance is a function of the material (plastics have a low thermal conductivity, metals and ceramics have high conductivity) and the area.  ICs are rated at maximum junction temperatures and define the thermal conductivity to the case.

So, the variables include the thermal conductivity of the processor, the junction temperature at which the processor runs (a function of the number of transistors and who fast they switch), the area on top of the processor on to which you can attach a heatsink, the ambient temperature around the heatsink, the size of the heatsink, the airflow over the heat sink, and the thermal conductivity of the goo between the heatsink and the case of the processor (BTW, Air is a sucky thermal conductor, so a heatsink without goo is worthless).   To make things even more complex, many modern chips have thermal pads under the parts that are intended to be soldered down to ground layers on the printed circuit board to allow the board itself to act as a heatsink.  These pads are impossible to solder without a surface mount reflow process.  I think the Ras-Pi has a processor with this type of thermal pad.   This type of thermal pad is not compatible with socketed devices.  In fact, sockets probably do a pretty good job of insulating the entire bottom of a processor.

Last edited by ewaller (2013-10-19 16:36:59)


Nothing is too wonderful to be true, if it be consistent with the laws of nature -- Michael Faraday
Like you, I have no idea what you are doing, but I am pretty sure it is wrong...Jasonwryan
----
How to Ask Questions the Smart Way

Offline

#17 2013-10-19 16:18:53

Xyne
Moderator/TU
Registered: 2008-08-03
Posts: 5,616
Website

Re: [Solved] Side effects of compiling on slower machines.

ewaller wrote:

In fact, sockets probably do a pretty good job of insulating the entire bottom of a processor.

I have found that you can easily use the board to dissipate heat by applying thermal grease to the socket prior to processor insertion. Using this method, the cores never go above ambient temperature and power consumption is radically reduced.* tongue








* In case this was not clear to someone, this is a joke, Applying thermal grease to the socket will damage the system and render it unusable.

Offline

#18 2013-10-19 18:13:25

ANOKNUSA
Member
Registered: 2010-10-22
Posts: 2,141

Re: [Solved] Side effects of compiling on slower machines.

ewaller wrote:

(BTW, Air is a sucky thermal conductor, so a heatsink without goo is worthless).

Mabybe I'm reading too deeply into this, but are you reccommending that some (small amount of) thermal compound be applied along the length and breadth of a heat sink, as opposed to just the contact point between the sink and the CPU casing? In many laptops, the heat sinks consists of a "conduit" running from the CPU to the fan and "fin" array, rather than the array of "fins" on a block found in desktop sinks. So it seems like compound smeared along that length in a laptop would act as a dust magnet, and could possibly drip to the board.  I don't know if that would pose any problems or not, I'm just really curious.

Offline

#19 2013-10-19 18:25:54

ewaller
Administrator
From: Pasadena, CA
Registered: 2009-07-13
Posts: 12,265

Re: [Solved] Side effects of compiling on slower machines.

No, not really.  The effective area is the area right over the die.  Since the package itself does provide some heat spreading, so it probably best to completely coat the smaller of the two surfaces.  Compound exposed to just air is pointless, as you point out.     What I meant was that a bottom plane of chunk of aluminum or metalized plastic(the heatsink), sitting on the top plane of a chunk of plastic or ceramic (The chip) are, for are practical purposes, separated my a small airspace (measured in 10s of  microns)  regardless, it is an airgap, and it dramatically increases the thermal resistance.  Both surfaces must be wet by a goo that is both airtight and has a high thermal conductivity.


Nothing is too wonderful to be true, if it be consistent with the laws of nature -- Michael Faraday
Like you, I have no idea what you are doing, but I am pretty sure it is wrong...Jasonwryan
----
How to Ask Questions the Smart Way

Offline

#20 2013-10-20 23:43:39

ANOKNUSA
Member
Registered: 2010-10-22
Posts: 2,141

Re: [Solved] Side effects of compiling on slower machines.

Gotcha; thanks for clarifying.  I really just asked because I'd only applied fresh compound on Thursday night, so I figured I'd apply the rest if need be.

Offline

Board footer

Powered by FluxBB