Linux on MacBooks and CPU voltage - can it destroy my CPU?

litemotiv · 2010-03-05 21:57:54

Peasantoid wrote:

** The guy's probably an Apple meat puppet, come to think of it...

and they're probably laughing their asses off right now in ##mac

frigaut · 2010-03-08 19:21:06

...and if the guy is not a troll, you are unnecessarily insulting him... Unless you're sure, you shouldn't cast these kind of opinions. It's nothing else than replying to possible FUD by FUD.

Anyway, let's poll: I know of 2 MBP (5,1 and 5,2) than have run exclusively linux for respectively 14 months and 7 months.

shadewind · 2010-04-18 00:17:12

I'd like to revive this thread. Has anyone come to any conclusions whatsoever about this?

AbtZ · 2010-04-19 08:54:18

Would just like to chime in and put a rest to this. Basically he claims the electron tunneling phenomenon causes permanent damage to the chips. As such he clearly has no idea what electron tunneling is*. Furthermore CPUs aren't even small enough (yet) for electron tunneling to be a problem.

Dismiss it as FUD.

See this post for more info:
http://gromit.dlf.org.au/pipermail/dlf- … 05651.html

*For those interested it has to do with quantum physics and electrons existing in probability clouds, which extend a few nm into materials -- not them physically burrowing through stuff. http://en.wikipedia.org/wiki/Quantum_tunnelling

Zeist · 2010-04-19 11:15:37

I run arch randomly on my macbook and I haven't seen any issues. In fact the only real difference in power is that with some playing around with linux power saving settings I can actually get longer battery life on linux than on os x.

geniuz · 2010-04-20 08:30:14

Doesn't seem to make any sense to me either...and why am I not surprised this comes from an Apple guy? Running linux on a Macs is nothing new and I think a LOT of people are doing it, so I think by now there would've been some serious report claiming there's something not quite right with running linux on Macs. Besides, show me ONE case where this might have happened...seriously I've been scavaging google for some time and I couldn't find one person claiming his CPU died because of running another OS. Also, even if Apple didn't bother to inform its users of this flaw, I'm positive a company like Intel would.

A quick search also revealed that this is probably BS since I've read that firmware regulates these sort of things in OSX and not the OS itself. And SHOULD this is in any way be true (which if you ask me is very unlikely) it is a serious flaw in the design of Apple machines and by now a lot of people running Windows only on their Macs would've complained...but again, no investigations or claims what so ever seem to have been made on this issue.

Come to think of it, if this sort of stuff would be handled by the OS, wouldn't there be like a MASSIVE amount of vague software available to o/c your CPU and things like such...? And if you honestly deleted your Arch partition because this guy said so, I think he's probably lhao at the moment.

I'm not really an expert on this matter, and since I just installed Arch on my MacBook last week I cannot confirm like others my machine hasn't died. But I base my opinion on the plane and simple laws of information, if one guy says something is wrong and I can't find ANY piece of information backing it up what so ever (quite the opposite actually) then I just assume its BS. Common sense right?

EDIT:
I've found this thread after googlin electromitigation and some people that seem to know much about the matter discuss it, what I could get from it is that electromitigation is indeed a problem, but its effect is only seen after very long periods (we're talking years, sometimes even decades in very extreme cases only) before it becomes fatal. Anyway here's the source and make from it what you want (look at post #16):
http://forums.anandtech.com/showthread.php?t=1558653

Also in the following article there is a short definition of electromitigation and a discussion, bottom line here is that it is very hard to detect an even harder to predict. So that totally busts his statement of the 9-12 months to failure:
http://arstechnica.com/cpu/oc-myths-2.html

The more I look into it, the more I realize is that this phenomenon is really vague, I mean it isn't even given that this process will INEVITABLY occur...from what I make of it there is just a probability that it will occur and this risk rises with voltage (this is assuming there even is an increased voltage as this guy states). So to wrap it up, the more I research on the subject the more it confirms my feeling that this is FUD, simply because this phenomenon is hard to detect and predict and therefore you can tell anyone anything about it and they will never be sure. You will only scare them, make them feel uncertain and eventually they will even start doubting before they decide to only use OSX.

Last edited by geniuz (2010-04-20 12:48:46)

shadewind · 2010-04-23 13:50:49

If anyone is still wondering, I've made some research about Boot Camp and its drivers:

First, there's a bunch of regular PC drivers from manufacturers like Nvidia, Intel, Realtek et.c., nothing special there. These are things you can download directly from them if you'd want to.

Secondly, there's a folder for Apple drivers with expected stuff like the trackpad, iSight, keyboard, stuff like that. The only mysterious driver here is the Apple Null Driver. When checking it's contents, there's no binary files at all. So there's no real drivers in it, just an INF-file which just gives Windows some names for devices, among them the SMC chip.

Thridly, we have the BootCamp.msi package which is the installer itself. Here is where it starts to get interesting. Inside that package is a bunch of files, AppleOSSMgr.exe, AppleHFS.sys, AppleMNT.sys, BootCamp.exe and finally MacHALDriver.sys. AppleOSSMgr is an app which chooses the booting OS for the next reboot. AppleMNT and AppleHFS are file system related things.

BootCamp.exe and MacHALDriver.sys is what's interesting. A disassembly of MacHALDriver.sys shows that it does actually provide access to the SMC as there are calls to Windows port I/O functions with the 0x300 and 0x304 addresses, the same ones listen in the linux applesmc.c kernel driver. It does not, however, do any of this on it's own and is merely there to provide access for userspace applications.

The userspace application in question is BootCamp.exe which uses the MacHALDriver. SMC values are accessed by 4 letter "keys" that can be read and/or written. There are only a couple of these these keys that BootCamp.exe actually accesses and these are: ALV0, ALV1, ALI0, MSLD and LKSB.

ALV0 and ALV1 reads the value of the ambient light sensors. ALI0 provides information about light sensors. LKSB changes the value of the display backlight. MSLD reads the value of the "clamshell key".

This is nothing weird of course. BootCamp.exe reads the values of ALV0 and ALV1 and sets the value of LKSB to an appropriate value. If MSLD is read, it puts the computer to sleep. It does not, however, touch anything else at all. No fans, no voltages, no power management, nothing. Here's proof that there is nothing that Boot Camp does that Linux can't. Actually, the Linux SMC driver can set fan speeds, read temperatures and a lot of other things tha the Boot Camp driver cannot.

Add to that, it's not even possible to set voltages through the SMC. Here's a complete list of SMC keys with descriptions:
http://www.parhelia.ch/blog/statics/k3_keys.html

If you look through that list, you can only find two values related to the CPU core voltage, VC0c and VC0C. It's rather obvious that these values are only for monitoring purposes and cannot be modified. The hint is that VC0c is raw ADC _input_. The VC0C value is an interpretation of that to a real temperature scale. Although both are listed as K_VAR_ATOM_RW which would lead one to believe they are modifiable (RW), it's not uncommon for input registers on microchips to be writable. The only thing that happens is that the value is changed but the next time the real value is polled by the chip, it's again replaced with the actual value.

Hope this helps

litemotiv · 2010-04-23 17:17:11

Excellent information shadewind, thanks :thumbs:

Peasantoid · 2010-04-23 19:05:01

Aha! Thanks for the clarification, shadewind. Finally we can dismiss Branes as a) misinformed or b) FUDing.

Hodapp87 · 2010-05-17 01:55:40

Peasantoid wrote:

Aha! Thanks for the clarification, shadewind. Finally we can dismiss Branes as a) misinformed or b) FUDing.

I have dealt with Branes in the channel many times. Almost everything he says about Linux or its users is insults and FUD, founded primarily in his irrational hatred of open source down to its very concept.

augustl · 2010-11-24 10:05:55

Thanks shadewind, great research!

Arch Linux

#26 2010-03-05 21:57:54

Re: Linux on MacBooks and CPU voltage - can it destroy my CPU?

#27 2010-03-08 19:21:06

Re: Linux on MacBooks and CPU voltage - can it destroy my CPU?

#28 2010-04-18 00:17:12

Re: Linux on MacBooks and CPU voltage - can it destroy my CPU?

#29 2010-04-19 08:54:18

Re: Linux on MacBooks and CPU voltage - can it destroy my CPU?

#30 2010-04-19 11:15:37

Re: Linux on MacBooks and CPU voltage - can it destroy my CPU?

#31 2010-04-20 08:30:14

Re: Linux on MacBooks and CPU voltage - can it destroy my CPU?

#32 2010-04-23 13:50:49

Re: Linux on MacBooks and CPU voltage - can it destroy my CPU?

#33 2010-04-23 17:17:11

Re: Linux on MacBooks and CPU voltage - can it destroy my CPU?

#34 2010-04-23 19:05:01

Re: Linux on MacBooks and CPU voltage - can it destroy my CPU?

#35 2010-05-17 01:55:40

Re: Linux on MacBooks and CPU voltage - can it destroy my CPU?

#36 2010-11-24 10:05:55

Re: Linux on MacBooks and CPU voltage - can it destroy my CPU?

Board footer