Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 22 Jul 2000 17:18:03 -0500
From:      "Jeffrey J. Mountin" <jeff-ml@mountin.net>
To:        "Rodney W. Grimes" <freebsd@gndrsh.dnsmgr.net>
Cc:        lioux@uol.com.br (Mario Sergio Fujikawa Ferreira), freebsd-stable@FreeBSD.ORG
Subject:   Re: Signal 11 on 4-Stable buildworld; bad memory or what?
Message-ID:  <4.3.2.20000722162331.00d5ec70@207.227.119.2>
In-Reply-To: <200007211953.MAA39777@gndrsh.dnsmgr.net>
References:  <4.3.2.20000721132938.00b49400@207.227.119.2>

next in thread | previous in thread | raw e-mail | index | archive | help
At 12:53 PM 7/21/00 -0700, Rodney W. Grimes wrote:
>Note also that OC'ing can cause _permanent_ damage to semiconductors,
>and that clocking down to spec may still leave you with hardware that
>does not function correctly.  The predominant failure mode is caused
>by something called electromigration that has to do with metal migrating
>into the semiconductor due to operating at excessive temperature.

Certainly, but excellent cooling is a must when OC'ing, even if one is just 
doing a quick test to see if the CPU can be OC'ed.  Electromigration can 
happen when running a CPU within specs, except for sufficient cooling.

>Electromigration is a non reversable process, and it only takes a few
>microseconds to occur should the temperature at a metal/silicon interface
>rise above a certain threshold, this spot heating occurs on the microscopic
>level and external temperature monitoring has very little, if any,
>visibility into it.  This problem becomes more and more critical as
>geometries shrink.

A design flaw can also attribute.  My C instructor ran into to this years 
back with an Alpha when using 2 registers in a tight loop.  The layout had 
them adjacent on the chip, resulting in excessive heat explaining the odd 
failures.


Didn't care to add more to OC'ing, but what the hey.
Just run for cover afterwards.  ;)


OC'ing non-Intel processors has shown the smallest gain.  As mentioned on 
the list they already run hot.  Trying to squeeze more out is asking for 
trouble.  A good OC'ing text *should* mention this.

While I no longer have any OC'ed servers, I do make sure to have better 
processor cooling and prefer 2 fans per CPU.  Should one fail, the cooling 
is sufficient, since most are designed for OC'ing.

http://www.computernerd.com/

I've used products from the first for some time now.  The FC-PGA pages are 
buried at:

http://www.computernerd.com/futu33.html

Those that like to OC may wish to check out:

http://www.computernerd.com/futu36-1.html

For links to various cooling suppliers check out:

http://www.heatsink-guide.com/



Jeff Mountin - jeff@mountin.net
Systems/Network Administrator
FreeBSD - the power to serve



To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4.3.2.20000722162331.00d5ec70>