Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sun-barr!newstop!sun!amdcad!brahms!phil
From: phil@brahms.amd.com (Phil Ngai)
Newsgroups: comp.sys.ibm.pc.hardware
Subject: Re: Trying 25MHz? (Speedup 386SX from 16MHz to 20MHz: SUCCESS!)
Message-ID: <1991Jan23.220011.22557@amd.com>
Date: 23 Jan 91 22:00:11 GMT
References: <1991Jan22.160249.5110@d.cs.okstate.edu>
Sender: usenet@amd.com (NNTP Posting)
Organization: Advanced Micro Devices, Inc; Sunnyvale, CA
Lines: 67

In article <1991Jan22.160249.5110@d.cs.okstate.edu> ong@d.cs.okstate.edu (ONG ENG TENG) writes:
|Well, a certain Mr. M (thank you) from Finland told me that his friend
|successfully sped his 16MHz 286 up to 25MHz with a small fan.
|Hhhhmmm... beside the CPU that is warm (at 20MHz), the other chips
|seems pretty cool.  The C&T chips, the other decrete components, DRAM's,
|seems OK.  The only thing that is hotter than the CPU is the *big* serial
|chip on my multi-serial card (that's the one chip you have to add for
|an extra serial port, I got 4 on mine for 4 serial ports).  Anybody 
|wanna stop me (yeah, like you would stop the astronunts (sp?) from going
|to Mars!)?

I don't want to stop you, but I think what you are doing is not wise.
However, it is your equipment to do with as you wish. It is your data
at risk.

|Could anyone define hot on a chip (that requires heatsink)?  I know
|when a chip is too hot (i.e. ready to burn out), but mine is kind of 
|slightly below that.  I can put my finger on it for minutes and my
|finger would not burn, but it is quite uncomfortable.

Most chips are rated to operate properly at a temperature that
you probably could not stand for more than a few seconds.
(max ambient temperature of 70 degrees C. the military stuff
can go to 125 degrees C, or hotter than boiling water.)

The temperature of a chip is not a sure-fire indication that it
is not being run beyond specification. Nor does running a single
chip within specs mean that the board will function as designed.

Your obession with temperature and clock rates is misdirected.
A chip is not like a mechanical device which overheats when
it is overloaded. The real dependences are as follows:

1) Most chips' maximum speed of operation decreases with
increasing temperature.

2) Most chips' power dissipation is a roughly linear function
of clock speed.

3) Most chips' maximum speed of operation increases with
decreasing temperature. (this is the same as 1 but I wanted
to be explicit about this)

4) Most chips' are tested to a small margin beyond their specifications
to allow for the lack of perfection in the equipment used to test them.
We want to be sure that "even if our speedometer reads 5 MPH low, we
are still within the speed limits". And, they are tested at the
maximum temperature specified. So at room temperature, there will be
some margin.

But how much? Unless you are an engineer and fully analyze the circuit,
you don't really know if it will still be reliable after your changes.

As far as I'm concerned, it is foolish to risk your valuable data
to this kind of folly. Although I work for a company which makes
semiconductors, I don't speak for them. I don't think they would
disagree with me, however.

There are companies that make chip coolers (Velux). However, they
don't intend that you just crank up the clock rate. The entire
design needs to be analyzed and any timing problems corrected.
Otherwise you are gambling with your data.

--
When someone drinks and drives and hurts someone, the abuser is blamed.
When someone drinks and handles a gun and hurts someone,
the media calls for a gun ban.