I'm posting this to the group and to Mr. Mika personally.
Some may start screaming that it is off topic for the newgroup.
But I have to argue that fact.  Since overclocking cpu's and generally
upping performace is of MAJOR interest to ANY playing with povray..
I say it fits RIGHT IN!  Sue me!


Nieminen Mika wrote:

>   Overclocking seems to be an obsession. As soon as intel (or whoever)
> publishes their new cpu, someone overclocks it.

        Agreed.  There's people out there with money to burn that simply
buys new motherboards/cpu combinations just to see HOW fast they can drive it.
Then publish the results.  (search www.metacrawler.com for overclocking)


>  Because of this, people seem to think that they can just buy a cheaper one
> and overclock it to the same frequency as the expensive ones. They never
> seem to think why they are selling it with a lower clock frequency and much
> cheaper.

    To a point.  I don't believe because a chip is rated at say,  100 mhz,  that
it won't
RELIABLY run at 133. (or possibly higher?)  Intel and other chip makers don't
rate their
chips RIGHT at,  and UP TO the the point of failure.  very often they leave such
a WIDE
margin of error that,  well,  that is what the overclockers are PLAYING with.


>   The reason is obvious: Because it doesn't work perfectly with the higher
> frequency. They sell it with a frequency which is suitable for it. If it
> worked fine with the higher frequency, of course they could sell it with
> that frequency and receive more money.

        Uhhh,  that is EXACTLY what Intel did. In fact,  it has been PROVEN that
the 150 and 166 original
Pentiums came from the same batch.  They needed so many for each market.
Difference in chip?
NOTHING.  NOTHING AT ALL.


>  So when you overclock your cpu you are actually setting it to a level
> which have been detected as unstable.

        Yes,  agreed.  But there ARE certain things that overclocking involves.
Things that INTEL did NOT consider and put into their "test" conditions.
For instance,  Intel for obvious reasons must test their chips prior to shipment
under worst-case conditions.  Such as HEAT.  One of the "tricks" to overclocking
a machine REALIABLY is HEAT.  You cannot drive a chip faster than what it was
intended
for,  and have it HEAT up considerably.  You bet it will fail.  But very often
people will
drop 20-30 dollars and buy a LARGE fan and REALLY large heat sink to go with it.
Result?  A chip that is running at almost room temperature.  Intel can't assure
that the chip
will ALWAYS run at "room" temperature.  Again,  they must plan for worst case.


>   This unstability has not necessarily to be immediately visible. Everything
> may seem to work just fine. You use the computer for weeks or months and
> notice no problems.

        I have to agree with you there.  But understand,  this overclocking thing
has become
such an "art form" to some people,  that utilities have been written that will
run the chip
through some pretty demanding chores,  repeatedly.  If there's a failure...
providing the program
is ran long enough,  it WILL APPEAR.  (but yes,  I have seen random things go
wrong that I
definitely attributed to speed increase.  I guess it's the chance you take.


>   And all this for what? I once read a comparison in speed with various type
> of programs with and without overclocked cpu's in a computer magazine. It
> really looked like the programs were running a lot faster, but they just put
> the absolute values, not the relative values (ie. the program runs n times
> faster in the overclocked than in the regular cpu).

        Depends... sounds to me like that article didn't have it together.  And
really had NO
idea what was involved.  I jumped my CPU from 150 to 166.  Speed increase?
Not in any real world applications.  Math intensive I did... such as povray.
But a jump of 16 mhz isn't really nothing at all.  And povray didn't report any
blazing
speeds.  (but then again,  does it EVER???)  However,  there is a LOT more to
"overclocking"
than just the chip speed.  There is also a matter of driving the BUS at a faster
speed.
And THIS combined with faster chip cycles is where it really starts to pay off.
The standard bus speed a few years ago was 33 mhz.  (note,  I did NOT say CHIP
SPEED)
People started taking the old original pentiums and upping the bus to 50 mhz.
NOW THERE
is a speed increase.  PERIOD.  Take the normal pentium,  jump it from 150 to
166.  Up the speed
of the bus from 33 to 50.  Believe me,  your going to see more than a 1.2 speed
increase.  More like,
2.0 x and possibly higher.  (You'll even see speed increases in the transfer from
the hard drive)
Bottom line,  Unless you can make the chip run 100+ mhz faster,  overclocking
in,  and OF itself
isn't worth it.  However,  if you can manage to speed up the bus AND speed up the
chip at the same
time.  YOU BET it's faster.  A LOT faster.


>
>   Often people think that the only problem with overclocking is the increased
> temperature of the cpu. If this was true, you could take a 8086, deep-freeze
> it and overclock it to 500MHz. Of course it doesn't work, no matter what the
> temperature is.

        Uhhhm,  your not going to believe this.  But that was done already.
And yeah,  it allowed them to run the chip at outrageous speeds!
What they did was take something like a 486 or something rediculous...
(I know it wasn't a pentium)  and hook a compressor up to it.  They
somehow surrounded the chip in a casing,  then used something like compressed
ammonia or something to super cool the chip.  (if I remember correctly,  they
lowered the
temperature to something like -30 degrees F )  They were able to (reliably) run
this
486 or whatever it was,  at something like 600 mhz.  It was a pretty interesting
article.
And as far as heat is concerned,  it is a big threat.  The heat generated alone
can literally
cook the chip.  Although there are other things to worry about.  That is the
first symptom
over overdriving a chip.  And must be contended with.

>
>   No thanks. I have never overclocked my computer and I never will.

        I understand your fear.  Some of us though have seen this done realibly
SO MUCH,
that it doesn't worry us.  Then again,  some of us,  simply don't have anything
to lose.
(are machines are so old it doesn't matter to us)



                                Happy tracing and best regards,

                                              Johnny

Post a reply to this message