JIM HOLSONBACK wrote in a message to ROY J. TELLASON:

 JH> Hello, Roy.
 JH> I've been meaning to write since way back on 11/21, when I
 JH> overheard you and Pascal Schmidt talking about - -

 RJT> Apparently the speed can go up there quite a bit,  the package was
 RJT> wrong.    Looks like if I were to get my hands on a K6-3 chip I
 RJT> could maybe go all the way up to 500MHz or so.  They do say that
 RJT> going to a FSB speed of 83MHz is "overclocking the chipset" but
 RJT> don't seem to think that's a problem.

 JH> OK, I've read about taking older Socket 7 boards and hardware
 JH> hacking to get around the cpu core voltages and clock multiplier
 JH> jumper settings which they were originally designed for, but not
 JH> for a PCChips board. The one I read about was for Asus P55T2P4,
 JH> which had Intel HX chipset. That internet info you got able to be
 JH> sent as an email attachment?  I'd like to get a look at it.

I think so,  if I can find it again,  or at least if I can find the link. 
I unfortunately tend to do stuff like grabbing lots of info here and there
and forgetting where it was that I got it from...

Somebody in here mentioned "M571" (I think?) and this was the
first or second thing that Google came up with,  a pointer to the site, 
and the first thing I saw on the page was a picture of the board I had
here.  :-)

I'll have a look,  see what I can do.

There wasn't much in the way of "hacking" required to get all
sorts of stuff working,  the people that sold the board didn't know that
much about the chips and how setting apparent multipliers actually works
differently for some parts.  I think *one* setup required a resistor to be
plugged across a set of pins rather than a jumper,  but you used a jumper
to make that happen,  sort of.  Kinda nifty,  actually,  and after reading
that stuff I have a better understanding of how that stuff works.

 JH> Not that I think there is much danger of you springing for a K6-III
 JH> chip anytime soon - - last I saw,  a K6-III 450 sold at Ebay for
 JH> $40 plus shipping.  But K6-2 of the speeds up around 450-500 are
 JH> sometimes now selling at Ebay for fairly reasonable prices.

I still have yet to get there.  Maybe I should give it a shot with my
current provider,  since I'm seeing better connect speeds?  Or maybe not, 
maybe I should wait until after I install the upgraded software that's
sitting here...

 PS> I think the 500 and 550 MHz K6-2 versions want a 100 MHz FSB (at least
 PS> my 450 MHz wanted that). Slower versions can do with a lower frequency
 PS> (66 MHz), using a higher clock multiplier. I don't know whether a 500
 PS> MHz K6-2 would want to run at 6 * 83 MHz.

That sounds about right,  I do recall the text I was reading on that board
mentioning an 83 MHz bus speed,  and how some peripheral cards might have a
problem with that.

 JH> Yes, K6-2 will let you play with the clockmultipliers - - I have a
 JH> K6-2 475 in this box (meant to run at 95 x 5) and I'm only running
 JH> it at 400 (100x4).  I can't say I recall ever seeing a Socket 7
 JH> board which had jumper settings for clock-multiplier of 6.  AFAIK,
 JH> the K6-2 line topped out at 550, and K6-III at 450.

That agrees with my recollection too,  I think (it's been a while),  though
there may have been a K6-500 mentioned in there someplace.

 JH> The writeups I saw about hacking and overclocking old Socket 7 
 JH> boards which were originally designed for chips down in the 200-233
 JH> MHz range strongly recommended mounting a cooling fan over the 
 JH> voltage regulator circuits for cpu.

One of the boards I'd been using had a pair of those regulators and one
time one of them got *real* hot,  I couldn't keep my finger on it,  it was
so hot. But then the next time I checked it out it was barely warm,  some
odd behavior going on there for sure.

I notice that some of these boards are using just linear circuits on big
heat sinks,  while others are using switching regulators,  at least for the
one of them.  It's when I see a coil nearby that I get a clue.  :-)  Those
seem to be the better-made boards,  too.  The one we're talking about is
one of those.

 JH> Of course, a lot of the faster chips (e.g. K6-2 500) draw a good 
 JH> bit more power than something like a P200 MMx, for which these 
 JH> older boards were originally designed.

That's something I haven't looked at yet...

---