> >Generally, overclocking is going to cause some problems with something as
> >sensitive to interrupt timing as Linux, but I'd guess the problem is that
> >your using dog-slow SIMMs. Get some 60ns SIMMs and see if that doesn't clear
> >things up (>90MHz should use 60ns SIMMs).
> >: Any comments?
> Yup!
> My 100MHz Pentium runs perfectly well with 70's. In fact, it runs the
> very same speed with 70's as it did with 60's. While overclocking the
> chip is probably a bad idea all the way around, especially due to the
> potential damage from heat generation and the very fact that the chip
> is certified for exactly the speed that it't rated at, I suspect that
> there is too much worry invested in DRAM speed. I've seen perfectly
> fine DX4/100's running with 100ns chips. I've seen 'em work fine with
> 60's. No special tweaking of the BIOS, nothing fancy.
I've got to disagree with the worry of doing any damage due to heat.
AFAIK, the Pentium 75-100MHz series use the same exact die (same chip
inside). The only two differences are that Intel has tested each chip to
come up with speed ratings (this is debateable) and the markings on the
chips are different, depending on what the speed tests say. I have
looked at the Intel Pentium data book, and the packaging for the three
different CPUs are exactly the same. I reasoned that if one chip ran
hotter than the other, they would have some differences to the
packaging, such as a wider and thicker ceramic base. But there's
absolutely no difference, and by that logic, they should run at the same
temperature. I could be wrong on my reasoning here, as I'm not at all
knowledgeable about chip design, so correct me here if I'm wrong. The
real conclusion came when I took a digital thermometer and put it to my
heat sink and let my machine run at 75MHz, 90MHz, and 100 MHZ, each for
an hour. BTW, I've got a PB 406CD w/ a Triton chipset and an alleged
75MHz Pentium, with a passive heatsink (i.e. no fan on top, just fins,
although there is a fan built into the case to blow over the heatsink).
The differences in temperature were negligible and the small differences
didn't suggest any pattern to the `faster=hotter' theory. During the
test, I played Mechwarrior 2, which eats raw processing power to do the
3d graphics. Also, it was running under plain DOS, which unlike Linux
and OS/2, doesn't put the CPU to sleep when idling, therefore the CPU
always runs at full blast. I should run this test again, but instead of
speed, I should test for CPU temp for DOS/WinDoze vs. Linux, both under
various loads.
All and all, I've been running my machine for more than 6 months at
100MHz, 33% faster than the rated speed with absolutely no weird
glitches at all.
Plus, my machine stays on 24 hours a day running Linux. I hardly ever
shut it off.
Quote:> I do agree, however, that overclocking the chip, coupled with a
> sensitive, high performance OS will probably bring out any limitations
> in the system. I think that the best advice is to run the chip at its
> designated speed. Want a faster CPU? Buy one!
Yeah, yeah, yeah. I've heard this same thing from a few jealous friends
who upgraded to Pentium 100s around the time I did my overclocking. I
did the above experiment just to prove to them that there's no real
danger to overclocking a Pentium 75-100 series CPU. People have been
overclocking their CPUs for ages. There used to be an overclocker for
8088s that could get you to (if I remember right) 12 MHz on a 8MHz rated
CPU (8088-8). Macs have also had their share of overclocking methods,
mainly replacing the clock module and a few other parts. It's no biggie.
Intel has to have a cheaper, inferior model chip to make the higher end
one worth more. It's cheaper for them to accomplish this by speccing out
their chips from one assembly line with false results, depending on how
many of each model is ordered, than having 3 separate assembly lines.
Intel does, however, have some kind of low-pass filter on the clock
input of the Pentium 75-100 chips that will keep you from running it
over 100MHz. Supposedly, though, people have clocked Pentium 120s up to
~200 MHz with no problems.
I do agree, however, that overclocking your machine is not for the
casual user. I did this in the same 'do it yourself' sense that makes me
love Linux so much. Sure, I could do everything that I use my computer
for in Windoze and not have any setup hassles, but it's not as much fun,
and when I'm done getting something in Linux to work, I've got a great
feeling of accomplishment that I made something work better than it used
to. But, I only do this because I feel that I can handle this kind of
work. I would not feel sorry for the person with no technical experience
that sets their CPU speed switches to 100MHz and their Pentium is
actually capable of only 75MHz (i.e. Intel tested it to be under spec,
and it truly is), then brings their PC in to be repaired claiming it's
`broken'.
*rant mode off*
My real opinion is that it's quite OK to overclock your CPU, provided
you can run it for any length of time without any glitches. I am not
saying in any way, shape, or form that overclocking your CPU won't make
your computer die or malfunction in any way. If you overclock your CPU,
and it works, good for you. You've just saved a lot of money. If it
doesn't work and your computer doesn't work like it used to, that's your
problem. My experience has been that it works.
+--------------+--------------------------------------------------------+
| Andrew Litt | "You've got to be kidding me! Are you suggesting that
|
| KC5APJ | soup is a biped?"-Crow T. Robot, MST3K
|
| MSTie#47556
+-----------------+--------------------------------------+
Dwarf,HC11,TMBG,|
|FZ/MOI,K2BSA,HP48,Newton,Phish,X-Files|
+--------------------------------+--------------------------------------+