Yes, good discussion is great - and a learning experience. See stuff (-->)
below.
Lawrence said:
Hi Chris,
While I agree with your remarks about potential weirdness, a little decent
testing, and incremental advances avoids most of that. Besides, 'weirdness'
is just as capable of cropping up in a 'factory' box as an OCed one. That's
partly why this news group exists. If everything integrated perfectly,
there'd be nothing in here to talk about.
--> Correct, however when Windows XP installs on a system it tracks the
components and what each item does according to the specs set by the
manufacturer. Future operating systems will get even better at it. When
one of the conditions becomes a variable, that's where trouble can come in.
It is like in beta testing programs or games, valid problems in the test
software are only true if the system is not over-clocked. If it can be
reproduced without over-clocking then it is a program bug. If not, guess
where the fault is placed?
But I pretty much totally disagree that XP *necessarily* runs best with
everything set to 'factory'. When you consider that CPUs are stamped out
of one big wafer -and that where the individual chips were in relation to
the center of that wafer (center chips being more stable at the highest
speeds, while edge chips tend to be more stable at lower speeds), then the
entire industry is based upon a spectrum of performance.
--> Granted, some companies produce "over-clock" chips with some point of
leeway, but are we really good enough to know when we've reached that point?
More power, or in this case running a chip beyond the recommended capacity,
isn't necessarily "better." What's gained? Bragging rights, or can one
really detect milliseconds of increased response time? The two major CPU
manufacturers can't even agree on how to describe what is important.
So, if you are willing to go slow with your overclocking efforts, and not
crank the chip right out of the box, you can get some really sweet results
using tried and true methods that tens of thousands of overclockers
successfully use, the world over.
--> I agree. But where do you draw the line whether a program is crashing
is the result of a faulty graphics driver or because the card has been
over-clocked. How do you know?
It's a Bell curve of performance. You just have to look for the sweet
spot.
--> For what, again? Beside the issue of Windows XP recognizing which
chips you're running and tuning itself for the factory specs on that
particular item, what is gained? Shortening the life a CPU, graphics card
or RAM? I've honestly never understood how friends of mine - people who
over-clock everything in sight - can be buying three CPUs a year because
they burn out, for what? I really notice no difference in performance,
except of the individuals as they brag to each other, or complain about
different companies and the products being too fragile.
--> The main difference I've noticed on machines and performance is
increasing the RAM. (And don't try it on Windows XP with mismatched
sticks!). Bumping a system up to 512 from 256 will make a huge difference.
Thanks for posting your thoughts. I really appreciate kicking these ideas
around.
--> Glad to have the discussion. I hope you can enlighten me. I never saw
the "advantage" with Windows 95, etc., etc., etc., and still don't.
