How many of us ever test multiple boards to compare sw vs. bios
overclocking? Indeed, who buys lots of the same model of board
and then does the same tests to show what variation there is from
board to board? Nobody I reckon, in which case anyone saying in
an absolute way that one method is better than another is not
speaking from any solid statistical base. Opinions are formed
based on personal experience, and we remember our bad experiences
much more than our good ones.
I've read many reviews where a particular board was described as
not being that good for overclocking, but then read forum posts
where some people have obtained very decent results. Likewise, a
review praises a product but posts appear from people who bought
the item but had big problems. There's clearly great variation
between multiple units of the same product, never mind different
products. For example, when I bought two X1950Pro AGP cards in
late 2006 (stock core clock of 580), they came from the same
batch and I fitted both with the same better cooler (ACCELERO X2);
however, one card would oc easily to core/mem of 641/783, whereas
the other would only go to 621 with a barely changed RAM clock.
It's easy to say a particular method is best, but in almost
every case such comments are based far too much on limited
individual experience with a small range of boards, sometimes
even just one board. The comments posted here already show this
to be the case, ie. people claiming one or the other is better
while others give examples where each method has worked well for
them, or badly.
In the end, who cares? Try both, see what works best for you.
It's not as if the method one uses is any kind of fashion
statement; I get the impression there are some who feel that
sw overclocking is somehow not 'real' overclocking, that in
some way it's too easy. If anything, making it easier to do
is a good thing as this will broaden the base of those trying
to oc their systems, which will drive demand for parts that
can take it.
Ian.