Search:

please don't use that measuring spot if you don't have True RMS DMM.
You'll get 0.1-0.3v lower values..
vddr from bios = real vddr. there's no droop whatsoever.

So, if unsure about your DMM, use...

ok, not cheap, but not a trueRMS one :)
and if you don't have a trueRMS dmm, then measure those points for AC, if there is any present, then find another spot.

nothing wrong with voltages, you are using to cheap dmm and measuring from a wrong place for a cheap dmm ;)

try these points:

http://alpha.frontier86.ee/~markku/975/vmch.jpg

well it's still stepping B2..

yup those 700s ran at 933 at stock.
but weird thing was, all 700s i tried out, none of them were stable @ 1ghz :)

btw kunaak, the idea of NBCC was discarded. even by Tony ;)
but nice to see that bsel trick has an use now with those 1333 fsb bioses..
when i tried it a while back, i had asus p5wdh and of course...

oh, never realized 4M conroes had 16-way assiociativity..
it's a pretty old trick for intel.. starting from old coppermine celerons, which had less cache and also smaller n-way assiociativity than...

answer lies in coreduo datasheets :-P
just pinmod 133 fsb chip to 166 and thats it.

if $10 apart, then go for the better model- oc'ed difference will be 2ghz vs 2158mhz.
just make sure you buy a...

plus it's a single core.. that's why weak sandra scores.
a celeron is a celeron.

i think conroe L has also reduced cache assiociativity, like all celerons.
and that's the main thing hindering perfomance.
e4000 series are still using allendale core.
there's going to be even 1M...

475 is quite nice for default nb cooler.. it's about the same i got with 6400/watercooled nb and upto 2.3v.
there's no ovp that i'm aware of, but after 1.95v (or ~1.85v for 1401 bios and up), pcie...

you can further lower voltage using crystalcpuid.
~0.8v was the lowest i could go if i remember correctly.

some use custom heatsinks for mosfets (incl me), but tbh they don't get hot at all, the heatpipe is cooling nb, not mosfets..

thanks yotomeczek.. can you run also 32M or at least 8M like loc.o did?
thanks again

by the way, does anyone have access to both bad axe and p5b deluxe?
would be very interesting to see a comparison, first p5b 399 boot vs 401 boot and then bad axe 1066 strap vs 1333 strap.
i don't...

what was your limit before that bios?

thats normal, asus mobos don't update DMI after multi or fsb change, so most programs can't read proper cpu speed.

if hp3 is auto, then, at some point, p5w will disable it.. mostly :)
btw, did some testing with hyperpath3, compared mch configuration registers, both hp3 on and off.
the only thing i noticed...

vmch doesn't have vdroop btw.

but anyway.. if not going subzero then p5w dh for e6600 and up. p5bdel for everything else.
or p5b all the way :)
one thing's clear - don't get p5w if you'll gonna...

menlatin, how about installing uniprocessor hal?

there's no need for that with your p5w.. booting up at 266mhz and selecting appropriate mem ratios, effectively forces 1066 strap on nb.
though with boards with no such logic in bios, bsel mod...

this is true that c2d cpus have fsb limits..
with dothans, to overcome this limit, one could simply increase vcca.
with core2duos, vcca is connected to vtt, according to kyosen.
but i don't...

change fsb, then reopen sandra and refresh as many times as you like ;)
btw, most of the information, sandra aqcuires from DMI.
it reads clock speed from dmi, detects multiplier and calculates fsb...

this should do:
simply change fsb with clockgen/setfsb and see if sandra detects a change.
it detects not.
quote from 975x datasheet: "The MCH core clock is synchronized to the host clock."
So if...

and if you search further, you'll see that this is not NB clock changing, but sandras inability to read proper cpu speed = bug :)

nice..

any benches/screenies or did it hang at post?