i am not saying that is a true power psu, but just sharing the experience when using that psu, since evoangel is using that psu.
i am saying modular have strong +12v rail as non modular, for sure i am saying when both were true power PSU.
what is the point comparing a proton with BMW

if open for 24/7, e8400 is better choice.
before i use q9450 4ghz 1.312v open 24/7, tnb charge me rm3xx
now change to e8400 3ghz 0.976v, rm150

ya, when i c the bill i also got frightened, 2 months i move back to my home the bill is rm3xx, more than rm350..
after change to e8400, rm150+-
idle at 4ghz really eat alot of power, even more than full load of my processor now..

mayb that time OC more often, full load processor more..

for sure not, the bill will show how much u use at that month, and this has been for 2 months after i move back to my hometown.

now i am runing e8400 at stock, never hit rm200 as well.

mayb u should try prime95 as well, because quadcore are not so suitable to test with linx, due to it can process much more faster than dualcore.

it was not same, in intel quadcore, i have tried before..
let say at 4ghz clock, if u use 1600mb of memory, a quadcore will finish 1 loop of ibt test at around 3x sec, while a dualcore will need 7x sec.
as u all know when using ibt, the processor will go back to idle before start to calculate another loop, so the quadcore might not fail in the 1st 3x sec and back to idle again.
i have tried a quad can easily pass ibt test at lower voltage easily, but failed in prime95 within 1 hour

yea, it chould be insufficient vnb, vdimm or vcore, but sometime it able to pass ibt 10-20 loops, but not prime95.

heat is not the main problem to cause the failure in OC, my processor reached more than 100c but still running in linx (due to poor thermal paste applied).
i do agreed linx is better in stability testing but only for dualcore after i tried it with quadcore..
maybe u all with quadcore can try it out and share the outcome, the lowest vcore to pass linx 10 loops and prime it for few hours

hmm, what are u trying to say about the heat? i never said that prime given more heat than linx.
just that prime able to test the stability for continuously, while linx are not able due to quadcore finished 1 loop in a short time period.
eg, at linx 1600mb testing
a quadcore at 4ghz finish 1 loop at 35sec, 1st loop(35sec) -> idle -> 2nd loop(35sec) -> idle
while a dualcore 4ghz finish 1 loop at 70sec, 1st loop(70sec) -> idle -> 2nd loop(70sec) -> idle
so the linx are continuous stress dualcore at 70 per loop, while quad at half of the time.

omg.. since like u have misunderstanding..
i never say prime is more stress than linx, no doubt that linx is most stress program available..
instead i am telling that the nature of the software is not so optimize for quadcore compared to dualcore.

lol, yaya.. we look into diff things..
usually i test linx with more than 1600mb, as cstk1 suggested before, at this setting it surely more stress compared to prime95.
but linx will stress the processor whenever it calculate the loops, after calculate each loop, the processor will enter idle stage before starting to calculate the next loop.
so the linx process will be load 100% ->idle 1-2 sec -> load 100% -> idle 1-2 sec
so in this case, a dualcore will needed 2x time compared to quadcore to finish 1 loop, so the linx will stress 1minute in dualcore, but 30sec in quadcore then the processor went to idle stage.
so in the 30sec, the chance for the error to come out is lower than 1minute.
and i said prime95 is better in this case is due to it stress all the way until the error come out or we stop them, although prime95 is not as stress as linx, but in longer stressing time, the error to come out has become higher
linx is good when the processor is slower, eg, a celeron single core will need very long time to finish 1 loop, eg, 10 minute, so it continuously stress the core for 10minute, while a 4ghz dualcore need more than 1 minute as well..
so imagine now a C2Q 4ghz need around 35sec to finish 1 loop in 1600mb linx, when faster processor coming in future, it will able to finish 1 loop in no time(assuming still using 1600mb in testing), there will no chance for the error to come out..
that y i said the nature of the software is not optimize for quadcore testing. but if using 4gb ram in 64bit with 3xxxmb in linx testing, it will be a diff story

45c is not hot at all if idle, if full load u got a gem
i guess your setting not stable, but not the temp. maybe cause by vdimm, vcore or vnb..

---

Added on July 25, 2009, 2:22 pmi guess this will depend on motherboard, some board when enable all this stuff will down volt, but some not..

This post has been edited by hengmy: Jul 25 2009, 02:22 PM

try ibt 100 loops with max ram or prime 8 hours

i think u dont need 1.4v for nb, around 1.3v will do for 500fsb quad at p45.
u might need set the vtt to 1.35v, and since 475 fsb for quad is not low, u might need to tweak the gtl ref.