Foxconn Bloodrage-Quantum Force- -X58- -Retail-

[saaya]

The button on my battery back up is the only I can turn off power to board now my case button has no effect either.When I shut windows down my fans and lights stay on but monitor goes off .Here is a pic of 150w result .Still stuck at 24.I didnt really explain properly.I tried several settings for tdc tdp even higher than 150w

hmmm thats odd... so... you can use higher multis but under load it drops to 24? or you cant even use higher multis at all?
weird...

Saaya, a hell of a long post but I fully understand it now, thx a lot

well you understand what i mean
that doesnt necessarily mean its the truth, i always recommend taking everything with a grain of salt and question everybody and everything, even, and especially, the proven things and people! cause those are the things and people that tend to not get questioned.

Ill keep it disabled then since there's basicly no reason to have it enabled. I did have it enabled but the difference between idle and load was in some cases huge, 0.5V! Although, idle and load Voltages were higher to start with when it was disabled, but FWIW that's all read by CPU-Z.

hmmm but i think cpuz eads pretty accurately on BR... can you please try with CxE disabled and vdroop enabled and check? it should be much better then

Im slightly dissapointed in my system though Ive beautfull components, but the Uncore refuses to work at high speeds 4.3Ghz requires quite some Volts compared to 4.2Ghz, but I tried hard getting 20x210 instead of 21x200, but it's a no go. I tried lowering Uncore multiplier (with RAM being 50% of that of course) which worked, but what's the point of DDR1260 RAM

yeah uncore is a big limit for i7... im sure without that 2 to 1 limit to mem clock we would be seeing DDR3 3000 already... at huge timings, but still...

heres a hint, i heard 1156 has a 1.5 to 1 limit, so higher mem clocks with lower uncore will be possible

from what ive seen the realworld impact of mem ocing on i7 is tiny... ocing uncore seems to be a bigger boost than ocing memory itself.
and in some scenarios low speed tight timings seem to be better and other benchmarks/apps pefer high speeds with lax timings...
on x48 you could run low speed tight timings and high speed high timings and get about the same perf... the fastest was 1800 cas6, the second fastest was around 2000 cas7... i havent played that intensively with i7s imc yet, so im not sure about this stuff with i7... but i saw that tony from ocz focussed on low speed tight timings instead of high speed high timings when tweaking his i7 system...

from what ive seen most seem to aim for 2000+ cas7, still, same as on x48... so for that 4ghz uncore is more than enough... if you want more then you need to go subzero afaik... and you need a good cpu, the newer ones can run notably higher uncore with lower vtt, ive heard of 4ghz on water with 1.2v vtt not sure if that was 100% stable though...

The UnCore frequency depends on the Vram ( DDR3 Voltage ).
Higher ram voltages allow the UnCore to gain stability at higher frequencies.
And of course, the maximum UnCore stable frequency depends on the CPU itself and not the board.
And keep in mind that there's nearly a wall on every i7's UnCore freq on air, restricted by the temps ( need to go subzero to be able to get pushed any further ) and not the voltages.

vtt and vdimm, yes...with good mem that doesnt need high vdimm you can reach huge speeds without much vdimm, so in my experience vdimm is mostly needed if the mem needs it, not the uncore. but its similar to x48, more chipset voltage got you a higher mem clock stable or the same clock stable with less vdimm. its about the same with vtt and vdimm on i7. naturally... its most likely the same memory controller...

and the x48 memory controller stopped scaling around 1.8v, and for i7 it seems to be around 1.7v, slightly less...

and just like with x48, the first chipsets needed high voltages, some even 2v, and the later the stepping the less chipset voltage they needed to be maxed out, the latest x48 chips can do high mem clocks with 1.4v... same with i7, the first chips needed high vtt for high uncore and mem clocks, the more recent ones can do it with much less vtt...

Im on water

But... in what terms of Vdimm are we speaking here, cause 1.75Vdimm didnt do it Also wtf is the scaling of Uncore then? Cause it's stable at 16x200, but 16x210 isnt like a lot more

what are your cpu clocks?
i think uncore doesnt clock well if its above cpu clockspeed?
maybe thats why it needs so much vtt to get higher...

did anybody ever check how the efficiency is with uncore clocked the same as core compared to higher or lower? i remember anandtech did that but only briefly and they didnt check it at high overclocked speeds with fast mem speeds...