Printable View
Hi guys,
got a few problems here with my i7 setup.
Mobo is Asus P6T Deluxe (Bios 0804), i7 920 and 2GB corsair 1333 Ram.
i7 is cooled by chiller so temp problems cant be :p:
Setup is 19x211 prime stable, but i cant prime the 20x200. Above around 170 BLCK i only got freezes in prime after few seconds at multi 20.
I dunno which Problem that is :(
More Voltag on IO or QPI doestn help. Multi 19 is all fine no Problems :(
Anyone got help for me ? seems im the only one with this prob.:mad:
i donīt think so ;) :p:Quote:
Originally Posted by aNNixP
I thought the 920 was locked?
the 920 IS locked. i think that the chip just cant clock that high (4GHz).
Nope, it has some options. This is why we should ignore Fud. It gets stated as fact, when the things they post are in fact fiction.
CPU is "locked" (20 is default, 21 & 22 are turbo multipliers). The problem is that some i7-920 canīt run 20x200 (4Ghz) stable but 19x211 (also 4Ghz).
Try turning Turbo and EIST off. Also make sure TDP limit is turned off.
Sry Chrisch i forgot you :D
Exactly that problem which Chrisch confirmed, 19x211 all fine 20x200 no go :(
All done already ;)Quote:
Originally Posted by xoqolatl
My guess is the memory controller or a form of NB strap. Hows performance on 9*211 compared to people running 20*10?
aNNixP, how do you run your uncore (NB speed in CPU-Z Memory tab)?
All three CPUs I tested have had problems running core in sync with cache at above 3600MHz. Try lowering your uncore multi or increasing core multi by one step. I tried to test performance of sync vs async and it seems to make no difference. Fast uncore is very good for Pi and Wprime, +133MHz on cache gives you more than +133 on core.
@ xoqolatl
i tried all, lowest Uncore + lowest memory multiplier. But it doesnīt help! I can prime 19x200 over 3 hours, but if i set the multi from 19 to 20 and
raise the vcore it isnīt stable. If i lower se multiplier to 19 and raise the BCLK to 211Mhz is it stable ;)
I can also bench 19x218 but not 20x212 :(
PLL voltage then? My 920 needed 2.0V PLL, 1.5V core, 1.475 vtt to be stable at 4000 core, 3800 uncore.
My 920 need 1.8v PLL, 1.36v VCore and 1.24v VTT(QPI) for stable 4Ghz Core and 3.16Ghz Uncore.
But this is with 19 multiplier and 211Mhz BCLK. Only the 20 multiplier doesnīt work :(
and i donīt think that the CPU need more VCore, PLL & QPI voltage with lower BCLK / Uncore / QPI clock.
for example:
4Ghz: 19x211 / 3.8Ghz QPI (multiplier 18) / 3.16Ghz Uncore (multiplier 15) / 633Mhz Mem (multiplier 3/6) / 1.4125v VCore (Bios) / 1.24v QPI/VTT (Bios) / 1.60v vDimm (Bios) / 1.80v PLL (Bios) <- primestable
4Ghz: 20x200 / 3.6Ghz QPI (multiplier 18) / 3.00Ghz Uncore (multiplier 15) / 600Mhz Mem (multiplier 3/6) / 1.4125v VCore (Bios) / 1.24v QPI/VTT (Bios) / 1.60v vDimm (Bios) / 1.80v PLL (Bios) <- not stable (restart after 15sek)
Probably still needs a lot of bios maturity as well.
higher multi always needs more vcc, its the nature of squeezing more clock phases into the same cycle.Quote:
Originally Posted by Chri$ch
you raise the bclk and lower the multi you are reducing phases and decreasing the clock period by adding more frequency on each multipliers phase.
the opposite happens when you raise multi, decrease base clock. clock period on each phase increases, as does amount of clock phases. more phases requires more voltage and steadier supply. the cpu itself might not be able to throughput valid data with the heavier load of increasing the multi with such high base clock.
its much like we're used to with fsb. you can run a low multiplier, high fsb and need less voltage and encounter less difficulty than with a higher multi and lower fsb producing same out put frequency. all to do with phases and alignment.
you can always whack Vcpu_pll up and see if its enough to lock the phases in alignment with each other.
Vcpu_pll and Vcc would be in my humble opinion the way to start looking for answers.
Anyone else not a fan how Intel took AMDs locking route here. Are we going to see like "Blue editions" come out?
24/7 stable at 21x191 here (20x191 POST with turbo enabled), you need to raise VQPI, it needs to be kept within .0250~.0375V of VCore depending on VCore gates, try VCore at 1.3975V to start with VQPI at 1.375V, PLL works best on Auto right now on the ASUS board. Also, blck over 200 is a problem in the BIOS you have at 20x multi on POST.Quote:
Originally Posted by Chri$ch
Ok ill try it later :) thx. But i dont think that it will help.
I testet already all since the last 3 nights.:(
20x200 is not stable for me, my chip is a good chip, runs 3.2ghz with only 1.08v, so i know 4ghz should not be a problem.
try 19x211 :D
Similar story, 19 x 200 is fine but 20 x 200 is no go (920 + P6T Deluxe). 20 x 195 works at low vcore (1.35 bios) so I feel sure 4 ghz is within reach. I haven't tried 19 x 211 yet but will do that tonight, thanks for the tip guys!
Seems that im not the only one with the Multi Problem :p:
Any Solutions yet ?
I got the problem as well, if anyone has a solution please post, used bios 0903 as well, no help tho, my cpu is stable at 3.8ghz with 1.25v, so im sure 4ghz is attainable.Quote:
Originally Posted by aNNixP
With 0903 bios I can run stable at 4 ghz in Turbo mode (21 x 191), but still not 20 x 200.
Well, I'm gonna resurrect the dead here but I want to know if this is only occuring on Asus boards? It sure as heck sounds like it is, which would be indicative of either a propagating BIOS fault or a motherboard engineering fault...
Any ideas?