3.8Ghz solid but 4ghz crashes

mark1522

mark1522

mark1522

Associate
Joined
25 Aug 2006
Posts
57
Location
Wirral
See sig for rig, running 3.8 all night solid no problems temps high 60's with prime95 blend...ill get it down with lower vcores im sure, thing is when i try to go to 4ghz cpu test is fine all day but prime 95 blend crashes after 30 seconds

tried from 1.25-1.35v vcore cpu ratio is 19x 200 stable 3.8 but 20x200 wont work so tried 21x 191..1.256vcore and this is what setting im using to write this but if i start p95 blend it will crash

memtest ok..i suspect ram settings wrong, qpi is within .5v, ram bus is 1.65v

i can o/c cpu but am rubbish with ram..oh and its patriot viper 1600mhz running 8 8 8 24 atm
 
my 4ghz clock crashes if u use any vcore value under 1.34,try upping yours to around that just to see?and i read that the 21x multi is the most stable to use,so try going with that, im no oc guru its just a suggestion :)
 
I had exactly this situation, the thread bif linked is mine asking the same questions :P

Following Bif's advice its now running at 4.0ghz and did 16hrs Prime stable.

So set everything to auto (inc ram timings) and just change CPU volts, QPI/DRAM voltage and VDIMM

also as per Bifs settings turn off all C functions (C1E, Speedstep etc) and the two "spectrums"

Mine finally stabalised at

21 x 191
vcore : 1.325
QPI/DRAM : 1.33750
VDIMM: 1.65

Choose the ram timing ratio closest to your normal ram speed, and thats all i needed to change.

Your voltages may well be different, but just work on these ones, and leave everything else auto or disabled
 
well prime has run 5 minutes but temps 75-81 on 4 cores..will run 10 mins an then lower vcore 1 click then retry....thanks for your help bif and freak...i know theres loads of info ot there, been looking solidly for days, everything different etc, i just suspected ram cos cpu tests were coming back fine even at 4.1ghz but blend was crashing. wil keep going cheers guys
 
My brief experience with clocking LGA1366 CPUs suggests that they are extremely sensitive to QPI voltage. As I ramped up the clockspeed on my i7-930, it always needed more QPI voltage for stability. Vcore was not really too important, but a very small change in QPI would be the difference between solid stability, and a BSOD within 10 seconds of full load.

You can check here for more details of what I did. But my feeling is that upping the QPI voltage will solve your problems.
 
Last edited:
Duff-Man said:
My brief experience with clocking LGA1366 CPUs suggests that they are extremely sensitive to QPI voltage. As I ramped up the clockspeed on my i7-930, it always needed more AQI voltage for stability. Vcore was not really too important, but a very small change in QPI would be the difference between solid stability, and a BSOD within 10 seconds of full load.

You can check here for more details of what I did. But my feeling is that upping the QPI voltage will solve your problems.
Click to expand...

Yep, thats what i found too, my last tweak of QPI from 1.33125 to 1.33750 went from unstable to solid :)
 
this is running stable at the moment

Bios ver 0303
vcore 1.34375v....bsods 1 click lower
QPI voltage 1.33750v thanks to freakbro and bifday2k for info and others
Dram voltage 1.65v

pc4ghztemp.png
 
See if its still stable if you try lowering the Qpi, leave the vcore as it seems it needs that much.

But u may be able to get away with lower Qpi, worth testing i think, if its not stable, just put it back to the settings that are.
 
You must log in or register to reply here.
Back
Top Bottom