OP
- Joined
- Mar 31, 2010
- Thread Starter
- #41
Well, finished overclocking finally. For anyone interested:
-bclock got to 200 @ 1.175v in 10mhz increments and didn't bother playing around with it any more. 1h OCCT stable, hottest core got to 74C on average core temp was around 69C.
-Overclocked RAM a tiny bit, well tightented timings from 9 9 9 24 to 8 9 8 24 @1600 @ 1.54v. 1h memtest stable.
-CPU - started the overclock at 1.25v and that got me to 3.8Ghz (1 core spiked to 63 but on average temps we 57.5C) but after that hit a wall and had to increase voltage to 1.3! to get to 4Ghz. 1h OCCT stable 1 core spiked to 72C, on average was 65C, also 1h Prime95 stable on FFT large datasets. Hottest core spiked to 78C!, average was probably around the 72C mark.
Overall, I'm tossing up between 3.8Ghz and 4Ghz, not sure whether those 4Ghz temps and voltages are worth it for 200mhz. At the moment I'm probably 50 50 on reducing the multiplier a tad. Don't think 0.2Ghz makes much of a difference in my world.
The other thing I should probably do is to reseat the CPU. One core is roughly 5C higher in spikes to other cores. I tried thing when I was building my comp probably 3 times and each time a random core was always hotter than the rest. But overall, the difference was smallest so I just left it.
Anyway, that's that!
EDIT: Decided to go down to 3.8Ghz. Reduced the voltage to 1.25 and reduced multiplier from 20 down to 19. BSOD all the way from 1.25 to 1.3v. Admittedly this was not done with downrated QPI and RAM. So maybe that was the problem, but basically it ain't working at 19 multi. I haven't tried running OCCT or Prime95 with x20multi and mem multi of 8 and QPI whatever it needs to be to get to 3200. Are you supposed to run stress tests with all multis set to their final settings?
EDIT2: Well things have gotten a bit more interesting. When I said that it BSOD all the way to 1.3V that wasn't entirely correct, it BSOD up to some point and then it started restarting without BSOD. Now I have changed my settings back to the 4Ghz overclock and all of a sudden, was just sitting around in my web browser and all of a sudden it just restarted... no BSOD. I have a feeling I might have fried something in the PSU when I changed to multi x19 and went down to 1.25 straight away with memory and QPI at max settings. Then again, it wasn't under load! The plot thickens.
EDIT3: Just got another crash not under load. Went to revert back to stock settings and then found that my channel A was set to 8 9 8 24, but B and C for some reason were 9 9 9 24. When I was setting memory timings, B and C were greyed out, so I assumed that they changed accordingly... not sure what where why things went wrong, but basically they are all set to 8 9 8 24 now, so fingers crossed that was it.
-bclock got to 200 @ 1.175v in 10mhz increments and didn't bother playing around with it any more. 1h OCCT stable, hottest core got to 74C on average core temp was around 69C.
-Overclocked RAM a tiny bit, well tightented timings from 9 9 9 24 to 8 9 8 24 @1600 @ 1.54v. 1h memtest stable.
-CPU - started the overclock at 1.25v and that got me to 3.8Ghz (1 core spiked to 63 but on average temps we 57.5C) but after that hit a wall and had to increase voltage to 1.3! to get to 4Ghz. 1h OCCT stable 1 core spiked to 72C, on average was 65C, also 1h Prime95 stable on FFT large datasets. Hottest core spiked to 78C!, average was probably around the 72C mark.
Overall, I'm tossing up between 3.8Ghz and 4Ghz, not sure whether those 4Ghz temps and voltages are worth it for 200mhz. At the moment I'm probably 50 50 on reducing the multiplier a tad. Don't think 0.2Ghz makes much of a difference in my world.
The other thing I should probably do is to reseat the CPU. One core is roughly 5C higher in spikes to other cores. I tried thing when I was building my comp probably 3 times and each time a random core was always hotter than the rest. But overall, the difference was smallest so I just left it.
Anyway, that's that!
EDIT: Decided to go down to 3.8Ghz. Reduced the voltage to 1.25 and reduced multiplier from 20 down to 19. BSOD all the way from 1.25 to 1.3v. Admittedly this was not done with downrated QPI and RAM. So maybe that was the problem, but basically it ain't working at 19 multi. I haven't tried running OCCT or Prime95 with x20multi and mem multi of 8 and QPI whatever it needs to be to get to 3200. Are you supposed to run stress tests with all multis set to their final settings?
EDIT2: Well things have gotten a bit more interesting. When I said that it BSOD all the way to 1.3V that wasn't entirely correct, it BSOD up to some point and then it started restarting without BSOD. Now I have changed my settings back to the 4Ghz overclock and all of a sudden, was just sitting around in my web browser and all of a sudden it just restarted... no BSOD. I have a feeling I might have fried something in the PSU when I changed to multi x19 and went down to 1.25 straight away with memory and QPI at max settings. Then again, it wasn't under load! The plot thickens.
EDIT3: Just got another crash not under load. Went to revert back to stock settings and then found that my channel A was set to 8 9 8 24, but B and C for some reason were 9 9 9 24. When I was setting memory timings, B and C were greyed out, so I assumed that they changed accordingly... not sure what where why things went wrong, but basically they are all set to 8 9 8 24 now, so fingers crossed that was it.
Last edited: