I currently have an Asus P6T, i7 920 c-step, CNPS 9900 HS, 6gb 1600 OCZ platinum DDR3 memory, 2 BFG 285gtx OCX in SLI, and 1000w PSU to juice it. With that said, and who knows how many have before, I'm not entirely new to overclocking or computers, however I'm running into a concrete wall with the overclocking.
I've searched countless forums and tried countless suggestions, guides, possibilities, but am unable to push my cpu past 3.3 with all settings auto. I have updated every component that could be updated and although I'm sure I wasn't by the book I will try to note everything i did notice wrong that raises a few questions in my opinion.
Now, I did not go from the grassroots settings and raise the clock/voltage in 10 or so increments at a time - adjusting voltage as needed ... and I know most have you have labored endless hours to get the results you did, as i should...But when i follow a rough bios settings equivalent of "forum poster/computer article" and tweak as necessary nothing I do helps the near seconds of instability and blue screen PRIME 95 causes after initiating.
So if i was at about 3.8, 1.35v on both vcore/qpi, ram speed toward the default side, all spectrums disabled, turbo off, lowest qpi/uncore necessary i would boot into windows...hell, try a few games without problems(although they're not a stability test by any means) as soon as i ran prime 95 within minutes it would crash - with temps around 70-75(TJunction)....I'd usually up vcore/qpi and try again - the same bsod. Just to see if voltage was even an issue i would put it at about 1.42 to see if it could make it to the second thread of prime tests...nope.
I have tweaked using dozens of other nearly identical setups, upping and decreasing as needed. No luck. Now while i can see how this is lacking any solid numbers, or maybe clarity and i can get any add. info you need, but some of the potential problems i thought might be causing it were when i tried a different stress test...Everests' included stability test and unchecked everything but cpu, 10+ minutes without a problem versus the on the clock crash prime produced after just under a minute.
I dont know if that's indicating or showing a likely weak link, but when i did check cpu and fpu and tried the test again it copied prime's signature death kiss. It's getting fairly frustrating which I'm sure is part of the game, but i was hoping someone might have an idea what could be causing wall. I know all cpu's arent able to hit the norm...But, i figured it was worth a shot. Lastly, i'm not positive on this but i read somewhere that with a 3.8 or so OC, a handful of peripherals and 285 gtx in sli with modest overclocks that it might be pushing 800-900 watts under full load. I know prime doesnt put 100 percent on each component but maybe it could be a 1000w PSU that was "always" reliable can't hack it's expectations?
Try setting ur cpu to this
Bclk : 185
Ram speed : 1483MHz
Multiplier : Auto
Speedstep : On
HT : Doesn't matter
Turbo : Off
CPU Voltage : 1.2750 (Type 1.27 will make it 1.2750)
The rest auto
Tha'ts my setting for my I7 and P6T, remember Turbo must be off to get more than 3.3GHz++
The reason I go for 3.7GHz is the voltage, going to 3.8GHz and above needs a lot of extra voltage(Around 1.35+), and the performance difference isn't that great.
I set the Bclk to 185 to get 3.7GHz and leave the multiplier to Auto cause only Auto multiplier can enable the speedstep feature (For my case)
Thank you, If you took the time to suggest - I'll definitely take the time to try it. One thing i never touched i occasionally hear needs to be adjusted is taking the ram timings off auto...Is that something i should manually configure or irrelevant? Also, i heard something about differential amplitude, any thoughts on that setting? Thanks again, really appreciate you taking the time to help!
UPDATE - I tried what you (xyzionz) recommended, but with no luck I double checked everything you didnt mention was on auto and entered the values as you listed but windows couldnt boot. I tried upping the voltage a little to see if that'd help, maybe a a slim percent room for varying boards/configs - but couldnt boot with the 1.3v i tried. I left vcore on auto to see what would happen and it made it into windows, CPUZ said the vcore was a 1.312, but still failed after a few minutes of small--fft prime 95 testing. Ugh, well - no one said it was easy right...Thanks anyway though
Thank you but your threads was one of the many i did look through to see if there was anything different or likely and it was unsuccessful. I do want to add i looked through some of the notes i was taking initially and for the "goal" i was trying to get to of 3.8 nothing short of 1.41 vcore(auto put it as well) would work to boot...I'm wondering if i have bad cpu or I'm missing something crucial - maybe i could take a few pics of my bios and post them. Also, at about 1.4vcore my tjunction temp was roughly 85-87 with my CNPS 9900 on full blast, does that sound right? or maybe too much paste this time? ANyway, thanks everyone, i envy you all with your 1.3vcore tops necessary to clock to the moon...lol
On a side note....A majority of the time i blue screen the error given is something like "clock interrupt was not received on a second processor," which i've read means up the vcore a little - but, more than 1.42v for 3.8 seems abnormal.
I have almost the same setup, I am running the Corsair DDR3 1600 ram and 2x GTX 275's. I am running on 200mhz FSB @ 3.8. 4.0ghz actually runs slower. The key here may be your CPU, I have seen some c0 stepping processors not make it passed 3.3.ghz. Pretty much at luck of the draw if you have followed the guides. In theory, you should not need more than 1.35 volts to get your CPU stable in the 4ghz+ range.
Im having a difficult time, can use a 172 BCLK X21 with about 1.25v, but when i raise to 173, i get the continuous reboot cycle. Upped the voltage all the way to 1.4 and it still wont take it. my brother has the same identical setup, and does the same identical thing. only where his was once rock solid at 165 BCLK and 1.25v, gives us continuous reboot cycle. This just cant be.