Prime passed,Linx passed,surf web=FAILED

[E6700]

what is the problem with this sandy bitch cpu? at 4.6 with 1.375v it passed every test i run,8 hours prime,linx,memtesst without a problem. this cpu runs very cool idle 36-38c,prime max temp 75c. however leave the pc on for long period of time or while playing java internet games i get BSOD. i disable everything except for HT,mem set at 1600mhz 1.5v. I'm running out of ideas so i just up vcore to 1.38 for now, perhaps someone with this Asus P8P67 board can give me some advice ? thanks

 

[sangyup81]

which setting are you using on prime95? small is good for 8 hours but blend should be run for 24 typically

 

[Kenmitch]

I'm using a 2500K and the Asus P8P67 Pro MB. I'm sitting at 4.6ghz at 1.37v's 100% stable as I wanted to keep S3 sleep.

I was sitting at 4.7ghz stable I thought....Turned out just like yours as it would pass every stress test then out of the blue bsod surfing the web or doing something with very little cpu demand. I figure it was some voltage or setting but didn't feel like trying to figure it out for the xtra 100mhz of speed. I'd rather sit at 4.6ghz with all energy stuff enabled than hard locking my overclock and voltages.

I'm still using the 1502 bios my board came with. Debating if I wanna try another version to see if it helps or hurts my overclock.

When I started playing with my 2500K I found the following guide. Didn't follow it to a tee but it did have some pointers that set me in the right direction. Maybe give it a look over to see if it'll help you. The thread is kinda getting long but might have some usefull info for you or others.

http://hardforum.com/showthread.php?t=1578110

 

[john3850]

This is the 2rd post in regard to flash and crash when using 5850 or 5870
cards.
I have run java net speed tests since 2000 with no problems.
My 5870s cards and 10.1-10.2. drivers gave me many java display problems

Try using the latest ati drivers with the most recent flash update.

Many ati cards use 3 different speeds which may also cause a problem in java.
a) low 2d speed web
b) medium flash speed web with flash
c) high 3D speed games

 

[BonzaiDuck]

what is the problem with this sandy bitch cpu? at 4.6 with 1.375v it passed every test i run,8 hours prime,linx,memtesst without a problem. this cpu runs very cool idle 36-38c,prime max temp 75c. however leave the pc on for long period of time or while playing java internet games i get BSOD. i disable everything except for HT,mem set at 1600mhz 1.5v. I'm running out of ideas so i just up vcore to 1.38 for now, perhaps someone with this Asus P8P67 board can give me some advice ? thanks

I'm otherwise with John3850 -- at least to wonder if this is something other than your OC. He seems to have experience with that particular ATI graphics card. My brother came to me a few weeks back with his PC -- the OS had corrupted because he had just installed a bad or wrong graphics driver.

But even if stable -- How much have you tried trimming your voltage down with the VCORE Offset feature in UEFI-BIOS? I've seen screens of the P67 BIOS, and it's not that much different than my ASUS Z68.

The voltage information about these K CPU's seems sketchy. If it's a 32nm processor, then the "VID" upper limit posted by Intel only states that it's the highest VID they'd ever produce -- not the highest "safe" VCORE. The 45nm CPUs have an upper limit for the "safe" range of about 1.37V. You would make an educated guess -- a deduction -- that the 32nm cores should have a lower safe limit than earlier gen cores. Som OC'ers posting "Sandy Bridge" OC'ing guides are suggesting to keep the VCORE as low as 1.30V.

With these verbose comments, I'm attempting to give as well as take additional information. Nobody is quite sure about these Sandy Bridgers.

I'm currently at 4.53 GHZ with an IBT "Maximum" stress-level at load showing about 121+ GFlops and prevailing VCORE just barely at 1.29V with a maximum reported at 1.30/1.31V. At room-ambient 78F, my hottest core maximum temperature is 69C and the maximum package temperature is 72C.

And I'm not sure I've yet found a minimum stable voltage, but I know I'm getting close . . .

The motherboard "Auto" feature always adds more than necessary voltage, but with this new spate of boards, you can adjust the Offset and still run VCORE at "Auto."

 

[Accord99]

Are you using load-line calibration? While overclocking a 2600K/P8P67 Dlx, I found that with LLC set to one of the higher settings, while the load vcore when running an application that could load all the cores was at a high enough level to be stable (1.368v), when it ran an application that loaded only 1 or 2 cores, the vcore was lower (<1.3v) and at times too low to be stable and would blue-screen. At times the idle voltage would also drop to unstable levels (<0.9v).

I ended up using LLC at the default setting and tweaking the voltages so that idle voltage would remain higher 0.9v and the voltage when running 1 or 2 cores was around 1.34v.

 

[Kenmitch]

This is the 2rd post in regard to flash and crash when using 5850 or 5870
cards.
I have run java net speed tests since 2000 with no problems.
My 5870s cards and 10.1-10.2. drivers gave me many java display problems

Try using the latest ati drivers with the most recent flash update.

Many ati cards use 3 different speeds which may also cause a problem in java.
a) low 2d speed web
b) medium flash speed web with flash
c) high 3D speed games

Trying new drivers may or may not help the OP. Guess it's worth a shot tho just to see.

When on an overclocking adventure more than one thing can cause the same issue.

I'm guessing his problem started after he started cranking up the overclock on his cpu tho....Guess a response from the original poster with a little more of his troubleshooting would help to clerify things.

Without knowing how he went about overclocking the 2600k it's hard to guess. Reading the original post it looks like he's gone the hard locked overclocking route.

 

[E6700]

which setting are you using on prime95? small is good for 8 hours but blend should be run for 24 typically

blend,gonna try run it 24 hours today at 1.38v see if it pass.

This is the 2rd post in regard to flash and crash when using 5850 or 5870
cards.
I have run java net speed tests since 2000 with no problems.
My 5870s cards and 10.1-10.2. drivers gave me many java display problems

Try using the latest ati drivers with the most recent flash update.

Many ati cards use 3 different speeds which may also cause a problem in java.
a) low 2d speed web
b) medium flash speed web with flash
c) high 3D speed games

yeah i have latest Ati driver.

Are you using load-line calibration? While overclocking a 2600K/P8P67 Dlx, I found that with LLC set to one of the higher settings, while the load vcore when running an application that could load all the cores was at a high enough level to be stable (1.368v), when it ran an application that loaded only 1 or 2 cores, the vcore was lower (<1.3v) and at times too low to be stable and would blue-screen. At times the idle voltage would also drop to unstable levels (<0.9v).

I ended up using LLC at the default setting and tweaking the voltages so that idle voltage would remain higher 0.9v and the voltage when running 1 or 2 cores was around 1.34v.

enable LLC i get more BSOD so i have it disable.
I didn't mess with the offset,just manually set the voltage.
hopefully i dont get anymore bsod at 1.38 or iam gonna have to up more vcore until this cpu toast.

 

[Kenmitch]

enable LLC i get more BSOD so i have it disable. I didn't mess with the offset,just manually set the voltage. hopefully i dont get anymore bsod at 1.38 or iam gonna have to up more vcore until this cpu toast

SB's overclocking style is arse backwards from the past. No need to hard lock the voltages and disable all the energy stuff in bios. Using the offset voltage and a little LLC may give you better results. Gonna be hit and miss tho with the offset but once you figure it out it does have it's benefits. If it seems to work fine at 1.38v's after you test it out just save the profile and play around with offset voltage.

Your overclock is hard locked as in full speed and voltage at idle?

 

[E6700]

SB's overclocking style is arse backwards from the past. No need to hard lock the voltages and disable all the energy stuff in bios. Using the offset voltage and a little LLC may give you better results. Gonna be hit and miss tho with the offset but once you figure it out it does have it's benefits. If it seems to work fine at 1.38v's after you test it out just save the profile and play around with offset voltage.

Your overclock is hard locked as in full speed and voltage at idle?

yeah i prefer like that,this cpu runs very cool even when i up to 4.8ghz with 1.42v temps stays 38c idle, but i dont wanna run 24/7 with 1.42 so i back down to 4.6, dont know why every stress test it has zero problem but keep on getting BSOD when doing nothing.

 

[Kenmitch]

dont know why every stress test it has zero problem but keep on getting BSOD when doing nothing.

I thinks it's a SB thing. My guess is it's some kinda voltage that needs tweaked once you hit a certain point with the overclock. Maybe the first sign of throwing in the towel from your chip. To me looks like it lands somewhere around the second larger than normal vcore boost to get to the next multiplier.

Did you try bumping up the memory voltage a little bit? Sometimes it'll help

I went the other route and left all energy savings enabled in bios. Just using the offset voltage and a little LLC works great for me up to 4.6ghz at 4.7ghz I have the same problem as you. I'd rather keep my S3 sleep and idle at 1.6ghz 1.03v while I'm typing responses to posts or surfing the web

 

[BTA]

I may have missed it but have you tried a higher voltage on your memory? 1600 at 1.5v is usually pretty hard for most sticks to actually pull off.

 

[E6700]

I may have missed it but have you tried a higher voltage on your memory? 1600 at 1.5v is usually pretty hard for most sticks to actually pull off.

gonna try that next if i get another bsod with 1.38vcore.Running large fft prime right now.Thanks guys

 

[BTRY B 529th FA BN]

you are not stable at 4.6 and those volts, lower your CPU clock, raise your memory voltage + loosen the timings

 

[E6700]

you are not stable at 4.6 and those volts, lower your CPU clock, raise your memory voltage + loosen the timings

i enjoy overclocking,so i'm gonna find a way to make this cpu stable at 4.6. i upgrade from i5-750 @4ghz just to overclock

 

[BTRY B 529th FA BN]

Something aint right with the voltages somewhere... i'd invest some time in the RAM, tuning and fine tweaking...

 

[Kenmitch]

i enjoy overclocking,so i'm gonna find a way to make this cpu stable at 4.6. i upgrade from i5-750 @4ghz just to overclock

Glad I wasn't the only one

I didn't need the 2500k but couldn't fight off the overclocking demon in me. Good thing no compelling video cards have came out lately as I'd be in the same boat on that end of the deal. The nice thing is I can at least keep my case, power supply, and water cooling gear for the next adventure. GPU waterblock will go with the 5850 someday.

 

[Suspicious-Teach8788]

BSOD code please? Everytime you get a BSOD, log down the code.

 

[E6700]

BSOD code please? Everytime you get a BSOD, log down the code.

x124,which is why i'm working on the vcore first.

 

[nyker96]

there's only one conclusion, web surfing is more cpu intense than prime95. ;]

 

[Nemesis 1]

enable Vdrop control to< . turn on all your power saving features and in Cstate set to 6. lower your voltage to 1.37 first than try 1.36. Turn off turbo . Don't worry it still works if you let intel manage your cpu. Buy using the tools in bios . I have mine locked at 1.36 now at 4.5 ghz. But it turbos up to 5.2 at times normal 4.82. . All intel energy features turned on and I am using the IGP with Virtu. This chip is unning @1.352 and V drops to around 1.344V underload . So 1.344V is what my chip needs to be stable at 4.5ghz all 4 cores and allows for great turbo even tho its turned off in the bios.

 

[Idontcare]

blend,gonna try run it 24 hours today at 1.38v see if it pass.

Don't use blend, its the least advisable option in Prime95 for performing stability testing.

Use small FFT for testing CPU core stability. Use large FFT for testing uncore (L3$, IMC, etc) stability.

Blend fails to test either set of components rigorously because it alternates between small and large FFT's which means the CPU never really gets overly hot and stressed and neither does the memory pathways from the dram to the L1$, resulting in an OC'ed rig that appears to "pass Prime95" when it really doesn't.

That said, blue screening when web surfing sounds like a graphics issue most likely being brought on by flash apps in the web-page IMO. I'd be curious to know if your rig throws errors in furmark (if it doesn't BSOD when running it).

 

[Nemesis 1]

I'm using a 2500K and the Asus P8P67 Pro MB. I'm sitting at 4.6ghz at 1.37v's 100% stable as I wanted to keep S3 sleep.

I was sitting at 4.7ghz stable I thought....Turned out just like yours as it would pass every stress test then out of the blue bsod surfing the web or doing something with very little cpu demand. I figure it was some voltage or setting but didn't feel like trying to figure it out for the xtra 100mhz of speed. I'd rather sit at 4.6ghz with all energy stuff enabled than hard locking my overclock and voltages.

I'm still using the 1502 bios my board came with. Debating if I wanna try another version to see if it helps or hurts my overclock.

When I started playing with my 2500K I found the following guide. Didn't follow it to a tee but it did have some pointers that set me in the right direction. Maybe give it a look over to see if it'll help you. The thread is kinda getting long but might have some usefull info for you or others.

http://hardforum.com/showthread.php?t=1578110

When your in S3 check your Vcore voltage with a multi-meter let me know what ya find.