Problems overcloking I7 2600K - System freezes

R

Rephab

n00b
Joined
Mar 21, 2011
Messages
20
Hi guys!
First of all a big "hello!" to everyone cause this is my first post on this great forum and i hope you can forgive me for my bad english. :)

I read every post about overcloking Sandy Bridge, but i've some problem with my spec when i play 3D games, although everything is stable under Prime95, Intel burn test and AIDA64.
This is my configuration.

1204 bios, b2 EVO

Ai Overclock Tuner: Manual
BCLK/PEG Freq: 100.0
Turbo Ratio: "47"
Internal PLL Overvoltage: auto
Memory Freq: 1600mhz

Dram Timing Control: ....
Load Line Calibration: High
VRM Frequency: manual - 350
Phase Control: Optimized
Duty Control: defaults
CPU Voltage: Offset Mode
Offset Sign: +
Offset Voltage: 0.065
DRAM Voltage: 1,65 (as
CPU Spread Spectrum: Disabled

Advanced Menu:
CPU Configuration:

Hyperthreading: Enabled
Intel Virtualization Tech: Enabled
Enhanced Speedstep Tech: Enabled
Turbo Mode: Enabled
CPU C1E: Enabled
CPU C3 Report: Enabled
CPU C6 Report: Enabled

I've also found stability with:
- Everything Auto (same temps as the first conf)
- LLC Ultra High, Voltage Offset + 0.035 (too high on temps, but BSOD with 0.030 settings)

AIDA64 reports a voltage peaks of 1,400 with the second and third confs and a peak of 1,382 with the first conf.
Maybe could i work on voltages?

Temps peaks with Prime95 are 80° for 2 cores... maybe too high! But i use one Scythe Mugen 2 with the stock fan. I really need to change the fan!

As i said before, i've some problem while playing 3D games. The strange thing is that these games don't have a massive load on CPU: 20% on world of Warcraft, near 40% on Crysis and so on.
After some random time, the screen goes black as it is in standby mode and the system freeze (whitout BSOD or any error message).
Do you have any suggestion about that?
Thanks for help!
 
Hi Rephab,

My first question would be, do your games do the same thing without an OC? The issue may not be related to the OC, possibly your psu wattage? Those 6970's sure pull a lot of watts, alternatively, do you have the newest ati drivers, and have you tried it with just one 6970?

Have you tested your system memory?

Sadly, thats about the extent of the help I can probably suggest, I'm not familiar with your board, or the particular quirks of it to be specific.
 
I believe it's a memory issue. When running 3D apps the GPUs can use a lot of system RAM. You don't specify your memory setup. I suggest you enter your memory timings manually. If that's not it try bumping the VccIO a little (max 1.2?) and maybe even the VccSA (max 1.1?). The high density DIMMS you have may need these little bumps to handle the heavy work 3D puts on you RAM.
 
Hi guys and ty!
I've got no problem with default settings, and i use a Coolermaster Silent Pro Gold 1200 Watt as a PSU, so i think is no a "power" related problem! :)

I specified my memory setup manually (ops i've missed something in my spec... sry)
1600 Mhz @1,65V 7-8-7-24, as suggested by vendor G.SKILL

I didn't tested RAM with tools other than Intel Burn Test which uses all free memory.
World of Warcraft, Crysis, ecc, didn't usually use more than 4GB.

Which kind of test do you suggest?
Also another question: do you think temps are too high for a Mugen 2?
MB temps never go over 33° at full load, so i suppose to have a good air flow inside my tower.

Thanks a lot to everyone!
 
Have you tried the ram at 1333? When you say it ran fine at stock, is that with ram defaulted to auto as well? I haven't read on issues directly attached to yours with SB setups, but I have read some issue's with 1600 ram settings the specifics on it I'm not sure of. At work currently or I'd search out what I remembered reading.
 
I would disable C3, C3, C1E, Virtualization tech, speedstep, keep LLC on max setting. disable PLL overvoltage unless you are overclocking 4.9ghz + . Should be 100% stable with all that .
 
sok0 said:
I would disable C3, C3, C1E, Virtualization tech, speedstep, keep LLC on max setting. disable PLL overvoltage unless you are overclocking 4.9ghz + . Should be 100% stable with all that .
Click to expand...

Disabling all the SB benefits and keeping the LLC on max is a terrible idea IMO.
 
hmz said:
Disabling all the SB benefits and keeping the LLC on max is a terrible idea IMO.
Click to expand...

+1

From asus OC SB guide:
"Additionally it is recommended to keep 「C1E」and「EIST」option enabled for the best overclock scaling. This is different than previous Intel overclocking expectations where the best scaling was with disabled power states or power management options."
 
If he is overclocking, why does he need all of the power saving features that I told him to disable? Try it and see. Worked for 2 people that I know. Worst that can happen, your system is still unusable?
 
I did a BIOS upgrade to 1502 and i can see some improvement unde stress test.
@4.7 now i only need a +0.030V with LLC set to high!
I ran Intel Burn test at Extreme setting with 32 Thread and 6500MB of Ram without any kind of problem. After that i logged World of Warcraft and after some minute playing... freeze! Damn!

I've the same behaviour leaving clock at 47 and changing these settings individually (change, boot, run stress test, play WoW):
- RAM = Auto
- everything auto except VRM Freq to 350
- LLC to UltraHigh
- EPU ON/OFF
- Phase control: optimized/extreme
- Duty control: T.probe/extreme

And i've no freeze with ram timig @1600 7-8-7-20 and everything else at stock settings.

Now i'll do some try at 4.4.... i'm so sad...
 
Rephab said:
Hi guys and ty!
I've got no problem with default settings, and i use a Coolermaster Silent Pro Gold 1200 Watt as a PSU, so i think is no a "power" related problem! :)

I specified my memory setup manually (ops i've missed something in my spec... sry)
1600 Mhz @1,65V 7-8-7-24, as suggested by vendor G.SKILL

I didn't tested RAM with tools other than Intel Burn Test which uses all free memory.
World of Warcraft, Crysis, ecc, didn't usually use more than 4GB.

Which kind of test do you suggest?
Also another question: do you think temps are too high for a Mugen 2?
MB temps never go over 33° at full load, so i suppose to have a good air flow inside my tower.

Thanks a lot to everyone!
Click to expand...

I have the same RAM as you (I think) and I'll tell you this.... that RAM does not play nice with sandy bridge. I've had more freeze ups (usually over night or on long daily idles).

I RMA'd it to the Egg and and picked up the 1333 7-7-7-21-1T instead. I could have just lowered the speed on what I have, but it's 50 bucks less for the 1333. So if I'm going to run it slower, might as well save the money.
 
No good news...

I did a try also with these settings:
- Vcore manual @1.420
- LLC regular
- Phase & Duty Extreme
- Multi @46
Still freezes!

But i noticed something: when the game freezes and the screen becomes black (and with the monitor led blinking yellow as it was in standby) i'm still able to talk in teamspeak for 5/10 seconds.
So isn't a "sudden" freeze but... i don't know what...

I'm waiting for some indication while run at default settings... damn...
Thanks guys!
 
@Adam
But i've no problem running @1600 with stock CPU settings, and no problems under Intel Burn Test that uses every single byte of ram if launched at Extreme setting.

I see that you've also 2x 6950 in CF. I was thinking about some strange problem with a Xfire conf, but it's only a "last idea". :-D
 
I agree with ADAM1203's statement about the ram you are using... The build I put together for my brother is almost identical to yours and we tried all sorts of settings and it wasn't until we picked up a set of 2x4 GB 1333 memory and ran it at 1.5 volts were we able to keep it 100% stable with the 2600k OC'ed to 4.7GHz.

Plus he is only running one 6970 vs the two you are running... but that shouldn't matter.

Do you have all of the memory slots populated? or just 2?
 
blert said:
I agree with ADAM1203's statement about the ram you are using... The build I put together for my brother is almost identical to yours and we tried all sorts of settings and it wasn't until we picked up a set of 2x4 GB 1333 memory and ran it at 1.5 volts were we able to keep it 100% stable with the 2600k OC'ed to 4.7GHz.

Plus he is only running one 6970 vs the two you are running... but that shouldn't matter.

Do you have all of the memory slots populated? or just 2?
Click to expand...

Every slot's populated.

Maybe you'll rght, but the question is: why everything's fine with Prime, Intel Burn Test and Aida stress test?
I've this problem only while playing 3d games... that's frustrating!
 
Rephab said:
Every slot's populated.

Maybe you'll rght, but the question is: why everything's fine with Prime, Intel Burn Test and Aida stress test?
I've this problem only while playing 3d games... that's frustrating!
Click to expand...

The reason is because of the way the memory is being accessed.... or in my case, not accessed.

I have heard of specific WOW issues, but the freezing you are having is resultant of your RAM.

I know GSkill is touting this as "P67 ready" but with the CAS timings and clock speed of the memory... it's really not.

I have a M4E board and did EVERYTHING under the sun.... the breaking point was when my wife almost through the damn PC out the window.

The native speed for SB is 1066 or 1333, anything more can't be guaranteed for stability.

Besides, if you do the math, the difference is literally 1.5 nano seconds.....

Other than having an e-boner to say you have the most pimp GSkill Rips X's there's no point.... especially when you have to explain to your friends why your 3000 dollar rig keeps freezing.
 
Rephab said:
Hi guys!
First of all a big "hello!" to everyone cause this is my first post on this great forum and i hope you can forgive me for my bad english. :)

I read every post about overcloking Sandy Bridge, but i've some problem with my spec when i play 3D games, although everything is stable under Prime95, Intel burn test and
Click to expand...

Are you using the Intel RST driver?
 
Rephab said:
No good news...

I did a try also with these settings:
- Vcore manual @1.420
- LLC regular
- Phase & Duty Extreme
- Multi @46
Still freezes!

But i noticed something: when the game freezes and the screen becomes black (and with the monitor led blinking yellow as it was in standby) i'm still able to talk in teamspeak for 5/10 seconds.
So isn't a "sudden" freeze but... i don't know what...

I'm waiting for some indication while run at default settings... damn...
Thanks guys!
Click to expand...

Oh,
Your video cards are crashing

That's a VPU recovery that failed.
Same thing happens with mine if you overclock the cards too far (though I'm using a single card). You can actually still do some windows functions for several seconds after the black screen. Then once the cards COMPLETELY time out, then the system is frozen.

In SOME rare cases, you can get the display back by pressing winkey+R, and type cmd.exe, and then alt enter twice. But this only works if keyboard control hasn't been fully locked out. (in XP, all you have to do is press control+ESC, then type "r" for run, then cmd.exe to bring up that command prompt. Since W7 hid the run menu, the same shortcut might not work).
 
Last edited:
Adam1203 said:
The reason is because of the way the memory is being accessed.... or in my case, not accessed.

I have heard of specific WOW issues, but the freezing you are having is resultant of your RAM.

I know GSkill is touting this as "P67 ready" but with the CAS timings and clock speed of the memory... it's really not.

I have a M4E board and did EVERYTHING under the sun.... the breaking point was when my wife almost through the damn PC out the window.

The native speed for SB is 1066 or 1333, anything more can't be guaranteed for stability.

Besides, if you do the math, the difference is literally 1.5 nano seconds.....

Other than having an e-boner to say you have the most pimp GSkill Rips X's there's no point.... especially when you have to explain to your friends why your 3000 dollar rig keeps freezing.
Click to expand...

I did some try with every ram setting in Auto, but nothing changes.
What i could try is setting it manually to 1333Mhz and, if it could help, setting Volts to 1,65 while leaving timings to Auto.
Do you think that i can replicate the problem with Memtest?
I'm at work atm, i'll do some try later.
 
Falkentyne said:
Oh,
Your video cards are crashing

That's a VPU recovery that failed.
Same thing happens with mine if you overclock the cards too far (though I'm using a single card). You can actually still do some windows functions for several seconds after the black screen. Then once the cards COMPLETELY time out, then the system is frozen.

In SOME rare cases, you can get the display back by pressing winkey+R, and type cmd.exe, and then alt enter twice. But this only works if keyboard control hasn't been fully locked out. (in XP, all you have to do is press control+ESC, then type "r" for run, then cmd.exe to bring up that command prompt. Since W7 hid the run menu, the same shortcut might not work).
Click to expand...

I also thought something about VGAs, but i didn't OC them, neither i changed BCLK freqs.
Are there other settings that can affect VGA's stability?
 
Hmm...
Best way to trouble shoot this:

Try with only ONE card.
After you verify the first PCIE X16 is fully functional, then switch to the second card in that same slot. Once both cards pass, (and hopefully they will), move either one (won't matter, now that you verified the cards are good) to the second X16 slot and test it there.
Will just take a couple of reboots and a driver reinstall, hopefully. Maybe you can have some fun gametime while testing.

This will determine if either one of the cards, or slots ,have a problem.
This may take awhile and it's annoying to test, so report back after you're done.
 
Ok i've found it thanks to Falkentyne! :D

I tried to lower the Radeon power settings to -20 and i had no problem with Unigine while i had the same problem with stock settings.

Every card was connected to the PSU by a single cable with a 8pin + 6pin PCIE connector. I changed this setup using two additional modular cables. Now every card has two cables: one with a 6 pin connector and one with the 8 pin one.
I've no problem with this setup!

But now i've a question: is it possible that a 1200W 85+ gold that come with 4 cables 8+6 pin cannot afford a decent power with only 2 VGAs?

Another strange thing is that when i run 3D applications, i can ear a strange sound coming from the cards (atm i don't know if from only one or both): an "elettric" sound, like interferences (you know... fzzz fzzzz fz zfzz fzzz :D). Maybe that means "Goodbye! Time to RMA!"? And maybe this malfunction draws a lot of useless power?

Anyway thanks for the great support!
 
Sorry, I don't really know anything about modular psu's or how those high end psu's work. I know some of them split the rails, while some have a huge beefy +12v
 
Rephab said:
I did some try with every ram setting in Auto, but nothing changes.
What i could try is setting it manually to 1333Mhz and, if it could help, setting Volts to 1,65 while leaving timings to Auto.
Do you think that i can replicate the problem with Memtest?
I'm at work atm, i'll do some try later.
Click to expand...

You could probably get away with 1.5v. Honestly, I would just RMA the RAM. It should run fine at 1333. I just got mine in today. I sent back my 2600K and RAM (my processor was a crappy OC'er).

Now I'm running at 5Ghz on 1.43v with my 1333 GSkill RipX. No problems so far.
 
Adam1203 said:
You could probably get away with 1.5v. Honestly, I would just RMA the RAM. It should run fine at 1333. I just got mine in today. I sent back my 2600K and RAM (my processor was a crappy OC'er).

Now I'm running at 5Ghz on 1.43v with my 1333 GSkill RipX. No problems so far.
Click to expand...

Thank you for your support Adam, but actually i've found no problems with my ram.
Also an intensive Memtest results in no errors.

Now i need to undestand more about my 6970s power consumption and my PSU power output on 12V lines.
 
Rephab said:
Thank you for your support Adam, but actually i've found no problems with my ram.
Also an intensive Memtest results in no errors.

Now i need to undestand more about my 6970s power consumption and my PSU power output on 12V lines.
Click to expand...

No problem man, I'm here to help! That's what's so great about these forums.... you can get a TON of information from people who just want to lend a helping hand.

With respect to your setup, can you list the specifics (use my signature as a template) so we know what you have in your system?

The 6970s will pull 21 amps each (I have two 6950s flashed as 6970s). So if you figure that you're on a 12v rail and watts=volts x amps then you're looking at about 252 watts each (max load). You can add about 5% on that if they are OC'd with increased voltage.

That said, if you have two you're looking at about 500W just in video cards. Then add to that your hard drives, fans, processor and so on..... about another 200W or so.

If you're not in the 750W PSU range then you're under powered.

With XFire 6970s you really should have 1000W.... you could go 1200W but that's not really needed unless you add a third card.

BTW, what version of driver are you using? You may want to try the 11.4 preview driver, it can be found here: AMD Catalyst 11.4 Early Preview for Windows 7
 
Last edited:
I've a Coolermaster 1200W Silent Pro Gold (85+ gold cert.) and that's why i think there's something "strange".
But why that happens only with overcloked CPU? I did nothing on VGAs!
 
Just thought I'd add that if you need to verify that your video card is recovering during game-play, check the events viewer, ATI drivers will add an entry to the event viewer whenever it happens. Sometimes, rarely, there will be useful diagnostic information.

I had this issue on a non SB system with Dragon Age II. Simply going to the non-Aero theme before launching the game resolved my issues.
 
Rephab said:
I've a Coolermaster 1200W Silent Pro Gold (85+ gold cert.) and that's why i think there's something "strange".
But why that happens only with overcloked CPU? I did nothing on VGAs!
Click to expand...

That should be plenty of power, the 1200 has 98A on the 12v rail. I have the 1000W with no issues what so ever. That's even with a Koolance WC kit.

What driver version are you on?
 
Ok i've update my signature! :p
Now i'm using. AMD Catalyst 11.2 on Windows 7 64
 
Rephab said:
Ok i've update my signature! :p
Now i'm using. AMD Catalyst 11.2 on Windows 7 64
Click to expand...

I see that you're running the RAM at 7-7-7-24 The stock speed is 7-8-7-24

First change that timing.

Second, try the 11.4 driver I posted above.
 
You're right, but the 7-7-7 is only a misspell. I mean 7-8-7.
I tried to download the new driver from the link above, but i got an "Unauthorized Download", i'll try to google it and also i'll search for some kind of changelog.
Thanks.
 
you cant link to AMD drivers, they wont let you.

just goto game.amd.com, and get the preview 11.4 drivers listed
 
Op here is mine. Please use your own multiplier and voltage. You'll see that's what you need to more or less to be going for 5Ghz

AI Tweaker" Page"
AI Overclock Tuner = Manual

BCLK/PEG Freq = 100
Memory Frequency = 1600mhz or whatever freq your Mem is.

OC Tuner = Disabled

EPU Power Saving = Disabled

EPU Settings = Disabled

Dram Timing Control. Set to memorys SPD timings.

Internal PLL overvoltage = Enabled You need the latest bios for this option

CPU Power Management

CPU Ratio - Manual

Turbo Ratio = By All Cores

By all cores = 50

Turbo Mode = Enabled

Load Line Calibration = Ultra High

VRM Spread Spectrum = Disabled

VRM Frequency = Manual

VRM Fixed Freq Mode = 370

Phase Control = Extreme

Duty Control = T.Probe

CPU Current Capability = 120%

Dram Voltage = Set to Memory spec.

CPU Voltage = Manual - 1.375v to start for 5000 Ghz. May require more or less. Be careful with the voltage

All other voltages = Auto

Advanced -> CPU Configuration Page

CPU Ratio = 50

Intel Adaptive Thermal Monitor = Enabled

Hyper-Threading = Enabled

Active Processor Cores = All

Limit CPUID Maximum = Disabled

Executebit = Disabled

Speedstep = Disabled

Turbo = Enabled

C1E = Enabled

C3 = Disabled

C6 = Disabled
 
Adam1203 said:
I see that you're running the RAM at 7-7-7-24 The stock speed is 7-8-7-24

First change that timing.

Second, try the 11.4 driver I posted above.
Click to expand...

Mmmm looking at my RAM XMP profile it says 7-8-7-20, not 24.
Do you think i can safely use 20 instead of 24?
Thanks
 
I'm an idiot!
This is my correct RAM modell 1600Mhz CORSAIR XMS3 Classic DC KIT CL7 rt [CMX4GX3M2A1600C7]

I bought G.SKILL for a friend of mine... sorry...
 
You must log in or register to reply here.
Back
Top