• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Solid Freeze - Easy Fix?

N

natr89

n00b
Joined
Jun 12, 2011
Messages
25
I am having an issue that is driving me absolutely insane. Posting here is my last resort and I am sincerely hoping that someone has an answer.

First things first, computer setup:
Antec P280
Corsair 800w PSU
ASUS P6T Deluxe V2
i7 920 OC'd to 4.0Ghz
Corsair H100
12GB Corsair RAM 1600Mhz
2 x HD Radeon 6970 in Crossfire, both OC'd
4 Monitors (all differing models, 1 acer, 2 dells (1 of them ultrasharp) and a vizio)

My overclocking has been a journey and a half trying to find something stable. After going all the way up to 1.55 on both Vcore and QPI and eventually decided to set them to auto with the BCLK set to 200 (of course for the 4.0Ghz OC). With them set to auto, I had no real issue until really taxing the machine. Once under pressure, I get a hard freeze. No sound. No distortion or the screens, nothing. Silence and everything is frozen. No BSOD, no memory dump, NOTHING.

Everything in my research points to a memory issue hence why I am posting here. CPU temp is a constant 45ish degrees Celsius and with the Vcore and QPI set to auto, CPU-Z shows it at 1.4.

What should I do?!?! This is drive me absolutely insane. I am getting ready to run a memtest 86 and will post those results here once I get them.

Anyone have any idea what this is or how I can resolve it? Thanks in advance for any help I get.

Nate
 
Memtest86 shows NOTHING after 2 full passes.

Also absolutely NOTHING in Event Viewer other than the Kernel Power critical for me hard shutting it down to reboot and get it active again.

Anyone?
 
Last edited:
Memtest86 shows NOTHING after 2 full passes.
Click to expand...
What about after 20 passes? You really need to run memtest86+ more than 8 hours to catch subtle memory problems like a single bit flip after a cell heats up. I have seen memtest86+ pass for hours before finding the first error on quite a few occasions at home and at work. An other option is hclmemtest. This seems to find bad ram faster (in only a few hours instead of a day). Run several instances of that simultaneously to cover almost all of your ram.


If all the memory testing passes then try IntelBurnTest maximum burn followed by Prime95.
 
I was having the exact same problem with my P6T. Ran memtest for several hours and would never show errors. I ended up picking up a Sabertooth X58 last week and everything has been stable since.
 
thefordmccord said:
I was having the exact same problem with my P6T. Ran memtest for several hours and would never show errors. I ended up picking up a Sabertooth X58 last week and everything has been stable since.
Click to expand...

I was afraid someone was going to say MOBO. I had issues with my first P6T and had to send it back and they ended up sending me this one that I am using now. UGH. Not cool.....

Any other suggestions?
 
drescherjm said:
What about after 20 passes? You really need to run memtest86+ more than 8 hours to catch subtle memory problems like a single bit flip after a cell heats up. I have seen memtest86+ pass for hours before finding the first error on quite a few occasions at home and at work. An other option is hclmemtest. This seems to find bad ram faster (in only a few hours instead of a day). Run several instances of that simultaneously to cover almost all of your ram.


If all the memory testing passes then try IntelBurnTest maximum burn followed by Prime95.
Click to expand...

I will run this today and see what I come up with. If something shows, I will post it. If not, I will run the prime95 and see if we freeze.

Thanks!
 
Vcore and QPI set to auto is a very bad idea.

Have you tried:
1. setting CPU clock skew to 300ps?
2. Turning off CPU and PCIe "spread spectrum" This is fine to have enabled when running at stock, but it will muddy the signals and can cause issues when overclocking. Leaving this enabled is for tin-foil hat people who are scared of somebody harvesting data from EM produced by the computer.
3. Setting CPU Differential Amplitude to 800mv

Having the QPI and CPU voltage set to the same thing can also cause issues overclocking. If you have a good D0 920, 1.4 should be fine for CPU voltage.. you may even be able to go lower.
QPI voltage is very dependant on how good the IMC on your CPU and what RAM you have.

My old 920 required 1.65 QPI voltage for running DDR3-2000 at 2000 with a BCLK of 200. Intel's updated stance on QPI voltage is that it is safe as long at it is within 0.5v of CPU voltage.

With 1600Mhz RAM, QPI should be fine at 1.35v or lower unless your IMC is horrible. Even my old 920 ran the RAM fine at 1600 with a 1.35v QPI.

You can also try setting the PCI-e speed to 105Mhz.. sometimes it makes a difference.. sometimes not.

You may want to try raising the IOH and maybe the ICH voltage by 1 or 2 notches. IOH seems to like a bit more voltage at higher BCLK settings.

You may also need to adjust the CPU PLL voltage. Some CPUs like them at the lowest setting.. others like them higher. I generally run at 1.84 or 1.86 for the CPU PLL voltage.
 
drescherjm said:
What about after 20 passes? You really need to run memtest86+ more than 8 hours to catch subtle memory problems like a single bit flip after a cell heats up. I have seen memtest86+ pass for hours before finding the first error on quite a few occasions at home and at work. An other option is hclmemtest. This seems to find bad ram faster (in only a few hours instead of a day). Run several instances of that simultaneously to cover almost all of your ram.


If all the memory testing passes then try IntelBurnTest maximum burn followed by Prime95.
Click to expand...

Ran it for about 3 hours. No errors found.
 
cyclone3d said:
Vcore and QPI set to auto is a very bad idea.

Have you tried:
1. setting CPU clock skew to 300ps?
2. Turning off CPU and PCIe "spread spectrum" This is fine to have enabled when running at stock, but it will muddy the signals and can cause issues when overclocking. Leaving this enabled is for tin-foil hat people who are scared of somebody harvesting data from EM produced by the computer.
3. Setting CPU Differential Amplitude to 800mv

Having the QPI and CPU voltage set to the same thing can also cause issues overclocking. If you have a good D0 920, 1.4 should be fine for CPU voltage.. you may even be able to go lower.
QPI voltage is very dependant on how good the IMC on your CPU and what RAM you have.

My old 920 required 1.65 QPI voltage for running DDR3-2000 at 2000 with a BCLK of 200. Intel's updated stance on QPI voltage is that it is safe as long at it is within 0.5v of CPU voltage.

With 1600Mhz RAM, QPI should be fine at 1.35v or lower unless your IMC is horrible. Even my old 920 ran the RAM fine at 1600 with a 1.35v QPI.

You can also try setting the PCI-e speed to 105Mhz.. sometimes it makes a difference.. sometimes not.

You may want to try raising the IOH and maybe the ICH voltage by 1 or 2 notches. IOH seems to like a bit more voltage at higher BCLK settings.

You may also need to adjust the CPU PLL voltage. Some CPUs like them at the lowest setting.. others like them higher. I generally run at 1.84 or 1.86 for the CPU PLL voltage.
Click to expand...

Thank you SO MUCH for this detail. I set everything to the following:

CPU clock skew - 300ps
CPU spread spectrum - off
CPU diff amp - 800mv

CPU voltage 1.4
CPU PLL - 1.84
QPI voltage - 1.35
BLCK - 200
PCIe - 105
DRAM frequency of 1203 (recommended from somewhere else even though ram is 1600 to allow for head room)

IOH and ICH - 1.2 (up from default of 1.1)

Everything booted up properly, but 10 seconds into a Prime95 test prompted a BSOD which gave the following:

"A clock interrupt was not received on a secondary processor within the allocated time interval."

HELP? :)
 
natr89 said:
Thank you SO MUCH for this detail. I set everything to the following:

CPU clock skew - 300ps
CPU spread spectrum - off
CPU diff amp - 800mv

CPU voltage 1.4
CPU PLL - 1.84
QPI voltage - 1.35
BLCK - 200
PCIe - 105
DRAM frequency of 1203 (recommended from somewhere else even though ram is 1600 to allow for head room)

IOH and ICH - 1.2 (up from default of 1.1)

Everything booted up properly, but 10 seconds into a Prime95 test prompted a BSOD which gave the following:

"A clock interrupt was not received on a secondary processor within the allocated time interval."

HELP? :)
Click to expand...

Hrmm... that is usually CPU voltage.. but can also be other things.

What voltage is your RAM supposed to be at? Do you have it manaully set to that voltage?

Try CPU PLL of 1.86.

What speed to you have QPI set to?

Is uncore speed set to 2x what you have the RAM set to? If you leave it on Auto it should default to 2x of what you set the RAM speed to.

You can also try raising the CPU voltage by one or two notches as well as the QPI voltage by a notch or two.

Try one thing at a time and then test.
 
HALLELUJAH PRAISE THE LORD.

You are a saint my friend. I set the RAM to auto and uncore to auto as well as bumping up the Vcore and QPI and the PLL to 1.86 and we are golden. I set the RAM voltage manually to 1.6 (tested at 1.5 says my research).

Everything looks golden. Ran an 8 iteration Prime95 test for over 30 minutes and it was smooth as butter. Temps didnt even go over 82C which is amazing for that many iterations.

Thank you again SOOOOOOOOOOOOO much for your help!
 
NOOOOOOOOOOOOOOOOOOO!!!!!!!!!!!!!!!!!!!!

It just froze again, same exact issue as before. GRRRRRRRRRRRRR.

I just upped the QPI to 1.425 to see if that makes a difference with a 1.45 vcore.

Anything else I can try?!!? This is so damn frustrating!
 
You can try setting the RAM timings manually. My ASUS board always seemed to change timings every time I did a cold boot.
 
natr89 said:
What are the specifics for doing this?
Click to expand...

First you need to get the base timings of what your RAM is supposed to be... might be listed on a sticker on the RAM itself.

You can also use CPU-Z to get a few more timings. Depending on what speed you are running the RAM at, the timings will be different.

I usually set the timings to what the RAM is supposed to run at for the speed I want to get to, not the speed I am necessarilly running it at, and then tweak from there.

If your RAM has an XMP mode, you can use the XMP mode setting in your BIOS to have it manually set the base timings... write the rest of your BIOS settings down first because using the XMP mode will change other settings as well.

Then reboot and look at the sub timings.. what they are set at automatically is usually a good place to start.
 
cpuzz.png


Current CPUZ settings. Does that help?

http://imageshack.us/photo/my-images/842/cpuzz.png/
 
Click on the SPD tab. That will give you default settings for different speeds.
 
I was still having issues galore so I bumped my Vcore all the way up to 1.575 and my QPI up to 1.55.

I have not had any issues and with my H100 I am running at 50C under no load/idle. Normal use is about 55-58C. When playing game it maxes at about 63-65C. Not too bad eh?

How bad is it to run my Vcore at that voltage 24/7?
 
Temps aren't the whole story, voltage alone is enough to kill a chip, though it's true that voltage and high temps will kill it faster. You are running a 920, though, and while I may be wrong I think those were a little more forgiving regarding vcore than current i5/i7s, so you might be okay. I just wouldn't recommend that high for 24/7 use.
 
Still not fixed. I feel like I have tried EVERYTHING. Even the really high voltages froze on me. This is getting ridiculous. I have updated all possible drivers on my machine and I still cannot figure this out....


HELP
 
Your RAM could still be bad regardless of what Memtest shows. Run your computer with one stick at a time and I bet you find a bad one. Memtest is about 50/50 in my experience.
 
Been wanting to add more anyway. I will double up and update when I can run some tests on 24GB.

Thanks.
 
You must log in or register to reply here.
Back
Top