i7 980 - passes stress tests but crashes under normal usage

[dkm777]

I'm still tuning my overclock in my sig. I got it to pass Prime, IBT and LinX for many hours, but it crashes/BSODs sporadically under normal usage and gaming. I'm at work right now and can't tell what the error code is, but I google'd it and all info points to "hardware failure". Of course, resetting everything to stock fixes the problem. I'm running 1.288Vcore, 1.268V QPI/VTT and everything else (IOH, PLL...) at minimum manual voltages (it says that minimum = stock in the BIOS help panel). I have good cooling, maybe I should raise something, I just don't know what. Also, is there a surefire method of chasing problems like these if recommended stress tests do not produce errors?

 

[jmarti445]

Check your video card, your drivers, and your memory. I don't think it is the cpu, as 4.0Ghz is within the realm of reason for that CPU.

 

[dkm777]

You might be right about the memory. The mem in my sig was RMA'd and I'm picking the replacement today. Currently I have 3 sticks of XMS3 taken from two dual kits but they do pass a few hours of memtest86. I'll be testing the new memory overnight and then see if everything runs right over the coming days.

 

[jmarti445]

Corsair memory I've found = bad for overclocking, I've never been able to successfully overclock XMS memory more then a few Mhz over stock

 

[dkm777]

Corsair memory I've found = bad for overclocking, I've never been able to successfully overclock XMS memory more then a few Mhz over stock

Really? My XMS'es are rated for 1600MHz CL8 and i run them at 1604MHz CL9 but the fact that they are not properly matched probably ruins everything. Anyway, the replacement will be a Kingston HyperX 1600 CL9 kit, since the guys i bought my Vengeances from no longer carry Corsair triple channel kits. Hope it runs rock solid for many years.

 

[jmarti445]

Whats the voltage on the memory, I7's are sensitive to DDR3 voltage.

 

[dkm777]

When it was overclocked it was 1.65V as per XMP profile. Right now it runs at 1.5V and 1333MHz. Or do you mean the Kingston HyperX?

 

[aigomorla]

do you have any power features on?

like C1E?

I noticed when you overclock, and you have power options on, when the cpu is trying to go on a lower power settings, then sometimes at overclock i would run into the same problems as your experiencing.

 

[kbp]

By chance.... A Vertex 3 or an M4??? lol

 

[dkm777]

do you have any power features on?

like C1E?

I noticed when you overclock, and you have power options on, when the cpu is trying to go on a lower power settings, then sometimes at overclock i would run into the same problems as your experiencing.

Yes, I have all the power saving features enabled.

By chance.... A Vertex 3 or an M4??? lol

Nope, I wish though.

Anyway, I have installed my Kingston RAM, but set the voltage to 1.5V and leaving memtest for 24 hours. If it produces no errors then I'll proceed to test under Windows. Fingers crossed.

 

[aigomorla]

disable the power saving features and see if thats the root of your problem.

if it is, you may need to lower your bclk and go more to multiplier overclocking.

When your cpu goes into power saving mode, it drops the multi, and leaves your bclk at your overclock as well as drop your voltages.

Well you can tell what trouble that can cause if u have a high bclk.

 

[dkm777]

Still running memtest, but when it finishes I'll try disabling them. But the strangest thing is when I built this rig it ran just fine using same settings until my RAM failed. After that it was nothing but trouble - first I ran into a few pieces of software that wrecked my Windows installation and now I can't overclock anymore. This box is cursed I tell ya. Or maybe I damaged my mobo by constant turning off and on after freezes and BSODs.

 

[jmarti445]

Check your voltages as well, use utility that you can check your rails while putting your system under a load in windows, sometimes it can be representive of a bad rail on your power supply. I've seen power supply's give these type of problems before. It could also be indicative that you may have damaged your power circuitry on board. I one time caused a MSI Nforce 750a platinum board to catch fire due to me overclocking my 1090T to 4.0Ghz. CPU wasn't damaged and still running great but the board I cannot say the same. I doubt it is a problem with the CPU, I've never had a CPU go up via overclocking knock on wood, in the 10+ years I've been doing it. Its usually the memory or the power supply or the motherboard's power circuitry getting stressed due to not being within spec.

 

[dkm777]

Check your voltages as well, use utility that you can check your rails while putting your system under a load in windows, sometimes it can be representive of a bad rail on your power supply. I've seen power supply's give these type of problems before.

Will do. When using HWMonitor I did notice that +5V was kinda low. It would suck royally if it's the PSU going as I bought it in another country and essentially have no warranty.

It could also be indicative that you may have damaged your power circuitry on board. I one time caused a MSI Nforce 750a platinum board to catch fire due to me overclocking my 1090T to 4.0Ghz. CPU wasn't damaged and still running great but the board I cannot say the same. I doubt it is a problem with the CPU, I've never had a CPU go up via overclocking knock on wood, in the 10+ years I've been doing it. Its usually the memory or the power supply or the motherboard's power circuitry getting stressed due to not being within spec.

Also another possibility. The VRM heatsinks are searing under load. But it's the Sabertooth, it's totally indestructible :biggrin:! (serves me right for buying into marketing hype; should've gotten a Gigabyte GA-X58A-UD5 with its 16 phase power)

 

[dkm777]

Got home from work, didn't found any errors during Memtest86+. That's 24 gigs of Kingston HyperX Genesis 1600MHz CL9 running at 1.5V instead of the recommended 1.65. Now testing with LinX. Checked the voltages under HWMonitor and they are:
+3.3V - 3.28V
+5V - 4.87V
+12V - 11.49V
They're all lower than normal. Does that point to a PSU going bad? What I didn't tell you about my PSU is that it has survived (although that is questionable now) a massive surge thanks to a nearby lightning strike. That was some time this summer. It took our security system, my father's HAM equipment and plenty of circuit breakers. The PSU was working fine up until now so I thought - went better than expected. Not so sure anymore.

Update: looks like uncle D is back in business. After replacing RAM I no longer get crashes, freezes or BSODs. What's more - I got not one, but two "golden" kits. I now have 24 gigs of Kingston HyperX Genesis 1600MHz CL9 rated for 1.65V, but running ant 1.5V without problems. In hindsight, I should have gone with this RAM right from the start - it was cheaper than Vengeance and would've saved me quite some time and headaches. Oh well.