Gigabyte GA-X79-UD3 Core i7 3820 Random Freeze

[Andres Gonzalez Biront]

Hi!

I built this new system a month ago. I thought it was stable but I'm having strange issues, mostly at idle, but I think it's random. I'll just copy what I wrote on XtremeSystems.

I'm now having problems with my OC, and I started blaming the Motherboard.


To get my 3820 stable at 4625Mhz I've used 1.44V Load on LinX. I've also tested Prime95 with AVX, and it's good for at least 2 hours (I didn't test more).


It works perfectly as long as I don't let the PC idle for a long time, because it will freeze. It's randomly crashing, with no BSoD, just a system freeze. Also, I've noticed a very strange behavior, sometimes when it freezes my hole home network crash. I have a 8-port 1gbps switch and a 5-port 1gbps router with DD-WRT, 4 servers connected at 1gps (some of them with 2 NICs), a vCenter Server 5.1, 2 ESXi, 2 VMware Worstations 9, a lot of VMs. EVERYTHING STOPS WORKING while the PC is freezed. I have to power down my 3820 to recover networking operation on my LAN. You simply CAN'T PING TWO COMPUTERS ON THE SAME SWITCH when my i7 3820 crashes!!! It looks like it is storming the network. It's driving me nuts.


I have tested 4625Mhz at 1.475V just to be sure (1.44 is stable). But it failed.
I have lowered from 4625Mhz to 4500Mhz (36x multi, 125Mhz) and 1.44v (just to be sure). But it failed.
I have changed the RAM latency from C9 1T to C10 2T (it's memtest86 stable at C9 1T). But it failed.


It fails randomly, mostly on idle, or when... when I'm playing tetris


Never on heaven, prime, linx...


Any ideas? Suggestions?


The BIOS version is F12T, the latest BETA. The same problem was happening with the F12Q BIOS.

Edit:

Some additional information. Sleep is disabled, so it is not related to that. C1E and all the power-related features are Enabled, but they were disbled, I've enabled those features for testing. The behavior is the same, the system will randomly crash (once a day or two), mostly at idle or with low load.

[Vinster]

Have you looked at the event viewer to see if anything was captured (or errors prior) from the crash. sometimes it will point to an issue and then hard crash prior to a BSOD being able to respond.

and have you tried running the system at stock speeds at all to see if the problem is replicated when no overclock was present?

I recently had an issue where my threads (1-8) would randomly "lose power", this was according to the event viewer system tab. this would cause a similar issue as you are describing... but the issue was only present when I overclocked, didn't happen when the CPU was at stock speeds. I could game, bench... all fine... do nothing or browse the web and boom... hard lock. in my case I had speedstep disabled, 3 years never an issue until recently.

Vin

[Andres Gonzalez Biront]

Have you looked at the event viewer to see if anything was captured (or errors prior) from the crash. sometimes it will point to an issue and then hard crash prior to a BSOD being able to respond.

and have you tried running the system at stock speeds at all to see if the problem is replicated when no overclock was present?

I recently had an issue where my threads (1-8) would randomly "lose power", this was according to the event viewer system tab. this would cause a similar issue as you are describing... but the issue was only present when I overclocked, didn't happen when the CPU was at stock speeds. I could game, bench... all fine... do nothing or browse the web and boom... hard lock. in my case I had speedstep disabled, 3 years never an issue until recently.

Vin

There aren't any logs before the crash. Just some random service starting and stoping. How did you manage to solve it, anyway?

[Vinster]

check all the logs... System, Application, Windows PowerShell, Hardware events... in my case re-enabling speedstep and setting my system state to C6E fixed my issue. but my issue was different than yours and completely different platforms.

Vin

[Andres Gonzalez Biront]

I've solved it. I wanted to let you know, if someone get a similar error, it was the RAM. I was running 4x4GB DDR3-2000 9-10-10-30 1T, it was Memtest86 stable, so I thought it was good. But after lowering the clock to 1666Mhz the problem disappears.

Finally, I went back to 2000Mhz but with 10-11-11-30 1T timings. Now I don't have the lock ups anymore :)

[Vinster]

Nice, glad to hear.

Vin

[ibach]

How the hell have you made such a clock with i7 3820?! I have a lot of trouble going over 4300 with allmost exact hardware. The moment I raise BCLK over 101 I lose network and everything on gsata...
Please post BIOS settings in detail.

[Vinster]

How the hell have you made such a clock with i7 3820?! I have a lot of trouble going over 4300 with allmost exact hardware. The moment I raise BCLK over 101 I lose network and everything on gsata...
Please post BIOS settings in detail.

don't play with memory, start with the multiplier. and an OC is not a guarantee, anything above stock is a bonus.

I recommend you start your own thread, include all the bios settings that you have currently set and the community can go from there and help.

Vin

[Andres Gonzalez Biront]

How the hell have you made such a clock with i7 3820?! I have a lot of trouble going over 4300 with allmost exact hardware. The moment I raise BCLK over 101 I lose network and everything on gsata...
Please post BIOS settings in detail.

Don't raise the BCLK, use the multiplier. First of all set the CPU multiplier to 36X or less, to ensure you don't go beyond 4.5Ghz, then set the vCore accordingly (for 4.5Ghz depending on the quality of your chip it should be between 1.3 and 1.4V). When you change the strap to 1.25x, the BCLK for the CPU will be 125Mhz, so double check the memory, and leave it at 2000Mhz or less (you should try to hit 2333Mhz once the CPU is fully stable).

I'm now LinX stable at 2333Mhz 11-12-12-34 1T, 4500Mhz on the CPU. I just can't get it stable at 4625Mhz (my target was 4750Mhz), maybe I have a bad clocker.

Hope this helps:
This config was stable for two hours with Prime95 AVX, but not for LinX. I can't get LinX stable even with 1.472V, the BSoD I get is 0x09c, it should be the IMC/L3 having problems at that speed but I'm not sure how much voltage is safe. The VTT and VSA are now at 1.2V, maybe 1.25V or 1.3V will fix it, but I don't know if it's safe to run at those voltages.

[Vinster]

ibach, please do yourself a favor and work your way up to these results with the information provided. if you just punch them in and they don't work you wont have any direction as to the cause or resolution to the issue.

Baby steps wins the race here.

Vin