New 8320 Fails IBT at Stock

[Tyerker]

So, I just built my newest system last night. It's my first foray into water, with the Corsair H60. Today, to make sure the cooler was behaving properly, I decided to run the Intel Burn Test at stock. About 70 seconds in, it always says my system was unable to complete the test, it may be unstable, check the cooling system or lower my overclock.

Here are voltages and temps about 5-10 seconds before the test crashes:


After this, I tried turning all of my fans up to 100% (as you can see) and it didn't change anything. Also, to make sure it wasn't a bad fan port on my motherboard, I also swapped the pump's cable to a different fan port, and a fan to the port the pump was originally plugged into. The fan kept spinning, and I think I am feeling some vibration from the pump as well.

Any idea what my issue may be?

 

[Mandrake4565]

Lets see a ss of the Spd and Memory tabs in cpu-z

Additionally I don't like IBT for stress testing, too much heat and I find it's not a reliable indicator. Prime 95 blend 2 hours works for me.

 

[Tyerker]

I have also played about 30 minutes of DOTA 2 as a test and temps never got over 41C.

Here's CPU-Z SPD and MEM. When my RAM was 1600 IBT Blue Screened, so I turned it down to 1333.

 

[RGone]

Any idea what my issue may be? = Gigabyte 990FXA-UD3 = Was a pretty good offender when the FX 8 cores first came out. Bios mis-set the default Vcore and many wrote in a thread over at OCN about the problem. Manually setting Vcore to 1.35 to 1.375 was the fix most did while waiting to see if Gigabut was going to fix the problem in bios. I don't use Giga so I have not looked back for that problem in a long time.
Gone...Again.

 

[Mandrake4565]

Try setting the Cpu Nb Voltage to 1.25 I believe on your board it's called Cpu NB Vid control. After that try running prime with hwmonitor open then post it back here.

 

[Tyerker]

So turn up the voltage a hair on the CPU?

I ran a Prime95 blend and two of the cores had errors. Does kinda sound like an inadequate voltage thing, in that case. I'll bump it up a notch and see.

 

[Mandrake4565]

So turn up the voltage a hair on the CPU?

I ran a Prime95 blend and two of the cores had errors. Does kinda sound like an inadequate voltage thing, in that case. I'll bump it up a notch and see.

You're running stock settings? You shouldn't have to but you can give it a try, I'd also try what I said above with adding voltage to the Nb. God I hope this isn't another gigabyte ud3 disaster.

 

[Tyerker]

Got to 60C. One core failed. As advised I boosted CPU VID and CPU NB by a couple notches each.
Load Line Calibration is Extreme (per the Bulldozer Sticky)

 

[Mandrake4565]

What did you set the Cpu V manually to? If you look it went as high as 1.488 under load and it still failed prime at stock clocks. Also what did you set the Cpu Nb Vid to? I have my fingers, but this is looking like another Gigabyte UD3 nightmare.

 

[Tyerker]

I am using the UD3 Rev 1.1, so I imagine any "nightmares" would probably apply to me.

Installing the latest non-Beta BIOS and chipset drivers. I'll double check what the voltages are showing up as.

 

[RGone]

looking like another Gigabyte UD3 nightmare.= When the Visheras came on the scene the FX-83xxs were failing badly. Bios set Vcore too low. Heck they even had a female supposed to be from Giga in one of the forum's saying they were looking into the deal. A few users found default bios Vcore was too low and most just manually adjusted and then the same FX-8350s that failed at default settings no longer did so. I really never heard if Giga came with a bios to fix the issue. I don't really care for Giga anyway with all the heck they have put ther customers thru with UD3 and UD5 mobos.
Gone...Again.

 

[RGone]

UD3 Rev 1.1 = in Rev 1.1 was really the last fairly good UD3 we know about. Rev 4.0 is said to cure the crap of the 3.0's but not enough Rev 4.0's showing up to know for sure.

 

[Mandrake4565]

I'll double check what the voltages are showing up as.

I had a gigabyte board with my 2500k and the Hwmonitor readings were in accurate also. If I'm concerned about voltages I just check with a multimeter.

 

[Tyerker]

Ya, no multimeter here.

So RGone, are you saying the 1.1 is actually one of the better UD3 Revisions?

 

[RGone]

Yes the better as far as UD3's go.

Set APM too disabled in bios if there.

HPC = High Performance Computer to Enabled. If there.

 

[Tyerker]

CPU Voltage: 1.3875
CPU NB Voltage: 1.25

Upgraded the BIOS and now CPU-Z shows a quad-core... Back to the BIOS I suppose.

-edit-

Found it. Had "One Core Per Compute Unit" selected.

 

[Tyerker]

So, looks like the BIOS Flash, if nothing else, was a marked improvement:


The voltages are holding much more stable to what I set in the BIOS. I set 1.3875V, and they never went past the 1.408 displayed, typically hanging around 1.392. Also, instead of multiple cores failing after 1 complete test, I was able to complete 4 tests before one core failed.

Like I said, not perfect, but with the fear you guys put into me after my initial post, I must say I am relieved things are looking in the right direction, at least.

 

[mtdewcowboy]

Way too much Vcore for the speeds you are running
I run at 4.4 with 1.385 and LLC set to high
Something is off somewhere else making it drop cores

 

[Tyerker]

People seem to think this 990FXA-UD3 is apparently notoriously bad with 8-cores.

I've been able to run everything stable up to 4GHz with the 1.3875V. Tried booting 4.2GHz and couldn't boot. Probably time to start fiddling with the FSB now and see if I can do some more.

 

[Mandrake4565]

People seem to think this 990FXA-UD3 is apparently notoriously bad with 8-cores.

It all depends on the revision.

I've been able to run everything stable up to 4GHz with the 1.3875V.

When? Last screen shot you posted it was still dropping workers at stock clocks?