gigabyte 990fx ud5 problem

[Kei]

I spent the weekend sorting my 990fxa-ud5 out with an 8320. Been trying to get an overclock working and encountered similar problems as tripp. Clock speed and voltage jumping between 4.2 @ 1.35V and 2.9 @ 1.1V under load. Switching off c1, c6 and turbo and turning HPC mode on seems to have stopped it from jumping about. Temperatures definitely were not an issue as VRM sinks were warm at most and no temperature on HWmon was above 58 degrees.

Oddly enough, now it starts up for a split second, cuts out for about 3 seconds and then starts up normally. This occurs on a restart from windows or saving from the bios.

 

[Tripp]

yeah hpc mode was my problem too mate, how did you do your overclock?

 

[Kei]

I've tried simple multiplier changes as it wouldn't boot if I upped the bclk. (Like I have with my older phenom ii system) I'm limited by the cooler as its a noctua nh-l12 due to the skeleton case. 4.1-4.2 seems to be the limit thermally. (200x21 @ +0.025V LLC @ high)

 

[Tripp]

ahh i see what was you're max temp when stressing? i overclocked my on the bclk gotta tweek the overclock at the mo, my max temp was 62c so far. going for a custom loop soon so i should get better temps then

 

[Kei]

Highest temperature I've seen is 68 degrees on TMPIN2, TMPIN1 was 64 degrees. This was running 4.2GHz at 1.37V with C1, C6 and turbo turned off. (cool 'n' quiet & HPC turned on)

 

[Tripp]

i thought the max safe temp was around 65c, i got mine to 4.4 at 1.356v i held off going higher because of the temps.c1,c6 an turbo and cool an quiet off hpc on

 

[NickOmega]

yeah a you are the second person to say they used amd overdrive, i haven't used it myself as i thougtht doing an overclock in the bios was more stable. i will have to check the vrm's later as i have some stuff to do today, i will let you know tho fella. also my vcore voltage selector was weird to, you cant select a voltage you want it does it some weird way that is quite confusing. i was told to try and keep it around the 55c/60c (HW Monitor) at full load, does that sound about right to you?

Just to clarify, I wasn't using AMD OverDrive to actually overclock. It was just a suggestion I found online to enable and disable the boost in OverDrive to see if it stopped the throttling/underclocking during stress testing (as far as I could establish that was the only time it was underclocking).

The VRM heatsink on my UD3 was never hot (could still mean that it's not drawing the heat away effectively), NB is a completely different story however and feels far hotter than anything reported in HwMonitor.

Speaking of which, everything I could find points to TMPIN2 being the NB temp for my board, but it's a damn sight warmer than the 26° (identical value/min/max as CPU package) it's reporting at the moment.

Oddly enough, now it starts up for a split second, cuts out for about 3 seconds and then starts up normally. This occurs on a restart from windows or saving from the bios.

Mine does exactly that too :/

 

[Tripp]

Just to clarify, I wasn't using AMD OverDrive to actually overclock. It was just a suggestion I found online to enable and disable the boost in OverDrive to see if it stopped the throttling/underclocking during stress testing (as far as I could establish that was the only time it was underclocking).

The VRM heatsink on my UD3 was never hot (could still mean that it's not drawing the heat away effectively), NB is a completely different story however and feels far hotter than anything reported in HwMonitor.

Speaking of which, everything I could find points to TMPIN2 being the NB temp for my board, but it's a damn sight warmer than the 26° (identical value/min/max as CPU package) it's reporting at the moment.

ok dude i have sorted the throttling now anyway, the vrm's are transferring the heat ok as far as i am aware.

i was stress testing again last night doing another blend test, thinking that i had sorted the problem from the night before. i however haven i got a fail 6 and a half hours in on one core, does anyone have any advice please?

 

[Zoot]

ok dude i have sorted the throttling now anyway, the vrm's are transferring the heat ok as far as i am aware.

i was stress testing again last night doing another blend test, thinking that i had sorted the problem from the night before. i however haven i got a fail 6 and a half hours in on one core, does anyone have any advice please?

Have you tried going at the bus speed or are you just upping the multiplier?

Some (from what I've read) have had more success by upping the Bus speed.

 

[Tripp]

Have you tried going at the bus speed or are you just upping the multiplier?

Some (from what I've read) have had more success by upping the Bus speed.

i have been doing it through the bus speed, but i keep getting fails on a blend test after like 6/7 hours

 

[Kei]

Mine does exactly that too :/

Damn, wish gigabyte would release a fix for it as it's too annoying and doesn't do my hard disks any good.

 

[Kei]

I seem to have found the cause of the cut out at post and a way around the issue that also solves the fluctuating clock speed and voltage.

The cause is turbo mode, known as CPB or core performance boost in the bios. Turning it off causes the cut out issues. (bios FCb still exhibits it)

I've found that by having turbo and cool 'n' quiet on but leaving APM, HPC & C1/C6 off, the cpu clocks down at idle only and remains at a constant speed and voltage whilst under load. I set the turbo multiplier to be the same as the standard multiplier and it keeps the core speed constant.