Shed some light on this for me please

[keny]

I'm having a bit of trouble with this gigabyte board (990fxa ud3) as i cant seem to get the cpu to stop momentarily throttling down on the multi, ive dissabled all the usual power saving mumbo jumbo in the bios but it still does it .
I was just running a few early tests to see if it would clock at all and thats when i noticed it, i made a quick crappy video using my iphone as i know a picture speaks a thousand words and it shows the problem (i appologize for the quality and noise). any input on this would be greatly received as it is no good for benching as it is and may as well be a paperweight in the state its in now


VIDEO HERE

 

[kmo_9000]

I haven't used AMD cpu's in a while, but I remember installing some sort of "Cool'n Quiet" software to the OS to throttle it or something. That may be your problem, or you could double check to make sure the settings you changed are actually sticking in the bios.

You could also double check windows 7 power options to see how windows 7 is suppose to be managing the CPU.

EDIT: Didn't see you were overclocked to 4.9ghz. What temps are you getting at that speed? Does this happen when you run at stock?

 

[keny]

I don't actually know for sure what temps it's running but it's not hot I know that. I have disabled every option in the bios, including thermal control which would cause it to throttle when it got hot anyway, so it's not temperature related. Windows is at high performance and i don't know if it does it at stock, as I'm not really interested in running it at stock, I need it to run overclocked to bench with so I havnt even tried it at stock. It's a new one on me as every board I've owned you can disable the throttling via CnQ etc but this doesn't seem to work, and it's strange that it throttles under load and not idle (not to my knowledge anyway)
Mark

 

[RGone]

Keny, you got the first revision of that mobo (1.0) or the 1.1? The first one of those boards did n0t have a working APM = Disable in the bios that worked. So the cpu throttled under load.

 

[keny]

I have rev:1.1 bud, I'm currently running F5 bios and I'm wondering if a update may fix this problem. It does fluctuate the frequency no matter what state of oc, stock or not, it's just the amount that it fluctuates is greater the higher the oc. It's brand new and the NB on theese boards gets toasty as hell, is that normal I havnt messed with fx chips and this is my first one so I don't know if there supposed to run this hot but my old Deneb and thubans NB's never got this hot

 

[trents]

I certainly would try updating the bios. Gigabyte seems to have more problem with buggy bioses than the other major brand boards.

 

[keny]

Manage to sort this drama out, updated to the latest beta bios which gave me the APM option in the bios that RGone mentioned, disabled all the rest of the garbage and is now stable running benches @5ghz with maybe the odd fluctuation of a few MHz now and again

 

[trents]

Great! Keny, are you posting from your mobile device currently? I don't see a Sig.

 

[keny]

Yeah, mostly use my mobile to post and I don't have much of a sig anyway as I don't even have a desktop pc just boxes of boards and processors etc, mostly just bench my hardware and then if it survives I get rid of it and onto the next one, I should really make a desktop oc as I can hardly get on the laptop as my wife and daughter are always pounding Facebook or the net

 

[RGone]

Manage to sort this drama out, updated to the latest beta bios which gave me the APM option in the bios that RGone mentioned, disabled all the rest of the garbage and is now stable running benches @5ghz with maybe the odd fluctuation of a few MHz now and again

That APM thing is nice to have on the Rev 1.1 boards. I feel for those buying that first UD3 release.

What you cooling with? How much Vcore does it take you for 5.0Ghz? I just pushed 4.8Ghz for two hours of OCCT and yes they are hot and take some voltage oomph to get on up there. Mine is an older/earlier FX8120 and it is 'rumored the later FX-s clock a little better on lesser Vcore. I say rumored. I have no proof at all from my experience.

 

[keny]

That APM thing is nice to have on the Rev 1.1 boards. I feel for those buying that first UD3 release.

What you cooling with? How much Vcore does it take you for 5.0Ghz? I just pushed 4.8Ghz for two hours of OCCT and yes they are hot and take some voltage oomph to get on up there. Mine is an older/earlier FX8120 and it is 'rumored the later FX-s clock a little better on lesser Vcore. I say rumored. I have no proof at all from my experience.

I don't think it would be prime stable at the current vcore of 1.55 but it seems to run every bench ive tried up to now which is good enough for me, you're 4.8 is a good 24/7 clock and I would be happy if I could get that on 24/7. I'm cooling it with a TT Frio and I don't know how reliable the temp sensors are but it seems to run at around 45-47c mark when under full load, but that's with the frio's fans on full and it's running on the bench in open air so you could add atleast a few degs to that if it was in a case. I think I will see how far it will go on air, I think a few more 100mhzs is not out the question, I've read you're 8120 thread and I'm in the same boat as you were, being a noob to the fx, but able to spank a Deneb/Thuban should be a help as it looks almost the same bar the NB temp what's the max temp on these things? Is it the same as thuban? Plus what's the most vcore people are running as I'm curious as to how much they will take

Cheers
Mark

 

[RGone]

I guess in a few respects you and I are doing a similar thing. You run it till you are thru and the stuff goes on somewhere. Me I am playing with this used FX-8120. I have three other computers for work stuff. My work stuff is not overclocked. They are for work and not play or games. If I got to have the data...well it runs stock. Couple of old IT heads told me that about 12 years ago and I have followed that and have 99.8% up time without failure on my work rigs.

Playing, well that is different. I would not have a problem pushing the Core Temps and CPU Temp to 62c and 72c respectively looking at those temps in OCCT, HWMonitor or Open Hardware Monitor. They all scale very very close to the same temps since I have checked them.

I think I am accurate in saying this used FX-8120 has seen a good deal north of 1.55Vcore for hours at a time, before I got it. I am cooling it a little more than it was in it's previous life so it is not taking as much Vcore to go up the Mhz scale.

I can boot into windows a good bit higher than I can call stable as we understand stable generally. I can boot into windows and surf the web and post here in the forum at 4.3Ghz with 1.2865 set in bios using my W/c setup. Early testers of the FX-series late last year and first of this year, said they could do 4.5Ghz with 1.3Vcore...but I doubt seriously that was what we would call stable. I cannot do that.

I am trying to get 4.0 to maybe 4.2Ghz, 2 hours OCCT stable, using less than 1.34Vcore. Anywhere along in that range and those lower temps would be 8 cores at 4.0Ghz and probably doable for CM 212 EVOs like a walk in the park.

You are benching and I am playing another way. Hehehe. Good luck man and be well.
RGone...ster.

 

[keny]

That's what it's all about "the challenge" , if I manage to get this thing higher on air I will update and may even see if it will pass a 20 min prime at a high clock, I doubt it, but it's worth a shot.
Cheers
Mark

 

[RGone]

AMEN ! RGone...ster.

That's what it's all about "the challenge" , if I manage to get this thing higher on air I will update and may even see if it will pass a 20 min prime at a high clock, I doubt it, but it's worth a shot.
Cheers
Mark

 

[keny]

Had another mess about with this today and managed to bench at 5.1 and some benches would run at 5.2 but most crapped out at that frequency and I could not get it to boot at 26 multi so had to get the last 100 from reference clock, I went up to 1.625v and I got it to boot into windows a few times at 5.3 but within seconds it would bsod. I did notice in the boos the core control seems to only allow the disabling of two cores and not individual cores, which is wierd as I enabled core control and it only gives the option of disabling 2&4 as 1&3 are allways active no matter what, is this a fx architecture thing or is there another bios sub menu that I'm not seeing ?

Mark

 

[rescuetoaster]

I have rev:1.1 bud, I'm currently running F5 bios and I'm wondering if a update may fix this problem. It does fluctuate the frequency no matter what state of oc, stock or not, it's just the amount that it fluctuates is greater the higher the oc. It's brand new and the NB on theese boards gets toasty as hell, is that normal I havnt messed with fx chips and this is my first one so I don't know if there supposed to run this hot but my old Deneb and thubans NB's never got this hot

Mine gets quite hot, so I would say probably lol.