OC gone bad?

[Itachiii]

Hey guys, I recently started overclocking an old build which consists of:
FX 6100 BE
gigabyte ga-970A-ud3 rev 3 (updated bios to FC version)
2x HD 6850 xfire
2x4gb Kingston HyperX 8gb 1600MHz

So, I was able to overclock my CPU as a first step to 4.5GHz @ 1.398V (max 43C at Prime95 for 30 minutes) just by bumping the multiplier. After that I went to BIOS, and I did the following: HTT set to 231MHz and the multiplier to 19.5 (4.5Ghz). Bumped the CPU NB frequence to 2200MHz and the HT Link to 2400. Increased the CPU NB voltage to 1.225V and left the memory multiplier, timings and voltage on auto. Everything was smooth and fine again at Prime95 for another 30 minutes. In the end, I bumped the DRAM multiplier to x9.33 at start and set the timings to a preset that it has called expert (9-9-9-27-32). I started getting error on 2-3 cores on P95 like "illegal summout" and "resulting sum was X expected X". I raised the RAM voltage to 1.65v, I dropped the multi to x8 yet the same happened. I attach a few pictures of the errors as well as from CPU-Z to check them out as well.


Any thoughts in stabilizing it? I had turned off all the C1E state things ofc, I had LLC on extreme mode and my temps were not going over 43C

 

[RGone]

Straight up truth as we have seen and know it...

...That UD3 Rev 3.0 motherboard by Giga has caused more pain and grief than any other motherboard coming thru and there have been less than 10 of them and at last count about 6 or those 10 moved on to other brands of mobo and away from that board in Rev 3.0. Final outcome of the other 4 boards is unverified.

There is a beta bios for that mobo floating around somewhere. Have not had anyone in here in about 8 mos. or greater so have not had any need to know where that beta bios is still located for download. It was the only partial salvation for most users of the Rev 3.0 board. You might see if you can g00gle it up.

There was a fairly long thread in the AMD Mobo forum section that had a user with that board and he listed off at least 10 issues he had with that board and i vaguely remember that in that thread was a link to the beta bios for those UD3 Rev3.0 motherboards. Might look for that thread. Without that bios and what it gives access to in the bios, there is nothing much can be done for those boards and a decent overclock over the long term.

I just got thru reading parts of a +300 page thread at OCN about those boards beginning in Rev 1.0 thru 4.0 and there was never any real success for the Rev 3.0 users. They just got bye and cussed. Look for the Beta bios for the Rev 3.0 mobo and that is your best bet according to what we have been thru here in this forum.
RGone...ster.

EDIT:
Gigabyte GA-970A-UD3 rev.3.0 beta bios 970AUD33.FBd was the bios that came out in beta form a little over a year ago, but I don't see it anywhere now for download. It was reported to help that Rev 3.0 board a good deal and at least make it usable in the normal idea of things.
END EDIT.

 

[Itachiii]

...That UD3 Rev 3.0 motherboard by Giga has caused more pain and grief than any other motherboard coming thru and there have been less than 10 of them and at last count about 6 or those 10 moved on to other brands of mobo and away from that board in Rev 3.0. Final outcome of the other 4 boards is unverified.

There is a beta bios for that mobo floating around somewhere. Have not had anyone in here in about 8 mos. or greater so have not had any need to know where that beta bios is still located for download. It was the only partial salvation for most users of the Rev 3.0 board. You might see if you can g00gle it up.

There was a fairly long thread in the AMD Mobo forum section that had a user with that board and he listed off at least 10 issues he had with that board and i vaguely remember that in that thread was a link to the beta bios for those UD3 Rev3.0 motherboards. Might look for that thread. Without that bios and what it gives access to in the bios, there is nothing much can be done for those boards and a decent overclock over the long term.

I just got thru reading parts of a +300 page thread at OCN about those boards beginning in Rev 1.0 thru 4.0 and there was never any real success for the Rev 3.0 users. They just got bye and cussed. Look for the Beta bios for the Rev 3.0 mobo and that is your best bet according to what we have been thru here in this forum.
RGone...ster.

EDIT:
Gigabyte GA-970A-UD3 rev.3.0 beta bios 970AUD33.FBd was the bios that came out in beta form a little over a year ago, but I don't see it anywhere now for download. It was reported to help that Rev 3.0 board a good deal and at least make it usable in the normal idea of things.
END EDIT.

Hey RGone, thanks for the answer! Yea there is a Beta BIOS in Gigabyte's site, should I download it? It came out 4-5 months ago. Apart from the mobo being ****ty, my settings were fine? Maybe I did something wrong?

 

[RGone]

See if you can post just these pics...

...as outlined below and how HWMonitor is opened up for top volts all the way down thru package temps. Need the whole thing. Then the three cpuz screenies. In other words not really seeing the complete set of items we need to see in order to make a decent suggestion or suggestions.

One other thing before I put up the list of what to dos and the images below and that is the beta bios I refer to is over a year old. No idea what the last beta they have over at Giga is about. When I worked for DFI Inc, we had a beta bios that often preceeded an offiicial bios that incorporated most of what had been beta. Whether or not that is so with Giga is not something I would know. i only know that something like 13mos ago there was a beta bios that was a huge help to users of the Rev3.0 board that was such a pain. Me being me...well bios flashing ain't nothing to me. Did it so much with DFI that it is like sleeping with your brother to me. Hehehe. I likely would try one later than what you have for sure.

Now you put up a lot pics but the full pic of HWMonitor free was not showing as I will show below. And we need the 3 cpuz screenies only or mostly. I will show what we ask for below.

Oh crap I see you have Kingston ram. Deeyamskis. We have had a flood of users with Kingston ram and problems. Often have to set timings manually and up the ram voltage slightly and also up CPU_NB voltage as I see you have done some aleady, but maybe not enough.
RGone...ster.

This is what we need to see for sure and a real good starting point.

Normally during setup and testing we disable C1/E, C6, Cool N Quiet, APM, TurboCore and in windows performance manager itself we set to "performance" mode. ALSO if you have HPC in bios you would ENABLE it. That way there are not "other" settings messing with the overclocking process. Some of those settings are not available on all models of cpu but where in evidence we disable for setup of overclock process.

CPU Tab in CPUz from CPUID com

Memory Tab in CPUz from CPUID com

SPD Tab in CPUz from CPUID com

And this is screen capture of HWMonitor (free version) from CPUID com
HWMonitor has been scrolled enough and large enough to show Min/Max of Voltages and includes the CPU CORE TEMPS / "package" temps fully visible. Latest versions of HWMonitor show the CPU Core Temp as " a Package Temp" and is only shown as a single temp since there were never multple, individual core temp sensors anyway.

This capture is made of HWMonitor after it has been open and running on the desktop logging Min/Max temps and voltages while Prime 95 was running Blend Mode test on all cores for at least 20 mins and then the capture of HWMonitor was made and it shows the Min/Max temps and voltages before P95 Blend was started and while running P95 Blend mode and gives much greater insight into how the system is performing without guessing.

In order to attach screenshots of INDIVIDUAL images as suggested, first crop and capture the images with Snipping Tool found in Windows Accessories or equivalent. Then click on Go Advanced, a button at the bottom of every new post window. Then click on the little paperclip tool at the top of the Advanced post window when it opens. Clicking on the paperclip tool brings up the file browser/upload tool and the rest is fairly obvious.

 

[Itachiii]

Thx a lot for all the details!! I disabled all those things in BIOS but didn't change the power manager in Windows to perfomance TBH.HWMonitor is having trouble with voltages section. It shows a lot of downgraded voltages (e.g on the 3.3V it shows 2.4V) and it got me worried when I first realized it. So I swapped out the old PSU (a crappy 400W) before OC'ing and I swapped in a Coolermaster GX 550W. But the voltages on HWmonitor remain fault. I will take new pics and post them when I get back to OC'ing coz I'm out of town for the New year's eve.

 

[RGone]

HAPPY NEW YEAR.

I never worry about the P/S voltages since I always know what they are by DMM.
RGone...ster.

 

[Itachiii]

HAPPY NEW YEAR.

I never worry about the P/S voltages since I always know what they are by DMM.
RGone...ster.

Happy new year!!! Love, health and wealth for you and your family! What do you mean by "DMM"?

 

[Silver_Pharaoh]

There's a BETA BIOS on Giga's download page, is this it?
Was released October 7th 2014

Also, isn't Kingston+AMD a bad idea?

 

[RGone]

I would hate to think all Kingston is a bad idea lately with AMD but it is something to keep in one's mind when trouble with stability just "keeps" coming around in a particular system. Would I suggest to one of my buddies to use Kingston Ram and an FX processor? Nope, but that does not mean at some point it does not work.

No the bios I refer to is from about Oct/Nov of 2013 and not 2014. Working from memory here since I never owned one of those UD3's. I only know from what UD3 Rev3.0s said that it was the only thing that gotten them by. I just thought get that bios or one like it on the board before pulling all our hair out trying to tweak for stability.
RGone...ster.

 

[Itachiii]

I would hate to think all Kingston is a bad idea lately with AMD but it is something to keep in one's mind when trouble with stability just "keeps" coming around in a particular system. Would I suggest to one of my buddies to use Kingston Ram and an FX processor? Nope, but that does not mean at some point it does not work.

No the bios I refer to is from about Oct/Nov of 2013 and not 2014. Working from memory here since I never owned one of those UD3's. I only know from what UD3 Rev3.0s said that it was the only thing that gotten them by. I just thought get that bios or one like it on the board before pulling all our hair out trying to tweak for stability.
RGone...ster.

I installed the FC version (released at 28/1/2013). I don't know if it's the mobo or is it me? I got the DRAM multi to x8 (since x9.33 was not stable), but I couldn't figure how to change the timings.(???) Later on when I cooled down, I thought of maybe keeping everything at stock speeds and load up the most tight timings, play around with FSB(keeping the CPU multiplier low to preserve stable stock clocks) and the DRAM voltage till I get a stable set of timings so I can continue my OC'ing from there on?
1)Could timings be the problem?
2) would the procedure I described above work?

 

[RGone]

At the first page of the MIT settings about the third choice down is Advance Memory Settings. That is where you adjust the memory timings. At least that is what the manual for the UD3 Rev3.0 shows.

I would say actually that I would start at the start and run P95 Blend mode and post up the 4 screen captures as suggested earlier and then we all can get a look at what is going on. That is the procedure we have followed for a few years now rather than us guessing what is happening at a computer miles away from us in general.

IF you do not wish to do that, then yes you can tweak settings around on your end and see what you get. There is a flaw in the logic of tweaking ram timings at low cpu speed and then accelerating the cpu speed. I see such suggested now over many years, but I still don't believe or use such testing because the faster the cpu goes the ram must do its work at a faster and faster speed. Ram adjusted at slow cpu speed may well not be able to keep up with cpu when timings are so tight the ram squeeks.

Therefore I set ram to default speed and timings and add a slight added voltage to the ram and CPU_NB and then I begin to up the multiplier of the cpu to clock up the cpu. While doing this the ram should be a NON-issue with it at default settings of the ram maker. Makes sense to me.

Now that said in paragraph above, if upping the multiplier and testing at each step up in the multiplier results in "still" no stability...I say to self, the dang ram is a POS at default, so I go and lower the ram speed one notch below its rated speed and if the system seems stable when raising the cpu multiplier and thus cpu speed >> I run and get a hammer and break the ram. Hehehe. No not really. I just know the ram is going to need some serious tweaking once I reach a decent cpu clock speed. This is me. How I do any mobo or any brand cpu. YMMV.

RGone...ster.

 

[Itachiii]

At the first page of the MIT settings about the third choice down is Advance Memory Settings. That is where you adjust the memory timings. At least that is what the manual for the UD3 Rev3.0 shows.

I would say actually that I would start at the start and run P95 Blend mode and post up the 4 screen captures as suggested earlier and then we all can get a look at what is going on. That is the procedure we have followed for a few years now rather than us guessing what is happening at a computer miles away from us in general.

IF you do not wish to do that, then yes you can tweak settings around on your end and see what you get. There is a flaw in the logic of tweaking ram timings at low cpu speed and then accelerating the cpu speed. I see such suggested now over many years, but I still don't believe or use such testing because the faster the cpu goes the ram must do its work at a faster and faster speed. Ram adjusted at slow cpu speed may well not be able to keep up with cpu when timings are so tight the ram squeeks.

Therefore I set ram to default speed and timings and add a slight added voltage to the ram and CPU_NB and then I begin to up the multiplier of the cpu to clock up the cpu. While doing this the ram should be a NON-issue with it at default settings of the ram maker. Makes sense to me.

Now that said in paragraph above, if upping the multiplier and testing at each step up in the multiplier results in "still" no stability...I say to self, the dang ram is a POS at default, so I go and lower the ram speed one notch below its rated speed and if the system seems stable when raising the cpu multiplier and thus cpu speed >> I run and get a hammer and break the ram. Hehehe. No not really. I just know the ram is going to need some serious tweaking once I reach a decent cpu clock speed. This is me. How I do any mobo or any brand cpu. YMMV.

RGone...ster.

Thx once again RGone.. Nah, i prefer sticking with the good old fashion way, I will post screens tomorrow, because I want to get it done right! It's just so frustrating..

 

[RGone]

It's just so frustrating..

No telling how many of these FX things I have setup. The one I currently use has a 'dang good' CHV-z in it but now that I can look back my issue was the cpu. But 2.5 mos ago when I was looking for stability etc...I did not remember this cpu that I got used but had tested a year or so ago and sent back to its' owner after testing.

Sum beeatch ran me around and around because "I" wanted to setup things in a particular manner. A manner that was fine on heavy duty water but not wurf a shett on this C-type air cooler on top of the FX-8350. But the thing about this shett is we learn zero when it all works and tons when we have to actually work thru an issue. IMC on this FX-8350 is weak compared to my others. So when I tried to run CPU_NB fairly high > it needed CPU_NB voltage > but CPU_NB voltage was making cpu run hot. Shett shett. Point was I kept trying for high CPU_NB but knew I could not add voltage to the CPU_NB and thus P95 failure after failure until I slowed the CPU_NB. Hours of adjusting and playing for nothing because I plugged in the numbers I wanted instead of beginning at the beginning. If I had started slow instead of thinking the cpu would just do what I wanted...Hehehe, which it did not, I would caught the problem a few hours earlier. Lowered the CPU_NB after stopping and 'thinking' (ah how we hate that) and saying oh man, I got to lower that CPU_NB and then right smack dab into stability.

Yep I understand frustrating, since I seem to do it to my ownself off and on. Luck man.
RGone...ster.

 

[Mandrake4565]

Just on the Gonsters point on the IMC, I recently started doing a little testing on my chip to find out exactly how much Cpu Nb voltage I need to run my ram above 1866 Mhz. Then I want to find out how much more I would need to the Nb Freq above 2200 with it at 1866. I haven't had time to test running the Nb Frequency higher yet but in order for me to run my ram at 1866 Mhz I need 1.3875 Cpu Nb voltage. I testing using Occt 3 hours and also Prime 95 blend 3 hours, run separately not at the same time. I was able to run my ram at 1600 Mhz with the Cpu Nb Voltage set at 1.20 v but needed the additional .1875 v to run 1866 Mhz.

 

[RGone]

"Manny" your IMC sucketh too. But when we don't really test and test...we get in that jam I spoke of above. Thanks for sharing in the new 2015.
RGone...ster.

 

[Itachiii]

"Manny" your IMC sucketh too. But when we don't really test and test...we get in that jam I spoke of above. Thanks for sharing in the new 2015.
RGone...ster.

Since I don't have much of an experience in OC'ing, I thought that the bad "moments" are less than the good ones.. But I guess it dorsnt work that way.. So first things first, I'll start with the pics at stock clocks and then I'll move step by step.

 

[Mandrake4565]

"Manny" your IMC sucketh too. But when we don't really test and test...we get in that jam I spoke of above. Thanks for sharing in the new 2015.
RGone...ster.

Funny you say that Gonester and I'm going to post what I find in Just Putzun when I'm done. What I find interesting is I do need a bunch of voltage to run my Ram higher but benching I've run the Nb Freq around 2900 Mhz at times. I have always said that I feel I have a weak Imc on my chip but maybe it's more that mine is just a leaker. Now my benching is no where near stable therefore the reason why I want to test and see how it reacts when testing for stability.

Since I don't have much of an experience in OC'ing, I thought that the bad "moments" are less than the good ones.. But I guess it dorsnt work that way.. So first things first, I'll start with the pics at stock clocks and then I'll move step by step.

Itachiii, overclocking isn't hard per se, it's a process. It takes patience and a lot of people think that just raising the Cpu Mhz is overclocking, which it essentially is. Though, in order to get the whole system working in harmony and running at it's peak performance, there is a lot more to it. Having the Ram speed and Nb freq raised helps with the overall performance and overclocking those takes a bit more patience. I've been at this for the better part of three years now and I have learned more from benching and from the help of people on this forum, Rgone, Johan C/D, Bassnut just to name a few. If you are really looking to learn about Ocing and understand how to get the best performance from your rig it takes time. A lot of us do testing and post it in the Just Putzun Round thread. There is a lot of nonsense in there but also a ton of knowledge if one really takes the time to read through it.

 

[Itachiii]

Funny you say that Gonester and I'm going to post what I find in Just Putzun when I'm done. What I find interesting is I do need a bunch of voltage to run my Ram higher but benching I've run the Nb Freq around 2900 Mhz at times. I have always said that I feel I have a weak Imc on my chip but maybe it's more that mine is just a leaker. Now my benching is no where near stable therefore the reason why I want to test and see how it reacts when testing for stability.

Itachiii, overclocking isn't hard per se, it's a process. It takes patience and a lot of people think that just raising the Cpu Mhz is overclocking, which it essentially is. Though, in order to get the whole system working in harmony and running at it's peak performance, there is a lot more to it. Having the Ram speed and Nb freq raised helps with the overall performance and overclocking those takes a bit more patience. I've been at this for the better part of three years now and I have learned more from benching and from the help of people on this forum, Rgone, Johan C/D, Bassnut just to name a few. If you are really looking to learn about Ocing and understand how to get the best performance from your rig it takes time. A lot of us do testing and post it in the Just Putzun Round thread. There is a lot of nonsense in there but also a ton of knowledge if one really takes the time to read through it.

Hey Mandrake, yea I really want to learn about overclocking and of course patience and experience matter. I'll read through the posts and I'm pretty sure I will learn a lot. In the meantime, till I get to start OC'ing again my friend's pc, I wanted to tweak mine a lil bit (the comp in the description). Since it had the multiplier to x23 (max x24), I raised the BCLK to 170 to get something like 4.14GHz @ 1.318V (had to fiddle with the Dynamic Vcore, set the Vcore to normal and raise the DVID). Whatsoever, temps would not let my CPU operate at that clock, so much for higher. Ran P95 for ~10 seconds and Core#1 hit 100°C and the system shut down!!

 

[Mandrake4565]

Hey Mandrake, yea I really want to learn about overclocking and of course patience and experience matter. I'll read through the posts and I'm pretty sure I will learn a lot. In the meantime, till I get to start OC'ing again my friend's pc, I wanted to tweak mine a lil bit (the comp in the description). Since it had the multiplier to x23 (max x24), I raised the BCLK to 170 to get something like 4.14GHz @ 1.318V (had to fiddle with the Dynamic Vcore, set the Vcore to normal and raise the DVID). Whatsoever, temps would not let my CPU operate at that clock, so much for higher. Ran P95 for ~10 seconds and Core#1 hit 100°C and the system shut down!!

Cooling is the name of the game, if you really want to push. You should see some of the things we do to get our rigs to really go fast when benching. My rig used to visit my back deck in the winter to get it to really go.

 

[Itachiii]

Cooling is the name of the game, if you really want to push. You should see some of the things we do to get our rigs to really go fast when benching. My rig used to visit my back deck in the winter to get it to really go.
View attachment 158132

HOLY SH******. I really didn't see that coming HAHAHAHAHA