Issues with new 970a-ud3p motherboard

[Hatekraze]

-Antec 1100
-Windows 10 64-bit
-Gigabyte 970A-UD3P motherboard
-FX-8150 processor with h105 clc
-4x4gb sticks corsair vengeance 1866mhz (tall)
-Gigabyte 7870 oc edition
-TX750 psu

Ugh... im ready to snap this thing in half. Ive been fighting with it for around 15 hours now and havent made any real progress. Maybe someone here can help me figure this stuff out. If not, i may as well return it because even my Asrock extreme3 ran without issues up until the smaller VRMs became an issue overclocking. Everything was running well before hand on the old board, overclocked to 4.3ghz stable and running 1866mhz on the ram, stable for 12+ hours on prime95 all torture settings. Now this board hardly runs stock. Ill begin with a list of my issues so far

1) I hate that bios with a passion. Fought with it for hours before i could even figure out how to adjust anything. Nothing anyone can do about it, just wanted to rant

2) Ok, now first is the only drivers that are for windows 10 are audio, network, and usb 3.0. I have a separate audio and wireless card so those are pointless. When i try using the disk, it glitches up a ton and says its installing, but then most drivers failed as they arent compatible with windows 10. My bad, the board was recommended to me and i didnt think to check the operating system. But dont i need some other drivers to work? Or does the cpu and everything else work independently? All i know is when i boot up it pops up a window saying "Driver cannot release due to failure".

3) I cannot control my cpu 4pin fans with speedfan as i was always able to before. ANd on a side note, the amount of fan headers on this is disapointing, without my fan controller id be screwed. Ive tried any setting i can think of in bios and speedfan. Any changes i make in speedfan go unnoticed. Apparently the fans can be controled with the Easytune software (thats all the manual says about manual fan control), but that again is not windows 10 compatible to even try. So i guess im locked to relying on the bios to regulate temps. The bios set to manual gives a PWM slope to adjust from .75 to 2.5, so i take that as 2.5 meaning the fan revs up sooner. But i cant tell if it does anything since my temps stay down even in prime with the new cooler. From what i can tell about my temps anyways...more on that in the next point

4) This ones pretty simple, but a worry for me considering my goals to overclock. I used to have CPU and Core temps to monitor the individual cores and the socket temps. Now in speedfan the CPU temp is gone, but i still have the core temp. I think the cpu temp is replaced now as "Temp2" as it appears to go up with prime. In HWMonitor, it has CPU package temps displaying as the core temps would, and under other temps TMPIN1 appears to be like the socket temps were. If this is correct and i can feel safe monitoring these temps as such, im fine. Although its a bit anoying that it wouldnt label the CPU as such. I bought this motherboard to be able to overclock the FX(if i can ever get there), so i just need to be sure im seeing the right temps (EDIT: temp3 also seems to go up under prime, but a couple degrees lower than temp 2. Motherboard temp?)

5) The most important and most frustrating one... Anything above stock fails to boot for some reason. As i said before, i had it overclocked before at 4.3 stable without issue before. After that VRMs started to throttle, thus the "upgraded" board. Lets start with the ram. Stock in bios is 1333mhz, my ram is 1866mhz. I tried enabling the XMP profile with the stock 9-10-9-27 timings and it goes past bios but windows blue screens me with a kernel error. I looked into it a bit and it seems others have contacted gigabyte about it and they replied saying its a limitation with the AMD chips only being about to use 2 sticks at 1866mhz. Thats complete BS because i had it running for years on my old motherboard with the exact same setup aside from the motherboard. So if i stay with this motherboard, im locked down to 1333mhz or half my ram capacity. And for gigabyte to blame the chip is typical. As for the cpu, i tried my old overclock settings, and then i tried a few steps below that (with generous voltage), and i couldnt get it to boot again. It seems that it doesnt apply the voltage increase i set it to for some reason. Based on Coretemp and etc programs. I bought this motherboard to be able to overclock higher and i cant even get it to run stock hardware settings.

And now ill end my rant... Please help if you can with any of the questions. Is gigabyte just crap?

 

[CountMike]

Memory controllers in that chip are good to 1600MHz when both are used like in the case of 4 x4GB, at least those are AMD recommendations. RAM can also use higher voltage for stability.
Gigabyte is not "just crap" but UEFI sucks, you are better off using legacy mode. Also shouldn't relay at XMP profile, better off to set RAM multiplier and voltage manually.
Speedfan never worked for me right on AMD system, setting fan speeds thru BIOS works best. For following temps and voltages, AMD Overdrive works best.

 

[Hatekraze]

Thanks for the reply. I tried 1600 and it wouldn't boot up still. I'm currently working on overclocking with fsb to see what I can squeeze out. Then I'll try my hand at lowering the timings once I'm done overclocking the cpu. Although I've never adjusted the timings before, I believe this is my best bet. I'm not sure what's up with this board, but it seems to not be acting up anymore. The pop ups stopped on startups, and I've been able to raise the cpu clocks now. I didn't change anything so I don't get it, maybe something updated or installed in the background. I don't know, I just hope it's not an intermittent thing.

Also, sorry for the long rant. I may have come on a little strong lol Technology is great when it works, but when it doesn't I'd rather use it as a frisbee

 

[CountMike]

I hear you, one of my phones ended up as skipping stone on the water but desktops are just to heavy for that. I have a habit to set everything in BIOS to manual and turning off all turbo and power saving modes before attempting OC.

 

[Hatekraze]

I had done that before too. Like I said, I don't understand it. Oh well I guess. Do you know if cpu socket temps always show up? Like, if I'm watching that no temps rise too high I'm ok? Or is there a chance it's not showing at all. I'm pretty sure it's that Temp2 as it seems consistant to what the cpu temp would read. But it's creeping me out that the core temps have been a degree or two higher under load at the higher clocks. This cooler is much better than my previous setup, so maybe it is able to disperse the heat on the chip itself faster than the cores produce? Before the socket temps were always consistently 10 degrees higher than the core. But this rad has been great for cooling. I have the h105 38mm thick rad with push pull (upper 120mm slot is the stock sp120L fans, the bottom is a set of SP120 performance editions)

Also what is HPC mode and how does it affect the overclock. I've looked into it and most say to enable it. But would this be why I seem to be seeing core clock spikes in hwmonitor's min/max records? It doesn't throttle that I've ever seen. But for example, I'm clocked at 4.5ghz at the moment and ran it last night on prime. I had a core fail near the end, but hwmonitor showed min 4200mhz and max 4800mhz. I've also seen max 5200 with the usual min 4500 on a previous run. I'm thinking it could be my LLC settings. I'm set to medium LLC and 1.45v. Idle I sit at around 1.45v, and depending on load amount I see 1.426-1.46v in hwmonitor's records regardless. No higher or lower than that even with clock spikes. Perhaps this is hwmonitor getting incorrect readings? I've read that people see false voltage spikes but haven't run across a MHz spike. The low LLC setting causes a drop to 1.42v at 4.5ghz, and more vdroop as I increase the clock

 

[CountMike]

I keep LLC at normal so it doesn't hit over 1.42v even at 4.8GHz but rarely goes over 1.4 in normal use.
Hwmonitor is not reliable with AMD systems and FX temperature measurements are at best flaky. I'm getting about 5c difference between core and socket temps under moderate load but can jump to 10c under full prolonged road. That's all in AOD, I gave up on Hwmonitor, Speedfan and Coretemp because they show sometimes some ridiculous data. Prime isn't very AMD friendly either, AOD has adequate stability tests.

 

[Hatekraze]

Ok, thanks for the tips. I stress test with speedfan, coretemp, hwmonitor, and task manager (just for the up time clock so i can see if it freezes lol). Ive only used prime95 because it throws errors more readily apearantly. Theres two sides to the story... One says hey use programs other than prime95 because it throws errors too easy that apearantly never would have apeared otherwise. The other side says the whole point of stresstesting is to be 100% sure youre stable no matter what. I tend to lean towards the latter to be safe. But who knows, im not exactly an expert in overclocking lol

SInce this has turned more into an overclocking thread, i think ill make a new post in that section instead to get a few more experienced veiws. Thank you

 

[CountMike]

No problem but prime 95 and stuff like that is made for Intel but also stress testing the hell out of system is not a guarantee it's 100% stable, leave it running long enough and every system will throw some error because it can wear out everything. It's more of post mortem tst and for troubleshooting.

 

[Hatekraze]

Have you overclocked the FX chips much? Have you found a better way to stress test for it to be reliably stable? I figured it seems like prime95 being prone to throwing errors made it so if i cold pass prime for long enough anything else would be a breeze

 

[GhislainG]

Have you overclocked the FX chips much? Have you found a better way to stress test for it to be reliably stable? I figured it seems like prime95 being prone to throwing errors made it so if i cold pass prime for long enough anything else would be a breeze

Prime95 should not throw errors on a stable system; the make and model doesn't matter.

 

[CountMike]

I OC everything that gets in my hands. Started with 386 chips when some mods had to be done to MB. Oced old Athlons with Barton core using graphite pencils to unlock them. Phenom ii Black edition were also easy to OC but I guess that AMD gave up on barring unlocking with FX processors so they are all in essence Black Edition.
Up to now I had 8300, 8350 and now 6350. Experimented with BLCK/FSB OC but performance gains didn't pan out to be of much consequence, straight up multiplier OC with corresponding voltage settings proved to be most economical. I had this last 6350 up to 5GHz but settled it down to 4.8 as most economical under air cooler (Mugen3 and two fans).
All you need is some luck with silicone lottery and C0 stepping and of course to do it on good MB. This MB has socket good for over 140W and on 4.8GHZ it's about 138W VRM seems to be most defining when it comes to overclocking. Voltage droop can be severe otherwise.

 

[Hatekraze]

i upgraded from the extreme3 because of the vrms. Loved the board otherwise. How much better are the 83XX compared to 81XX? I thought about upgrading, but its pretty pointless from what i can tell. Only upside is that as i slowly replace parts for myself, the old stuff goes into storage for when we can build my girlfriend a better gaming tower than her dell laptop

And what is C0 stepping? I keep seeing stepping mentioned, but since everyone else seems to know what it is i never see a descripton

 

[CountMike]

You can find it in CPU-Z easy enough, basically it's grading processor by it's quality because they don't all come out identical from under the press so to speak. Another, coarser step is so called "Bining" and that's how they end up with let's say FX8300, 8320, 8350, 8370, they are all same but higher number it is, better they made it thru quality control. Even worse case are Athlon II and Phenom II x3 and even some x2, they had a core or two so bad that they didn't work at nominal voltages at all but could be possibly unlocked in some cases by using higher voltages and so get "Bx" processors.
Difference between FX x1xx and x3xx models is in the cores, first ones have Bulldozer cores and second Vishera cores that were improved later on. In practical terms, Vishera is better optimized and need less power for same work and so OC easier.

 

[Hatekraze]

Ok If I break this one I'll likely go to a 8350 so stepping isn't a stress test? Lol

 

[Hatekraze]

I have possible issue that maybe you can help me with. Using hwmnitor as i test on prime, im seeing odd values for my min and max clock frequencies throughout the hour long tests im currently doing. It appears to spike up or down at times, only the frequency and nothing wth the voltages. For example, testing at 4550mhz passed an hour of small ftts, min voltage 1.45v/max 1.478v, but the min/max records showed 4050mhz min/5050 max. No failed cores, no crashes, no vcore changes, no temp spikes, and with me trying my hardest to stare at the screen the next time i couldnt see the spike when it did a simular change up to 4800mhz at some point. My google searches have come up with nothing but temp and vcore spikes

 

[CountMike]

You could run AOD at same time in logging and immediate mode, maybe you can catch it there.
The thing about those jumps, during such heavy loads, program itself may not be able to function all the time and just gets stuck for a moment until it can get resources back.
Here's another nice test program: http://www.ocbase.com/

 

[Hatekraze]

Ok, so you think it sounds like its actually making the jump up? I still cant seem to find anyone else having the same extreme jumps like that. Plus, i spent pretty much the entire day trying to fine tune to be stable at just 4.6ghz without going over 1.475v, without any real luck lol So there seems to be no way it could jump that high without crashing.

 

[CountMike]

Dunno, I had lockups with less jumps than that, it's usually the program that dies first. One of the reasons and maybe biggest one they changed to Vishera from Bulldozer is disappointing power management. The last one, 8370 is as good as it can get in that field. FX 8370E has best low to high power spread.

 

[madmatt30]

Did you just keep the old windows install after fitting the new board ??
If do a lot of those speed fan/core temp monitoring issues will be down to that IMO.

Board is capable if running 1866mhz ram with 4 sticks , you may have to loosen timings manually.
1600mhz will run with native SPD settings though.

Easytune works fine on win10 BTW on my system.

 

[Hatekraze]

I kept the windows when I switched the board. I had just reset windows before hand and didn't think it made a difference when switching motherboards. Would it have better readings if I reset it again? I don't have a disk or anything to do a complete setup obviously, since this was an upgrade from win 7.

And I tried 1600mhz manually, as well as the 1866 xmp with correct timings on stock clocks otherwise and both times it wouldn't boot up. I didn't change voltages, the previous board didn't need any voltage changes to do so either. Memory is rated at 1866 9-10-9-27 at 1.5v and that's the xmp setting. I've been fine so far messing with fsb when overclocking, and therfore dipped over 1600mhz before without issue. Seems the best fsb I've had luck with is like 230 which puts me a bit under 1600mhz. My plan was to try and learn how to tighten up timings after I'm done with core clocks. Idk why but this board won't let me change the ram frequency. Hell, I can't even change any voltages other than vcore, cpu nb, and ram. There's not even any options to do so. So messing with fsb too much will screw me I think. Not overly impressed with it still. But at least I can overclock higher than the old 4 phase board. I should have spent a little more to try and get an identical AsRock board with better vrms. I'm probably just being picky lol

I'll try again with easy tune, but when I looked before everywhere I looked said it wouldn't work. I don't mind the mobo doing it now, but at the highest setting it still isn't set high enough to max out the fans even nearing dangerous voltage and heat. It also seems to go off of the core temps which means it tends to throttle up and down more than if I set it to the socket temp myself. But it does the job I guess