I have a Gigabyte Aorus Ultra Z390 which is like their 2nd from their top of the line, but I am finding I cannot overclock for crap. I had a 9600k and it took something like 1.37v just for 4.8 Ghz. So I bought a 9700k which was previously binned and tested to run 5.0 GHz at 1.295v and I am having trouble getting it stable at 4.9 GHz at 1.33v. So I am thinking this motherboard is just trash. Is this a common thing? I always though OC potential is basically almost solely respective of the CPU. Also I'd expect a lot more from this motherboard considering how expensive it is... Unfortunately every motherboard company out there is total trash when it comes to tech support so I am not going to even bother calling.
-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
Crappy motherboard perventing better OCs?
- Thread starter SPL Tech
- Start date
For the most part, the CPU is the thing when ambient overclocking.
That said, if we look up your board you'll find that it is 12-phase (6 actual - 12 doubled) and 50A FETs....not exactly the most robust, but I'm still surprised it is having trouble as those specifications, ballpark, are similar to others.
EDIT: Watch -
EDIT2: I just listened to it on the pot, lol (yay BT cans!), seems like it would easily be fine... especially with the CPU you are using. Im guessing the difference is in software reporting the core or something.
That said, if we look up your board you'll find that it is 12-phase (6 actual - 12 doubled) and 50A FETs....not exactly the most robust, but I'm still surprised it is having trouble as those specifications, ballpark, are similar to others.
EDIT: Watch -
EDIT2: I just listened to it on the pot, lol (yay BT cans!), seems like it would easily be fine... especially with the CPU you are using. Im guessing the difference is in software reporting the core or something.
Last edited:
- Thread Starter
- #3
Well I am finding the voltage in the BIOS is even higher than it is in CPU-Z. Like 1.32v in CPU-Z requires I set it to close to 1.4v in the BIOS. I also find the offset voltage control does not work for crap. It fluctuates from like 1.2v to as high as 1.6v in CPU-Z when I attempt to use it (under load). It's all over the place and it requires that I have to use massively high voltages to remain stable if I try to use offset (so I dont, I just use fixed voltage). I feel like MSI's motherboards had much easier to use BIOSes with adaptive voltage setting that worked well.
The temperatures match the voltage reported in CPU-Z. The CPU is getting pretty warm under Prime 95 even with a 280mm AIO cooler. I feel like CPU-Z should be accurate. It typically is. I dont know, but with both CPUs I am finding I am not able to get the clocks that other people are. I am needing a ton more voltage to remain stable at the same clocks that many other people achieve at much lower voltages and now I have tested two different processors which have the same issue with this mobo.
Last edited:
So... you are setting 1.4V in the bios to get 1.32V under load in windows, is that correct? Why not use LLC instead? IIRC, the turbo setting keeps things stable..above that it may overshoot. See if that helps.
Are you on the latest bios... or at least past F7? Have you tried another?
Have you confirmed voltage against Hwmonitor or even coretemp just to see? It seems right, but if you dont have voltage read points, it doesn't hurt to verify.
Offset adds to or takes away from the VID for a given multiplier/auto settings. If it goes too high for what was found to be stable, use a negative offset. In other words, if 5 GHz @ 1.3V is stable set manually, and on auto voltage goes to 1.35V a negative offset of 50mv should bring that voltage back to 1.3V...you'll have to play with it a bit.
Are you on the latest bios... or at least past F7? Have you tried another?
Have you confirmed voltage against Hwmonitor or even coretemp just to see? It seems right, but if you dont have voltage read points, it doesn't hurt to verify.
Offset adds to or takes away from the VID for a given multiplier/auto settings. If it goes too high for what was found to be stable, use a negative offset. In other words, if 5 GHz @ 1.3V is stable set manually, and on auto voltage goes to 1.35V a negative offset of 50mv should bring that voltage back to 1.3V...you'll have to play with it a bit.
Last edited:
- Thread Starter
- #5
So... you are setting 1.4V in the bios to get 1.32V under load in windows, is that correct? Why not use LLC instead? IIRC, the turbo setting keeps things stable..above that it may overshoot. See if that helps.
Are you on the latest bios... or at least past F7? Have you tried another?
Have you confirmed voltage against Hwmonitor or even coretemp just to see? It seems right, but if you dont have voltage read points, it doesn't hurt to verify.
Offset adds to or takes away from the VID for a given multiplier/auto settings. If it goes too high for what was found to be stable, use a negative offset. In other words, if 5 GHz @ 1.3V is stable set manually, and on auto voltage goes to 1.35V a negative offset of 50mv should bring that voltage back to 1.3V...you'll have to play with it a bit.
Yes, all correct. I am using LLC. I have the LLC set to the highest setting which is about a boost of +0.01v under load. I know what offset does, I am just saying it doesent work for crap on this motherboard. It's really hard to get a stable overclock using offset on this mobo without setting the voltage absurdly high. Even on fixed voltage, the vcore is still way higher than what it should be for a given overclock. I am on the latest BIOS rev. F10. The vcore reads the same in all software applications. Coming from MSI, Gigabyte's mobos are much harder to overclock with. MSI has adaptive offset voltage which is super easy to use and very stable. The downside is MSI's audio chipset completely sucks on the Z390 series and I never got sound to work on any of the mobos I tried so I have to move onto someone else.
Not sure what to tell you then.
The power delivery doesn't appear to be lacking. I dont recall trouble with it for a review.
Edit: you added this like 30 mins after I replied..
try a different motherboard is the only way...
Last edited:
- Thread Starter
- #7
Not sure what to tell you then.
The power delivery doesn't appear to be lacking. I domt recall trouble with it for a review.
Where did you get the info on the board's VRMs?
Google.
Typically any quality review mentions how the power bits are setup what amp the FETs are, etc. I know buildzoid does video on vrms...got lucky he did this board (link earlier in case you missed it). There are vrm databases here and there as well (well done ones for amd).
- Thread Starter
- #9
So I've been doing a bit more playing around. I am finding Cinebench is very easy to pass for some reason. I can pass Cinebench at 5.0 Ghz with AVX offset of 0 with 1.32v core. I can even pass Cinebench at 5.2 GHz. with 1.4v and AVX offset of 0. However, here is the thing. At 5.0 GHz. and 1.32v I cannot pass any of the other benches. Prime95, Intel Burn In Test, RealBench all crash quite quickly. So I up the voltage to 1.36 and that helps a bit, but I still end up crashing within a few min usually. So to pass the other benchmakrs at 5.0 GHz I would probably need 1.4v or so.
Also the temps are completely unsustainable. I am getting throttle limiting at 100C when I run these benches which is really strange because I have a 280mm AIO set to max pump speed with Thermal Grizzly Kryonaut TIM (the best performing non-conductive TIM currently made). I tried reseating the CPU cooler and that dident help. I think the chip just is genuinely getting too hot to cool. CoreTemp is telling me power draw of CPU is like 215w during the tests which is crazy high. In CineBench it's like 165W which probably accounts for the lower temps and easier stability.
So it sill doesent add up. Both CPUs I have tried require way more voltage and produce way more heat than other people claim they are getting with these things. Sure, it will pass Cinebench all day at a very low voltage, but it wont pass any of my other tests. So I dont know if just everyone out there is using CineBench for their stress testing or something, but from my seat these chips are completely incapable of the 5.2 GHz+ some people claim at under 1.4v. No freaking way. I cant even come close to that. Even if I bumped the voltage way up, high enough to be stable at 5.1 or 5.2 Ghz, the heat they produce cannot be cooled by any non-extreme cooler. You'd need LN2. I am hitting the limits of water cooling at only like 5.0 Ghz and 1.35v which is getting me temps of 90 - 100C in benchmarks.
Also the temps are completely unsustainable. I am getting throttle limiting at 100C when I run these benches which is really strange because I have a 280mm AIO set to max pump speed with Thermal Grizzly Kryonaut TIM (the best performing non-conductive TIM currently made). I tried reseating the CPU cooler and that dident help. I think the chip just is genuinely getting too hot to cool. CoreTemp is telling me power draw of CPU is like 215w during the tests which is crazy high. In CineBench it's like 165W which probably accounts for the lower temps and easier stability.
So it sill doesent add up. Both CPUs I have tried require way more voltage and produce way more heat than other people claim they are getting with these things. Sure, it will pass Cinebench all day at a very low voltage, but it wont pass any of my other tests. So I dont know if just everyone out there is using CineBench for their stress testing or something, but from my seat these chips are completely incapable of the 5.2 GHz+ some people claim at under 1.4v. No freaking way. I cant even come close to that. Even if I bumped the voltage way up, high enough to be stable at 5.1 or 5.2 Ghz, the heat they produce cannot be cooled by any non-extreme cooler. You'd need LN2. I am hitting the limits of water cooling at only like 5.0 Ghz and 1.35v which is getting me temps of 90 - 100C in benchmarks.
Last edited:
1. Makes sense... Cinebench takes seconds to run and isn't quite as stressful in the first place versus a stress test like P95 with AVX.
2. These temps aren't strange at all. I'd call it par for the course for your cooling and voltage.
3. Power draw seems about right as well. Welcome to modern Intel!
4. So, you're saying you need more voltage than others but don't know what application they used to give you these values? You can't say it's the board if you don't know that critically important piece of information!
2. These temps aren't strange at all. I'd call it par for the course for your cooling and voltage.
3. Power draw seems about right as well. Welcome to modern Intel!
4. So, you're saying you need more voltage than others but don't know what application they used to give you these values? You can't say it's the board if you don't know that critically important piece of information!
Last edited:
- Thread Starter
- #11
Of course I do-- CPU-Z, the same that everyone uses. All apps will provide the same info anyway as there is only one vcore sensor. The mobo's firmware determines the CPU voltage, not the software program installed on the OS you're using to view it.
For #4 I meant the STRESS TEST...you bought a CPU and said it does x.x Ghz @ y.yy volts.............under what stress testing application did they get these values? Hopefully that is more clear.
Hwmonitor and Hwinfo are also popular. CPUz and other applications can give different results even if it is on the same sensor. You've never had CPUz read erroneously and other software read right? It happens more than you think.
I know the mobo's FW/CPU determines the voltage... why did you say that? LOL
Hwmonitor and Hwinfo are also popular. CPUz and other applications can give different results even if it is on the same sensor. You've never had CPUz read erroneously and other software read right? It happens more than you think.
I know the mobo's FW/CPU determines the voltage... why did you say that? LOL
Last edited:
Similar threads
