Black screen running Aida64

[superdonair]

Hey guys, I started my overclocking adventure this morning and ive hit a black screen. All the way to 4.2 is stable but once I go to 4.3 at 1.3v i get a black screen about 8 minutes into Aida64. I've tried adding slightly more voltage but i still get a black screen. everything has power except the screen is black. i tried pressing ctrl+alt+del and i am able to open task manager and get back into desktop? Pretty odd. I updated the graphic drivers too but that didnt work. Setting i've changed: LLC is 5, cpu cache ratio: 255.50 (max limit) vcore: 1.35 and i changed ram to the xmp setting.

I also ran Memtest86 and have zero errors. I have not flashed my bios since Ive read you don't need to unless its not working right. Would flashing fix this issue? Or sadly is 4.2 as far as I can take it?? Please help thanks

z170a, i5-6600k, asus gtx 1070, h100iv2, seasonic 750w gold, 16gb ddr4 hyperx fury 2133

 

[MeanMachine41]

What temperatures are you getting in AIDA64 during test and before the black screen.?
It's best to have HWMonitor side by side with AIDA64. Take screen shots during test and before the black screen and link the screen shots here for analysis. You can use your favorite file host or use IMIGUR to provide the BBurl as a link.
Doing it this way will also provide info about your PSU voltages on all rails and GPU.
With your cooling and CPU, you should be able to get more, unless the CPU is sub par.

 

[superdonair]

What temperatures are you getting in AIDA64 during test and before the black screen.?
It's best to have HWMonitor side by side with AIDA64. Take screen shots during test and before the black screen and link the screen shots here for analysis. You can use your favorite file host or use IMIGUR to provide the BBurl as a link.
Doing it this way will also provide info about your PSU voltages on all rails and GPU.
With your cooling and CPU, you should be able to get more, unless the CPU is sub par.

Thank for the quick reply. Here is the screenshot: http://imgur.com/a/NbP8D

I unchecked Stress GPU to narrow things down, then I ran Aida64 for 15 minutes with NO black screen. And i ran a few passes in Cinebench. Temps seemed pretty good. This was 4.3@1.300 volts. I still need to run longer tests to confirm the stability is solid. So do you think its safe to say the GPU was the cause? Also should I run Stress GPU in Aida64 just on its own to see what happens or is there a better way to check if my card is faulty? Thx

 

[MeanMachine41]

Yes you can run the test on GPU separately in AIDA64 for an extended period.
Another test is to run Furmark, I use EVGA Precision x16 or the ultimate test is ROG "Realbench" which is a realworld tester and if you can pass that test your 24/7 stable.

Your IMIGUR link shows your temperatures are fine and the PSU is doing its job although a little overvoltage on the 12V rail under load.
So far as your CPU is concerned, you still have some Overhead to play with as temps are excellent at 4.3GHz. Raise Core voltage to 1.3875V and try 4.5GHz. Keep an eye on your temps and do not go above 1.4V on the Core.

Which particular MB are you using.? Some Bios Tweaks may be necessary for stability if the GPU tests good.

 

[superdonair]

Yes you can run the test on GPU separately in AIDA64 for an extended period.
Another test is to run Furmark, I use EVGA Precision x16 or the ultimate test is ROG "Realbench" which is a realworld tester and if you can pass that test your 24/7 stable.

Your IMIGUR link shows your temperatures are fine and the PSU is doing its job although a little overvoltage on the 12V rail under load.
So far as your CPU is concerned, you still have some Overhead to play with as temps are excellent at 4.3GHz. Raise Core voltage to 1.3875V and try 4.5GHz. Keep an eye on your temps and do not go above 1.4V on the Core.

Which particular MB are you using.? Some Bios Tweaks may be necessary for stability if the GPU tests good.

So I ran only the GPU test in Aida64 for roughly 15min, no issues. Then I ran a Furmark test with a score of 3770 at 58C no problems. After that I ran another Aida test with everything checked off for 20+ min, no black screen or issues. This was at 4.5GHz @ 1.3875V (mobo rounded to 1.390V) temps did not go over 61C. After the test I tried scrolling on this webpage and it was laggy but it went away after. Maybe the comp just had to cool down or something, just thought I'd mention anything unusual.

I am using an Asus Z-170A mobo. The settings I've change other than core ratio and volts are:
dram XMP
cpu/core cache current limit: 255.50 max setting
and LLC is on level 5 (options from 1-10)
everything else is set to Auto.
What I've read about LLC is that it levels out your volts and helps with vdroop? but can also spike your volts sometimes. Maybe that's whats causing that slight overvoltage.
What other tweaks do you recommend?
Also how long should I run Realbench and Aida64 for for the final stability tests?

 

[MeanMachine41]

Well I think you have a handle on your Bios. Yes,Load Line calibration set too high (7-8 should be OK) will increase voltage however its necessary at the higher overclocks. Also set CPU Current Capability to 120%.

I find its best to disable Cstates as this can interfere with a decent OC and it's just a power saving feature unless your a power conscious greenie then turn it back on later when your system is stable.

Now try for 4.6GHz at current settings and test again keeping an eye on your temperatures.
There are a few other Tweaks that can be done if it's unstable regarding System agent voltage and positive offsets. See how you go and if stable then save a profile in Bios.

If AIDA64 is stable after 30mins and temps in HWMonitor are less than 70C then run Realbench for 3 passes, it will approve the hash tag on each pass. If OK then try the benchmark test. You can then check your scores with others with similar builds on the Leader board.

 

[superdonair]

OK so I disabled Cstates, changed LLC to level 7 (lowest setting) and changed CPU CC to 120%
Tried 4.6GHz but got a black screen at 10 min in. Brought Current Capability back to auto then went onto my desktop but then half my wallpaper turned black. Turned back down to 4.5 and crashed in 10 minutes in again. Maybe level 7 is too low? Or I have to +/- the Current percentage. 4.5GHz@1.3875V seemed stable before these 2 tweaks. I wonder if this may be the threshold. Let me know what you think

This screenshot is 5 seconds after the black screen once I got back in at 4.6ghz: http://imgur.com/a/p5txM
And this one is at 4.5ghz: http://imgur.com/a/xztG5

I might try 4.6 again at 1.390V with cstates off and previous settings and see what happens.

 

[MeanMachine41]

Well your temperatures are OK in both tests and voltages are acceptable.
It's best to not stress all components together as it's more difficult to determine which could be letting you down. I usually do the CPU, FPU and Cache together for the frequency OC, then test memory, disks and GPU separately. You should also have the GPU at default till your CPU OC is stable. Then work on the GPU OC as this may be the problem.
I think you are at your MAX OC at 4.6Ghz and frankly its not worth chasing the extra 100Mhz as it hardly makes a difference unless your chasing benchmark scores.
Once your CPU is stable then bring your Core voltage down in small steps till its unstable then bring it back up till stable and that's your MAX OC.

 

[superdonair]

Oh ok, yeah that makes sense not to test all of the components together.
I haven't messed with the GPU settings at all but I will make sure they're all on default. I was just watching the temperatures with those programs. Still need to read up on GPU overclocking.
Yup I'm happy with 4.6Ghz and you're right its not worth getting another 100Mhz, I'd never notice that tbh.
It was kind of addicting pushing it a little more each time, I think I'm hooked now lol.

One last thing though, for an everyday 24/7 OC, what's the best thing to do? Should I raise the Ghz as far as I can without changing the voltage, or maybe run 4.1/4.2 and a little more volts set on Adaptive mode? or Offset? They seem kind of similar.

 

[MeanMachine41]

It's best to save profiles in Bios at various frequencies say one at 4.6GHz and one at 4.2GHz. I run the higher frequency in Winter when ambient temps are low then change my profile for Summer when temperatures are high. It depends on your climate. The rest is experimental and system dependent as all chips are different depending on your Silicone lottery draw.

With respect to Adaptive and Offsets.
Offset and Adaptive work differently. Offset is applied across the entire frequency range. Adaptive mode increases CPU voltage above stock VID and only when maximum default multiplier is exceeded.

Adaptive mode settings have the option of using Offset mode in conjunction with Adaptive but its just an additional option and not part of Adaptive mode. In some cases using Offset along with Adaptive is advisable.

Offset adds on top of normal voltage used by the CPU so even on idle your CPU will be pulling more voltage than needed.
Adaptive affects your turbo clock speed when turbo kicks in. With the added voltage you can also add an offset for non turbo clocks.
I find its best not to use Adaptive mode when stress testing.

 

[superdonair]

Yup made those 2 profiles just now and I'm just playing with the voltages on the lower setting. It's pretty much trial and error until you get it stable. It's interesting you never know what you'll get out of your chip. I think I have most of the basics down now but still have a lot to learn. It's not as bad as I thought it'd be as long as you do a little research first and go slowly. Thanks very much for your help and knowledge, I appreciate it a lot.

 

[MeanMachine41]

Your Welcome superdonair Yes it is addictive for sure. lol