profile back to stock

I have tried allsorts can anyone tell me how to delete old dis used profiles from my z170 mobo profile list , I want to start afresh being first time made a few mistakes

As far as I know, you can only over-write the existing ones with new profiles. You could copy the same profile to all currently occupied slots. If that is too messy for you, the only other thing to do is re-load/flash the BIOS. This will wipe all presets. You can save your current "good" profile to a USB drive beforehand and then load the profile into slot 1 after flashing the BIOS. However, be aware you can't load profiles from a BIOS version different than the one that created the profile. So, when it comes time to update to the next version, you will not be able to take your profile along with you. Make sure you make a copy in some other way until you have it all memorized.

thanks for that , as I am not used to this I get a bit frustrated, I tried a couple first profiled them then I did better on temps and voltage, so I wanted to keep it as neat as possible , when you are happy with your profile do you just say LOAD then play your game and how can you see its working under profile

Loading the profile in the tools menu will apply those BIOS settings. However, it will not become permanent until to use the Exit --> Save & Exit to leave the BIOS or Discard and Exit to abandon your changes. You can monitor the frequency and voltages with various software programs. Some of the Asus specific features can also be seen in the AI Suite control window. Be careful of making changes to BIOS settings from within the AI Suite application on the desktop. There is a somewhat intricate and inconsistent relationship between the two. Sometimes settings are over-written by AI suite. Sometimes they do nothing at all. Sometimes you will crash 3 seconds later. It is better to use the BIOS for changing any of those settings.

thanks for that got it at last. I understand a bit better now I load a profile in profiles say profile 1 when I want to use it what do I do and how can I tell if its working at say 4.5 and not stock

Once you Save & Exit, the system should re-boot. Those BIOS settings are now in effect. You will not be able to "see" into the BIOS screens from the desktop, but you can see some of the changes to frequency and voltage through any of the common monitoring programs. To see more subtle things like the effect of C-states on core frequency, you need a more detailed monitoring program like AIDA64. If you have already installed AI Suite, it does provide some of this information, but monitoring alone may not be a good enough reason to install it if you have not.

c I have done a overclock trying to get just a 4.6 I set my xmp to 1.270 volts I have trial version of aida64 and cpu-z all temps stayed ok but look at the voltage which would you use as correct aida 1.264 to 2.80 cpu-z 1.320 to as high as1.380 I wish to ask another question from you tomorrow but ill try to solve it first

If you are using Auto or adaptive voltage to set your Vcore, you will get some overshoot when using any synthetic stress program like AIDA, Prime, OCCT, etc. The Skylake chips in particular seem more prone to this than Haswell-E. As a general rule of thumb, it is recommended you use the fixed voltage when trying to find your stable overclock and benching, then convert it to an adaptive voltage for daily use when finished. There are also some Load Line Calibration settings (LLC) you may be able to raise to keep a tighter rein on the wandering voltage. Unfortunately, I am not using a Skylake chip so I can't tell you what is or is not normal for Vcore overshoot when using each of the testing programs. You might try a LLC level 7 and see if that tightens things up.

I set the voltage to manual, my only concern was the voltage difference between the the different monitors mind you the overshoot was not bad on the high one again many thanks

I suspect the two values are measuring two different things. Core voltage (VCore) - the value applied at one exact moment in time. CPU VID - which is the maximum allowed voltage overshoot. CPU VID is the ceiling. Some voltage overshoot is necessary and desirable in order to keep the system as stable as possible. The voltage regulation isn't so perfect that it can go from 0 to 1.265 to .775 to 1.265 and stop on a hundredth place value. If you search the terms CPU VID and Vdroop (voltage droop) you will surely find someone who has already described how these values work in a manner better than I can squeeze in this space. So what those two values tell you is your voltage was 1.28v at the moment you snapped the screen shot. During the testing, at some point (possibly only for milliseconds) your vcore reached as high 1.348v. That was the maximum recorded value - not the target and not the average Vcore. This overshoot is generally not important for normal use as it last fractions of a second in total and is not likely to affect your overall CPU temperatures. However, if you are very close to your thermal limits or if you run programs that utilize a 100% load for an extended duration, then you may want to keep an eye on this value. The Load Line Calibration can help with this at the expense of a higher average Vcore (less droop when not needed) which is likely to be most noticeable as a slightly higher VRM temperature.

 

In regards to your AI Suite questions, the software isn't perfect and it does make mistakes. In your situation it sounds like the program may have crashed before finalizing the BIOS changes and that left the BIOS in a state of disarray. That was likely a one time, single incident glitch. With it's default parameters, it will only stress test for 30 seconds per clock level. That might be enough to recognize a hopeless frequency and voltage setting, but those "on the borderline" values are hard to detect in 30 seconds. You can run it for longer, but frankly the AI suite is good for finding a general Frequency and voltage. Then you should fine tune it yourself, if you have any interest in this stuff at all. My experience with it on X99 was a bit shaky. Trying to tune it to specific voltages or frequencies often came back with errors. The one were it pretends to run stress tests as the frequency goes up indefinitely is a common one. No, you can't run 8.0GHz@1.40v on your chip. The percentage overclock value in AI Suite is just another way to show you "how much better off you are now". Obviously, 4.8 is 20% (0.8) more than the stock 4.0GHz.

thanks for that it will take me a bit of time to digest this but seems I have been looking at wrong readings or not understanding them by the way my LLC setting were set at level 5 if I go higher say to 7 would that not increase parameters,ie voltage temps. I just spent a bit of time digesting your feedback ,would you suggest I stick with aida 64 and not bother with realbench or is aida a softer touch my set voltage manual1.270 never went above 1.286 4hours approx. 4.6 63c tops

The LLC setting will affect the waviness of the voltage delivery. If you were to look at a line graph of Vcore, a level 1 setting would have considerable peaks and troughs compared to a very flat level 8/9 setting. I am not sure the max setting for your board. Z97 was 8. X99 is 9. The auto setting will try to dynamically change in accordance with load, but if you are running demanding programs or a high overclock, it's response may not be fast enough. Most people with high overclocks are OK 1-2 levels below the maximum.

 

The graph below is an old run with the LLC on Auto. You can see the how wavy the Vcore is at the beginning of the test. During that first minute, the OCCT program is applying highly variable loads as part of the warm-up routine and the voltage rises and falls. Starting at 1 minute, the serious testing begins. Just prior to each minute mark, the program initiates a new load routine. You can see the response in the Vcore at the lower LLC levels. After 2 events, the Auto routine recognizes there is a sustained high load and moves the LLC to the maximum 9 and there are no more significant overshoots. However, if you were looking at purely numerical data instead of a line graph, you might think it was running 1.263 Vcore the entire time, rather than the initial spikes.

 

http://www.mediafire.com/view/ivnztcx2kho07cm/2015-01-30-18h43-Voltage-VIN4.png

 

AIDA64 and Real Bench are not the most intensive programs out there, but I don't think the normal user needs any more than that. Real Bench adds some video elements the others do not. I love AIDA for it's monitoring capabilities and while I would normally pass on pay monitoring software, licensing that program is one the better decisions I have made. I initially used AI suite to monitor values and you likely will too. However, AIDA is a much better tool. I don't think you need to run these stress tests for hours unless your actual workloads are also near 100% utilization that last for hours. When you start getting into that length of time, you are putting a lot of thermal stress on your hardware that doesn't need to be there. 24 hours of Prime is ridiculous, unless you use the machine as 24/7 workstation or server. 4 x 15 minute stress tests is just as effective as 1 x 60 minutes. Besides, in all likelihood your crashes on unstable settings are likely to occur at sub-maximal loads and be related to rapid dynamic changes in voltage, rather than a consistent elevated value.