Overclocking i7-6700k, some help needed!

[ssmana]

Hi i just put together a new system. Asus hero maximus viii, i7-6700k, 16gb DDR4 3200 mhz. I have it set at @4.6 ghz and my voltage is 1.28. Now the reason i need help is for a couple of things. One, is when i monitor my voltage with software like HW monitor, it does not stay at the voltage i set it too in my bios. It jumps up and down, some times going past 1.3 volts, this got me worried because i am afraid that it is going to high and is unsafe. The second reason i need help is because i can not get it to stay at 4.6 ghz, it stays at 4.0 and then when i run games it goes to 4.6 ghz, i would like it to stay at 4.6ghz solid. Does anyone know why this happens. i did not use XMP or auto, i used the manual setting. is there anything in the bios that i can change so it stays solid at @4.6 ghz. My prior computer was a i7-4790k and i was able to get it to stay @4.6 ghz, but i forgot what i done it was so long ago. Plus the bios is similar but not identical.

 

[JackNaylorPE]

1. The voltage change is normal. When AVX instructions (or other advanced CPU features are called upon) are present for example, Vcore will jump up 0.13 volts, possibly more. This is why it is recommended not to conduct CPU stress tests with synthetics as they can hemmer the CPU too hard and too long with these instructions. An application based stress test which uses them like they are used in normal use, creates a more realistic load scenario, and while toy still get the voltage bump, it's not a constant hammering. Use RoG Real Bench to apply a more realistic scenario.

2. Games don't always stress a CPU very heavily. Even a GPU stress test like Fumark can only load the CPU to 40%. Again, RoG Real Bench is your solution here.

3. No, it is not advisable to run constantly at 4.6 GHz. This is like driving your car a6b 6500 rpm even when going 20 mph in a school zone. You get no advantage to it and you stress your system unnecessarily resulting in more wear and tear. If you still want to do this, do a web search on turning off Intel Speed Step and C States.

4.. It is not recommended to just jump from stock (.4.2 turbo) to 4.6 Ghz. Best to make small jumps so yu can see how settings and temps / voltages trend as you increase multiplier.

Might want to d/l as you continue in your endeavor. HWiNFO is the be all and end all of monitoring programs and Real Bench will allow you to stress the CPU and test the stability in a multitasking environment

https://www.hwinfo.com/
http://dlcdnmkt.asus.com/rog/RealBench_v2.44.zip

 

[ssmana]

1. The voltage change is normal. When AVX instructions (or other advanced CPU features are called upon) are present for example, Vcore will jump up 0.13 volts, possibly more. This is why it is recommended not to conduct CPU stress tests with synthetics as they can hemmer the CPU too hard and too long with these instructions. An application based stress test which uses them like they are used in normal use, creates a more realistic load scenario, and while toy still get the voltage bump, it's not a constant hammering. Use RoG Real Bench to apply a more realistic scenario.

2. Games don't always stress a CPU very heavily. Even a GPU stress test like Fumark can only load the CPU to 40%. Again, RoG Real Bench is your solution here.

3. No, it is not advisable to run constantly at 4.6 GHz. This is like driving your car a6b 6500 rpm even when going 20 mph in a school zone. You get no advantage to it and you stress your system unnecessarily resulting in more wear and tear. If you still want to do this, do a web search on turning off Intel Speed Step and C States.

4.. It is not recommended to just jump from stock (.4.2 turbo) to 4.6 Ghz. Best to make small jumps so yu can see how settings and temps / voltages trend as you increase multiplier.

Might want to d/l as you continue in your endeavor. HWiNFO is the be all and end all of monitoring programs and Real Bench will allow you to stress the CPU and test the stability in a multitasking environment

https://www.hwinfo.com/
http://dlcdnmkt.asus.com/rog/RealBench_v2.44.zip

I have it set @4.6 ghz with 1.28v, so from my understanding the fluctuation being monitored is safe even when it goes past 1.3 v and higher? Basically is it safe for me to keep the BIOS setting like this?

 

[JackNaylorPE]

It seems fine but I'd really hate to say 100% at this point. Can't compare when using different monitoring programs and different stress tests. Again, you can be synthetic stable and then fail using applications, especially when multitasking. If your test doesn't include AVX, and I don't know that the specific AIDA test you used does, then you are not seeing the max voltage.

I run at 1.38 and it will climb to 1.41 when AVX is not present ... when it is, I can see 1.49 - 1.50 but it lasts like milliseconds. I don't really feel comfy w/ voltage until I run Real Bench benchmark (3rd and 4th components of the benchmark include AVX)...I really don't feel comfy about mutitasking stability until I can get thru 2 hours of it as previous OCs that wee 24 synthetic stable, failed under RB. Not match value in knowing ya CPU can run stable synthetics unless goal is to get name on web site leader boards. Goal generally is to run stable and have "happy temps" using applications. So RB tells me that I am stable and I can generally get a higher OC than when using synthetics as the latter results in higher temps.

 

[ssmana]

It seems fine but I'd really hate to say 100% at this point. Can't compare when using different monitoring programs and different stress tests. Again, you can be synthetic stable and then fail using applications, especially when multitasking. If your test doesn't include AVX, and I don't know that the specific AIDA test you used does, then you are not seeing the max voltage.

I run at 1.38 and it will climb to 1.41 when AVX is not present ... when it is, I can see 1.49 - 1.50 but it lasts like milliseconds. I don't really feel comfy w/ voltage until I run Real Bench benchmark (3rd and 4th components of the benchmark include AVX)...I really don't feel comfy about mutitasking stability until I can get thru 2 hours of it as previous OCs that wee 24 synthetic stable, failed under RB. Not match value in knowing ya CPU can run stable synthetics unless goal is to get name on web site leader boards. Goal generally is to run stable and have "happy temps" using applications. So RB tells me that I am stable and I can generally get a higher OC than when using synthetics as the latter results in higher temps.

So i used the programs you suggested. I ran the test @4.6ghz and used 16gb Ram. It ran for 15 minutes and it passed. I will set it to run longer so i am sure that it is stable. Also the voltage VID is very low, i set it to 1.28 in the BIOS but using HWmonitor it runs averagely lower than that. Which makes me very happy because i would rather have a low voltage VID, min average max, than a max stable processor running higher than 4.6ghz. I am not sure how high i can get this chip, but i really do not want to. I want low temps and low voltage with a moderate boost. I am going to see if i can get it stable using 1.275 v. Thanks for the help, thank you for the two programs. I was in the deep blue ocean on this. I have done overlocks before but not the same chip. It is part of my nature to consider that there are always things i do not know. that is why i had to start this thread.