so with somehelp of people over at LTT forum i got a good idea of what i needed to do and i got to 4.7ghz at 1.480v. now that is way over what everyone says is the community safe point although intel specs at 1.5v? getting to where i am now was the last of any help i really got and has left me in a spot of trouble. under full load / utilization i hit 90c on cores 1 and 2. 85 - 87 on cores 3 and 4. i am using a someone passive fan curve that only really goes nuts at 80c. which has kept me from reaching 90c again.
stability tests i used are. intelburn test: -standard 10x -high 10x -very high 5x Passed.
realbench infinite loop benchmarking resulting in a pass. so did gaming tests and gpu benching.
im using a corsair h100hiGTX with 2x 120mm SP fans from corsair. and thats how im reaching my temps. below are some images of temps and scores etc. my main reason for help is should i be safe at this temp as my stability is on a very THIN line of 5+ / 5-. Or Should i just drop down to 4.6ghz?
Don't use burn test to test temps its almost as bad as prime95 in the amount of heat it generates at-least in my experience using it. Use OCCT or aida64. But that voltage is quite high for that clock and is going to be contributing to the high temp. I don't like going past 1.45v with skylake 1.4v is what i would say for 24/7. Try the following if you have not adjusted vccio and SA yet: VCCIO 1.15v system agent 1.2v Vcore 1.38v LLC around 5 or 6 but im not sure what the options are like on msi boards.
I got pretty high temps on my 6700k when overclocking, [email protected] and was getting 70s-80s in games, with a silver arrow and also when I built a custom loop with 2x180mm radiator space. In order to get temps down I had to delid, and got 15-20C drop in temps. I would say that your voltage seems a bit high, but every chip is different, maybe consider backing down to 4.6 or 4.5 as you aren't likely to see a performance difference.
That hasn't been their spec since fucking 45nm back it off or your chip will be extremely degraded in a month. Don't pass 1.35v at most, 1.3v it's be safe. I've degraded my 4790K in just 4 months with 1.38 volts so heed my warning.
You really got your priorities wrong, you need to limit how far you push, if you can't confirm it stable at 1.35v then you need to stop trying. Aim for 1.3 and I'd you need an extra little within that 1.35v buffer then do it but 1.4v is still too high. No amount of overclocking epeen is worth having to get a new chip in a year.
That's true for Haswell, however Skylake is an entirely different beast. It is fine going up to 1.4v for 24/7 provided of course you have appropriate cooling. With That said you are right if you cant get it stable you are better off doping 100mhz instead of jumping 0.5v.
If you do want to try for 4.7 and keep it below 1.4v you are going to have to put in manual volts for IO and SA as well as tweak the load line calibration. If you do that i would expect you to be able to get it stable. Sometimes you need to up the Vdimm a little bit but i don't think you will need to do that just keep it in mind as a last resort if you cant get it stable.
Max Voltage i would go to for SA and IO is 1.25 and 1.2 respectively.
Mainly because Haswell had part of the voltage regulation on the chip and Skylake does not. Witch is why Haswell would start to get very hot when you go past 1.35v, you would be inputting 1.8-2v into vccin (the on chip voltage regulation). With Skylake you don't have this problem.
when i was exploring my bios and just looking around i had left my Voltage at AUTO and had it on 4.7 it put the voltage at 1.498 and by god it was stable? hot as the sun but stable.... i think my chip is just not one of those "holy grails"
im fine with 4.6 at 1.35 unless my new cooling setup somehow keeps all my gear cooler so i can bump it up.