i7-6700k + Maximus VIII Hero

[Kenrou]

No matter what combinations i try the cpu will not be stable at anything higher then 4.7ghz regardless of voltage or temps (last post apparently was a fluke because i cant reproduce it). Same error every time now, CLOCK_WATCHDOG_TIMEOUT. New BIOS and USB firmware didn't help, only mouse and keyb, 1 or 2 sticks of ram, lower speeds and timings, even went so far as taking out all but one SSD. 4.7ghz 1.35v ~72c (can possibly lower voltage, haven't tried) with standard fan setting in BIOS.

Went all the way to 1.46v for 4.8ghz, no go... Could it the a bad cpu ? should i RMA ?





3DMark Fire Strike / Extreme / Ultra

 

[ATMINSIDE]

So you're considering RMA because you can't stabilize higher than 4.7GHz????
You literally have zero recourse to ask for a replacement CPU.

 

[Kenrou]

So you're considering RMA because you can't stabilize higher than 4.7GHz????
You literally have zero recourse to ask for a replacement CPU.

I have no idea if this is normal or not, its my 1st Intel since P3, that's why i asked i'm at 1.35v ~72c and cant go any higher, seems a bit... meh ? I'm well below both the safety margin for voltage and temps

Plus I'm not really used to getting errors with strange and scary names while overclocking, my experience with AMD is lack of voltage = freeze or core drop, not 3 pages of guys discussing wether the CPU or RAM or MB is fried...

 

[ATMINSIDE]

I have no idea if this is normal or not, its my 1st Intel since P3, that's why i asked i'm at 1.35v ~72c and cant go any higher, seems a bit... meh ? I'm well below both the safety margin for voltage and temps

Welcome to the silicon lottery.

The only time you have a case for an RMA is when the CPU does not function as intended at stock speeds.

 

[Kenrou]

Welcome to the silicon lottery.

The only time you have a case for an RMA is when the CPU does not function as intended at stock speeds.

I understand, just can't help feeling that it's a massive waste that it can't go higher

Can't (and won't) complain more though, speed improvement over my old 8370 at the same 4.7ghz is insane in every aspect

 

[wingman99]

I understand, just can't help feeling that it's a massive waste that it can't go higher

Can't (and won't) complain more though, speed improvement over my old 8370 at the same 4.7ghz is insane in every aspect

AMD and Intel don't run at the same clock speed AMD runs at a higher clock speed because they have a deeper pipe stage in the Instruction pipelining.

Some review sites can't get skylake stable above 4.6GHz.

 

[Kenrou]

So it's a sort of hardware cap or just bad binning ? If I can't get 4.8ghz stable at 1.425v ~85c there's no point in a delid/better cooling to try going higher right ?

 

[ATMINSIDE]

So it's a sort of hardware cap or just bad binning ? If I can't get 4.8ghz stable at 1.425v ~85c there's no point in a delid/better cooling to try going higher right ?

It isn't a hardware cap or bad binning, it is simply the silicon lottery. As I said, welcome to it.

No, delidding won't help.

 

[Kenrou]

No, delidding won't help.

Right, last question what about the cache or ring clock ? Why only put it a couple 100's below CPU speed, why not match or go higher, does it give temp or stability issues or simply not worth it ?

 

[ATMINSIDE]

It'll cause instability.

 

[wingman99]

So it's a sort of hardware cap or just bad binning ? If I can't get 4.8ghz stable at 1.425v ~85c there's no point in a delid/better cooling to try going higher right ?

I don't like the term lottery that is about winning and loosing of 1-3 people out of millions.

When they make CPU's they try to make them all the same however that is never been possible, like all things in life there is a variance with the finished product.

This is a picture of a try gate transistors of a Intel 22nm CPU you have to ask your self do they look perfect or functional variance. How small is 22nm

22nm http://www.intel.com/content/dam/ww...f/foundry/mark-bohr-2014-idf-presentation.pdf

 

[Kenrou]

Thank you all for your support, ill stick to 4.7ghz for now and be happy and proud considering this thread closed

 

[turbobooster196]

i dont have a 6700k, but his little brother, the 6600k.
Today i tested it a bit again, these are my settings on a MSI Gaming Pro motherbord.

multi at 4.7ghz
vcore at 1.330v in bios
Cache at 4100
5 loops of X264 stabilytie test vcore goes to 1.344v
max core temp 60 degrees, cooler hyper 212 evo, not bad at all i think

 

[Johan45]

Watchdog is likely Memory/cache. I would concentrate there myself. Seen a lot of those on the weekend playing with the X99. Realbench is encoding driven IIRC so a lot of memory usage. Have you modified the ram timings/voltage at all?

 

[Kenrou]

Using XMP 3200mhz 14-14-14-34 stock (sorry only now saw your question). You know one good thing about being pig-headed/stubborn ? sometimes you get results finally managed to stabilize 4.8ghz 1.36v, odd thing had to bump ram voltage, otherwise sticks to 4.7ghz.

Wanted to try my hand at overclocking the ram as well, but haven't the foggiest to where to start or which programs to stress with.



1036 4.7ghz uncore 4.7ghz - 1047 4.8ghz - uncore stock 4.1ghz

 

[Johan45]

Looks good Kenrou. Memtest or HCI in windows works for testing memory. Or even stick with the IBT /realbench. etc.. Be prepared for a lot of blue screens /no boot/ bios and OS corruption when dickin with the memory.
Link for HCI http://hcidesign.com/memtest/

 

[EarthDog]

So... you raised your ram voltage over its stock settings to stabilize the CPU?? Weird. That is like fixing your brakes because your AC isn't working.

Ok, that is being a bit dramatic, but, I am surprised that adjusting your ram stabilized your CPU as the ram voltage has nothing to do with the CPU.

But again, you raised that voltage over its stock settings? That sounds like bad ram to me. I don't have the Hero, but the Formula and the Extreme. With 3200MHz ram, I was XMP and go for 4.8GHz.

 

[Kenrou]

XMP sets it at 1.35v, maybe it was dipping below and making it hang ? I bumped it to 1.36v and it runs 4.8ghz fine now. This is what I get for picking new "untested" ram, had to use latest BIOS for it to work properly, maybe it didn't get rid of all the kinks ?

 

[EarthDog]

Maybe. Does the hero have any voltage read points? If so, test it with a MM and see. Otherwise, all you have is the BIOS.

 

[Kenrou]

HWInfo64 can read timings, voltage and temps from the RAM, so far doesn't seem to drop. Even at 1.4v haven't seen it go above 30c, and it runs SuperPi 32m in ~6m. I'll run more tests during afternoon