I5-6600K Vcore AsRockExtreme4

bmcat

Commendable
Aug 26, 2016
6
0
1,520
Hi Everyone, having a headache trying to figure this out, checked lots of forums but so far couldn't find any answer for my specific build, which is the following:

-AsRock Z170 Extreme 4
-Intel I5-6600K
-Zalman CNPS10X Optima
-Ballistix Tactical 16GB (4 GB x 4) DDR4 2666 (XMP Profile)
-Zotac Nvidia GTX 1060 Mini
-EVGA Supernova 750w B2 BRONZE
-Windows 10 Pro

basically I load up the OC profile from the UEFI to 4.2 OC, tried 4.4 OC, but since there was basically no improvement in gaming, I'm happy with the 4.2 OC, my chip doesnt go over 4.4 OC, anything more than that results in BSOD, a bit unhappy since I wanted to try 4.6 OC, but there is nothing I can do.

My issue is with the Vcore set at Auto in the UEFI, bumps this 4.2 OC to 1.392 Volts at full load and lowers it to 0.720V on standby. Also I have the LLC set to LV4, I tried it at LV1 and it gives me a crazy voltage of 1.43 on full load, which is a ridiculous amount of voltage for a 4.2OC.

my temperatures are steady around 65 Celcius on full load and 25 to 26 at standby.

Having this figures I set the Vcore at Manual in the UEFI, and setting it up at 1.3V and LLC at Lv4, I was getting a max voltage of 1.29 and temperatures around 55 Celsius on full load, which is less 10 degrees than having the voltage set at auto. My only problem with this, is that the CPU voltage doesn't reduce when in standby, which I believe wont help to keep my chip safe for at least 2 to 3 years.

Tried offset mode, that doesn't work at all, any figure I put, just results in BSOD.

Having this, I just need an opinion in basically what to do, because I cant figure it out what will be the safest option, hope you guys can help me.

Regards

OC_Figures.jpg
 
Solution
It's solved now, manage to get the settings right, my problem was loading the auto oc profile, I set everything to manual, cpu clock at 43, cache at 42, offset voltage at +45 and LLC at lv 2.

Results:
Max load voltage of 1.28
Max temp of 58 celcius at full load stability test in Aida 64
OC at 4.3 GHz instead of 4.2.

Happy with it now, finally

bmcat

Commendable
Aug 26, 2016
6
0
1,520
It's solved now, manage to get the settings right, my problem was loading the auto oc profile, I set everything to manual, cpu clock at 43, cache at 42, offset voltage at +45 and LLC at lv 2.

Results:
Max load voltage of 1.28
Max temp of 58 celcius at full load stability test in Aida 64
OC at 4.3 GHz instead of 4.2.

Happy with it now, finally
 
Solution

philipew

Reputable
Jul 26, 2016
168
3
4,715

The CPU Vcore Voltage increases at idle and decreases under load (to reduce the power because the current is higher under load). Why would you want yours to be any different? This is the normal way of these things.
 

bmcat

Commendable
Aug 26, 2016
6
0
1,520

I'm sorry, but your sentence doesn't make any sense, as far as I'm aware it's the other way around as you put it.

As for you second question, I'm overclocking, I want a stable temperature and a good CPU life at higher clocks than the factory ones, I want it different because I want to take control of it and don't want the bios to go wild and over-voltage my CPU and reach 80 Celsius degrees.
 

bmcat

Commendable
Aug 26, 2016
6
0
1,520

Call it what you want. Here are the definitions.

CPU clock speed, or clock rate, is measured in Hertz — generally in gigahertz, or GHz. A CPU's clock speed rate is a measure of how many clock cycles a CPU can perform per second. For example, a CPU with a clock rate of 1.8 GHz can perform 1,800,000,000 clock cycles per second.

The clock rate typically refers to the frequency at which a chip like a central processing unit (CPU), one core of a multi-core processor is running, and is used as an indicator of the processor's speed. It is measured in the SI unit hertz.
 

L0stChild

Reputable
Mar 8, 2016
635
1
5,165


then... what is cpu speed?
 

philipew

Reputable
Jul 26, 2016
168
3
4,715

As I wrote, the clock rate in Hertz (or cycles per second as you put it) is the frequency of the CPU, not its speed. This is why a CPU with a higher clock rate than another can be slower (i.e. have less speed) than one with a lower clock rate. You just have to benchmark this to prove it again and again and the reason is that the speed of a CPU is function of a lot more parameters than solely its clock rate e.g. its memory quantity and clock rate, its cache size, its architecture (like comparing Skylake with Ivy Bridge), etc... You still have a long way to go. Good luck with your overclocking ;-).
 

philipew

Reputable
Jul 26, 2016
168
3
4,715

Well you have to measure it. That's why all those benchmarks are around, and why there are all different. You can't compare 867 on Cinebench R15 (excellent) with 7600 (poor) on PassMark. For a car you just have to give a number of Km (or Miles) per hour and speed is clear. It is certainly not that simple for a CPU because there we refer to the speed "of execution", not of displacement which is much simpler (assuming the road is kept flat and dry, the wind is the same, etc...).

The speed of execution of a CPU depends on a number of factors such as the type load of the CPU (to name but one) - for example the speed of execution will vary considerably between arithmetic calculation (like Prime95 Blend test) vs. image rendering (like Cinebench R15 CPU test) because this relates to the speed of execution of its instructions in machine code, and not all the instructions are the same - some take fewer clock cycles than others. Multiple floating point calculations for example need a lot of "carry" micro code instructions which consume multiple cycles each and therefore can considerably slow down a CPU.

It's a bit like for measuring the amount of work a car can go through in one hour. Would you say that a (fast) Ferrari performs more work than a (slow) agricultural tractor? Yes, it's true on the freeway, but not in the field (or forest). This is where the car racer and the farmer would have very different (and divergent) opinions about the speed (i.e. the performance) of their respective vehicles.

Think about it...
 

philipew

Reputable
Jul 26, 2016
168
3
4,715

On the contrary, my sentence makes perfect sense and although I admit that it may be seen as being counter intuitive, it is absolutely correct (I propose to open the bet at USD1M cash - this is really easy money - I will soon be filthy rich then, can't wait) . Of course you have to understand this subject, have sufficient knowledge about what you are talking about, or alternatively know how to use the internet (things like Google anybody?) and find something that's plastered all over (nothing short of being common knowledge by now, especially after so many years).

Take a look here:
http://www.tweaktown.com/guides/7481/tweaktowns-ultimate-intel-skylake-overclocking-guide/index6.html
This is "TweakTown's Ultimate Intel Skylake Overclocking" and the graph under "Serial Voltage Identification" clearly shows how CONSISTENTLY LOWER the CPU Auto VCore (SVID) is under load, particularly at 4.8 GHz where it FALLS from 1.325 V all the way To A MUCH LOWER 1.284 V (!) and they mention under the graph "You can see the large drop from idle to load voltages." And these are not completely stupid people either (they do measure the voltage on the CPU pins directly with an electronic multimeter - no fancy CPU-Z action, smoke and mirrors).

Of course I have written about this more than once in other posts, but regardless, it's plastered all over the internet. Just look here:
https://pcpartpicker.com/forums/topic/130888-overclocking-i5-6600k Very first comment at the top: "That's completely normal. Its caused by a thing called Vdroop. When your CPU goes under load it causes the Vcore to drop."... too easy,

As I wrote previously, [Power P (W) = Voltage (V) x Current (A)] as in [72 W = 12 V x 6 A] and since the current is higher under load (this is not counter-intuitive), the motherboard (not the CPU) reduces the voltage under load (as a self-protection mechanism to avoid damaging the CPU) by an amount equal to Vdroop (V). Mine goes form 1.356 V (IDLE) to 1.344 V (UNDER LOAD) - clearly a drop of 12 mV (0.012 V).

If you "want a stable temperature and a good CPU life at higher clocks than the factory ones," you'd better start learning (applying yourself) about topics like "CPU Load Line Calibration" (LLC) and how it reins in the effect of Vdroop. I run my i5 6600K at 4.6 GHz after manually setting CPU VCore at 1.355 V in the BIOS completely stable for hours and hours of testing with Prime95 both Blend and Heat tests. From this OC, I run my 16 GB (4 x 4 GB) HyperX Fury Black C14 2133 MHz memory chips at 3200 MHz under 1.30 V (from 1.20 V), and my two GTX 970 graphics cards in SLI with their GDDR5 memory at over 4000 MHz (without changing the stock voltage) to cope with their heavily OCd GPUs. My case temps have been optimised with a laser guided infra-red thermal sensor to reach temps at or below 60 C max. under Prime95 Blend test with a budget single fan CPU cooler (no water in this fine electronics).

So, please don't tell me that I write a sentence that "doesn't make any sense". And concerning your "as far as I'm aware"... well, let me tell you that you are "obviously" far from being aware enough, and therefore you still have a long... very long way to go.

Ah!...Google... if I remember well (from 20 seconds ago) that's the name of a search engine on the internet... I suggest that next time, you check it out before you "expose naked" your complete UN-awareness, or better still, learn how to use it before you undertake a task such as the overclocking of a modern CPU. Good luck with your OC. ;-)