X HERO + 8700K. Can I disable "Ring Down Bin"?

[booya]

Is it safe? I have manual voltages without offset:
- vcore 1.2V
- LLC 6
- AVX 0
- Cache 8-45
- MCE disable
- Sync All Cores with Ratio limit 45

So now I have 800-4500 MHz with 1.2V at idle and ~1.212V under P95 with AVX. And with "Ring Down Bin" disabled, I have cache 4500 under full load. Otherwise, this thing gives me only about 4200 of cache or about.

Also, why is LLC higher than 5, considered extremely undesirable? Even in my case? And does anyone have any information about how accurate monitoring on this MB? What correction should I make to these numbers? For example, BIOS/AIDA64/HWInfo shows me 1.2 volts on vcore, how much should I add? Will 1.5% be enough? I decided that the maximum vcore for my processor is 1.25V.

When I had RIVE, I could measure the voltages at the measuring points and I've seen, that monitoring software shows lower voltages than digital multimeter. It was like:

vcore - 1.6%
VTT - 1.6%
VSA - 2.5%
RAM - 2%
PCH 1.1V - 1.6%
PCH 1.5V - 1.7%

[1ntel]

If you set LLC higher than 6 you will notice overshoot in voltage so usually anything above that is undesirable. (AFAIK with the 8700k and X Hero). Some voltage settings you will get small vdroop and small overshoot which is fine, it'll be around the voltage you have set, if unstable bump it up a very small amount until the vdroop is ok. I've not heard of LLC being higher than 5 undesirable, only higher than LLC 6 because of the overshoot in voltage under load.

Now for vcore up to 1.25v is the recommended maximum for high end air coolers & watercooling AIO's. Any higher and you will experience high temperatures in applications like rendering or high intensive games. If you wish to push higher (to a maximum of 1.4v or 1.35 i'd say for 24/7 usage) you need to delid the CPU - it's the only way.

Also stick with HWInfo64 - I've found this tool to be reliable when looking at voltages.

[booya]

1ntel
Thanks for the answer. I remember that Raja wrote, that LLC up to 5 level corresponds to the Intel specification. And overshoot can't be seen without an oscillograph. Here it is written that it is very bad - https://en.wikichip.org/wiki/load-line_calibration#Overshoot

As for the voltage, two of mine 3930k C2 degraded. One because of high VTT/VSA and vcore, and the last one only because of vcore 1.4V (constant voltage and frequency). The computer works 24/7/365. These CPUs were used 95% of time for office tasks. Therefore, I decided for myself, that 32 nm CPU should not receive more than 1.35 V, and 14 only 1.25. But about 14 nm I'm not sure. It is possible, that the number should be only 1.2V. I read that degradation does not occur from increased voltage, but from the currents. In that case, it's hard for me to understand, why my processors were degraded.

But I realized one thing, the temperature has little to do with degradation. All my CPUs were cooled by water. Yes, I used LinX and P95 to test my OC, but the total time was extremely short, considering how many this PC worked in 24/7 mode. Explicit consequences of degradation were noticeable after almost six months, but the system has lost stability much earlier, I just did not notice it. After the frequency and voltage were lowered, it did not happen again, but the processor, of course, could no longer operate at such frequencies and with such voltage.

I would also like to add, that AIDA64 and HWInfo64 show the same voltages (screenshot). But AIDA is more convenient, esp. for me (I have dark color scheme in Windows and HWInfo64 does not display everything correctly).

[1ntel]

booya wrote:
1ntel
Thanks for the answer. I remember that Raja wrote, that LLC up to 5 level corresponds to the Intel specification. And overshoot can't be seen without an oscillograph. Here it is written that it is very bad - https://en.wikichip.org/wiki/load-line_calibration#Overshoot

As for the voltage, two of mine 3930k C2 degraded. One because of high VTT/VSA and vcore, and the last one only because of vcore 1.4V (constant voltage and frequency). The computer works 24/7/365. These CPUs were used 95% of time for office tasks. Therefore, I decided for myself, that 32 nm CPU should not receive more than 1.35 V, and 14 only 1.25. But about 14 nm I'm not sure. It is possible, that the number should be only 1.2V. I read that degradation does not occur from increased voltage, but from the currents. In that case, it's hard for me to understand, why my processors were degraded.

But I realized one thing, the temperature has little to do with degradation. All my CPUs were cooled by water. Yes, I used LinX and P95 to test my OC, but the total time was extremely short, considering how many this PC worked in 24/7 mode. Explicit consequences of degradation were noticeable after almost six months, but the system has lost stability much earlier, I just did not notice it. After the frequency and voltage were lowered, it did not happen again, but the processor, of course, could no longer operate at such frequencies and with such voltage.

I would also like to add, that AIDA64 and HWInfo64 show the same voltages (screenshot). But AIDA is more convenient, esp. for me (I have dark color scheme in Windows and HWInfo64 does not display everything correctly).

When I overclock I can only go by what I have got available to me and that is the apps I use to see my voltages and my stability testing apps. I guess it depends how far you want to push your CPU.

All I know is that I have got my CPU at a higher clock with lower voltages than what it motherboard was providing at stock levels. The X Hero for me was pushing voltage of up to 1.36 momentarily.

I have my vcore at 1.24v and LLC 6. If I am getting an overshoot in voltage for a short period then I can't see it spiking any higher than a maximum of 1.3 and only for a brief period. I guess I would be more concerned if I was pushing 1.35v+

I think these motherboards market "easy" ways to overclock and, with things like multicore enhancement enabled by default where voltages are pushed in the 1.3v. A lot of people will be buying these and leaving the settings at defaults so I wonder how their CPU will hold up in the future?

I am at 4.8ghz, I did have 5ghz but needed a jump in voltage (and 4.8 to 5 is not noticeable for me - only benchmarks). 4.8 worked mostly at 1.15v but I wanted ultimate stability in many intensive apps so I over compensated the voltage as I had room and it's completely safe compared to what other people are pushing.

VCCIO is 1.000v and VCCSA is 1.100v. On auto these were much higher, 1.36v on VCCSA. Only required for RAM above 3600Mhz on this chip and, anything above 3200Mhz is not noticeable in day to day apps only in benchmarks really.

I think this motherboard just sets things stupidly high to guarantee it works on anything like high rated memory.

[booya]

1ntel wrote:
I have my vcore at 1.24v and LLC 6. If I am getting an overshoot in voltage for a short period then I can't see it spiking any higher than a maximum of 1.3 and only for a brief period. I guess I would be more concerned if I was pushing 1.35v+

Do you use manual mode or adaptive/offset? Now I have manual 1.215V + LLC 6 in BIOS and AIDA64 shows 1.216 at idle. The voltage stays the same under any load (heavy, like Prime95 and medium, like WinRar), which is interesting. But I would like to set the voltage lower, maybe 1.1-1.15V and raise LLC to level to 7. Then the processor would not have to work at a 1.2+ constantly.

1ntel wrote:
I think these motherboards market "easy" ways to overclock and, with things like multicore enhancement enabled by default where voltages are pushed in the 1.3v. A lot of people will be buying these and leaving the settings at defaults so I wonder how their CPU will hold up in the future?

I turned off MCE immediately as soon as I found out what it is. But I can write, that many are unhappy with how LLC works in adaptive/offset mode on Z370 motherboards from ASUS. I heard that there is no such problems on mb from ASRock. That's why I use manual mode, forcing the CPU to work 24/7 at increased voltage.

1ntel wrote:
VCCIO is 1.000v and VCCSA is 1.100v. On auto these were much higher, 1.36v on VCCSA. Only required for RAM above 3600Mhz on this chip and, anything above 3200Mhz

Well, even DDR4 3000 MHz may ask to increase vtt/vsa. 3000 CL14 will have almost the same performance as 4000 CL19, so the timings are matter.

1ntel wrote:
is not noticeable in day to day apps only in benchmarks really.

Arma 3 and Fallout 4 🙂

[DirtyMacho]

ok i might need some input.

I am even fine with 4.5ghz but there is not downgrading of cpu clock when not in use. or under heavy load. It is not jumping down it sticks to 4.3ghz for now. As for 4.5ghz i believe i needs to push to 1.27v

I am currently hitting 80degree at 4.3ghz but when i set voltage and core ratio to auto it is spiking to 95+ degree celcius. So i have to down volt to 1.25 or 1.26

let me know what is the best setting to set at 4.4-4.7 ghz . i want to under clock when no load. There is no point to clocking alway 4.3ghz as its summer and ambient is reaching 38-41 degree celcius .

So i have to ensure the corsair h100i v2 keep the cpu cool.

Right now i am running push pull so the temps are under 80 deg But before push only it hits to 91 degree on cinebench.

i dont know if this cooler is suffice or i need a 280 or even 360mm rad. Does the rad size matters for this cpu. For oc'ing it upto 4.8ghz or even 4.5ghz

I

Is there any table which tell what freq what voltage is min and max . it would be easy to set and see.

I believe 1.250 is higher for 4.3ghz .

[booya]

DirtyMacho wrote:
I am even fine with 4.5ghz but there is not downgrading of cpu clock when not in use.

In order for the CPU to downclock, some of these options must be enabled: Intel SpeedStep, Turbo Mode, Intel Speed Shift and CPU C-States. I did not turn them off.

DirtyMacho wrote:
As for 4.5ghz i believe i needs to push to 1.27v

My current OC (not tested yet, only Windows and short tests): 4.8 GHz, AVX - 3, HT off (in the near future, I will not need it). 1.216V in AIDA64/HWInfo. I think your CPU could handle 4.5 GHz at ~1.2V (without AVX offset and with HT). Even under heavy load.

DirtyMacho wrote:
let me know what is the best setting to set at 4.4-4.7 ghz . i want to under clock when no load. There is no point to clocking alway 4.3ghz as its summer and ambient is reaching 38-41 degree celcius .

I made some screenshots with my settings - link. Try them, except memory frequency and it voltage. It's for 800-4500 MHz, with HT and zero AVX offset.

DirtyMacho wrote:
Is there any table which tell what freq what voltage is min and max . it would be easy to set and see.

Just try not to exceed 1.25V under any circumstances and, just in case, add 1.5% to the voltage that BIOS/AIDA64/HWinfo/CPU-Z shows.

DirtyMacho wrote:
I believe 1.250 is higher for 4.3ghz.

Most these CPUs must operate at this frequency at a much lower voltage. Yours, for sure, too.