Overclock.net banner

4041 - 4060 of 4151 Posts

·
Registered
Joined
·
999 Posts
multi and Bus
It is a bug. RealBench 2.56 needs an update. It does not know how to correctly determine the CPU speed or the BCLK speed.

Some common timers within Intel CPUs run at 24.0 MHz or 19.2 MHz. Instead of RealBench determining the correct BCLK speed, it finds one of these timers and takes a wild guess that this must be the BCLK speed. After that it takes another wild guess. The CPU is running at 3693.4 MHz so ...

3693.4 MHz / 24.0 MHz = 153.9

It rounds that up to 154 and thinks that the CPU must be using the 154 multiplier which of course is impossible.

The 10900K uses a base multiplier of 37.

3693.4 MHz / 37 = 99.8 MHz

That looks pretty close to the actual BCLK speed that your CPU is running at.

Long story short, the multi and bus reported by RealBench are two meaningless numbers. Ignore those numbers.
 

·
Registered
Joined
·
251 Posts

·
Registered
Joined
·
182 Posts
Yes that push the performance extreme and not only in FPS, the most programs get a really good gain.
That does much more as 2% for 100mhz CPU clock, the second thing is it needs not much more Power, at 5ghz the CPU is really good at fps per Watt, better than an AMD.

In the most cases i drive only 5ghz 4666CL17-17 because same FPS but much lower power arround 20-35W in game.
 

·
Registered
Joined
·
546 Posts

·
Gaming Overclocker
Joined
·
2,031 Posts

·
Gaming Overclocker
Joined
·
2,031 Posts
Higher VRM SF means lower transients.
If you use an override voltage, and a fixed clock frequency, your VRM will be asked to supply a non-variable voltage.
My CPU scales from 640mv to 1500mv in a very short period of time.
In this case, the fast VRM response is the difference of to be stable or BSOD.
In your case, the only difference you'll see will be the temperature.
But even knowing this I'd test anyway... LOL
I've tested with 800KHz and seen high vrm temps after only 1h of Realbench (the vrloop 2 max may be bugged though):

2489309


So I went back to default, just in case anyone is using like 600 or 700KHz?

In all cases it's possible to find a good setup, in my case it's with selected hw so it's more possible as normal.
I see, delidded I suppose as well.
I've only put some liquid metal (Thermalright) on the top of the cpu and gained those 3 degrees, the American RockitCool stuff is still too expensive to get shipped to the UK (and customs cleared).
Anyway I'm gonna try this 4600 cl17 profile from my Ace's "Memory Try It" within the bios.
 

·
Registered
Joined
·
182 Posts
I know that is possible with 2x16GB to boot 4600Mhz on Z490 Ace, but the stable limit by snakeeyes from Hardwareluxx Forum was 4400mhz with GSat,
perhaps with newer Bios is possible.There was it also CL16-18 is easier than CL17.
 

·
Registered
Joined
·
200 Posts
Hi all. I finally got my hands on a 10900KF at a decent price. So far I am very impressed with the overclocking experience coming from a 10600K (which was much less forgiving in terms of voltages and temperature).

Obviously the 10900KF pulls more current, so Vdroop is greater than I was used to with the 10600K.

So far I am experimenting with HT off and trying to get 5.4 GHz all core (for gaming), I can set something like 1.41 V adaptive LLC5 (MSI) and see maximum Vcore readings of 1.326 V in HWiNFO. Previously, I got good results with even droopier adaptive LLC6 (MSI) on the 10600K, possibly due to better transients with the droop.

My question is, if I set 1.41 Vcore in BIOS, will any part of the chip be receiving anything in the range 1.41 V - 1.35 V that might potentially cause degradation (i.e before droop is applied), or am I simply safe with it as is (given the low Vcore reading of 1.326 V, max VID of 1.352 V and temps <66 C)?

For example, would it be better to go for a less droopy LLC setting and setting a lower Vcore in BIOS, but with the risk that it makes transients worse?

Any hints and tips would be much appreciated!

Edit: no longer pursuing HT off, both GTA V and AC:Oddssey had noticably lower minimums/dropped frames with HT off at 5.3 GHz vs. HT on (though I have yet to get 5.3 GHz HT on fully stable = occassional level 0 cache errors). Instead, currently running happily with HT on 5.2 GHz all core, LLC6 (MSI, very droopy), adaptive 1.41 V set in BIOS, 62 C and 1.29 V (under gaming load), max observed Vcore is 1.33 V (transient low load). I keep dipping into 5.3 GHz all core, but may need to go a bit harder on LLC to LLC4 or 5 (MSI); though I am not sure it will be worthwhile, depending on the voltage increase needed under load.
 

·
Registered
Joined
·
278 Posts
Can someone help me to set correctly the adaptive voltage, please?

Actually I have a fix voltage of 1.34v (LLC4 on Asus Z490-E) and in load of 1.261v. Now, I can set as adaptive voltage whatever I want (I tried both 1.34v and 1.261v): on load it always stays at 1.19v, so it does not change at all!

What should I do in order to reach 1.261v?
 

·
Registered
Joined
·
200 Posts
Can someone help me to set correctly the adaptive voltage, please?

Actually I have a fix voltage of 1.34v (LLC4 on Asus Z490-E) and in load of 1.261v. Now, I can set as adaptive voltage whatever I want (I tried both 1.34v and 1.261v): on load it always stays at 1.19v, so it does not change at all!

What should I do in order to reach 1.261v?
Using offset mode, or modify individual VF points often seems to overcome these sorts of issues. The other factor can be making sure that voltage control software such as XTU, throttle stop, (MSI) Dragon Centre are uninstalled as they can override BIOS voltage settings. Also moving up to the next most aggressive LLC notch up can sometimes fix this sometimes, I was seeing similar in my initial testing.
 

·
Registered
Joined
·
278 Posts
Using offset mode, or modify individual VF points often seems to overcome these sorts of issues. The other factor can be making sure that voltage control software such as XTU, throttle stop, (MSI) Dragon Centre are uninstalled as they can override BIOS voltage settings.
How should I use offset mode or modify the individual VF points?

I don't have such software.
 

·
Registered
Joined
·
278 Posts
I'm not an Asus user, you'll have to look up a video on how to do this in your BIOS. Might be easier for you to try LLC3 first.
I almost solved using offset mode (+0.050v).

However, the voltage goes rarely to 1.241v as with the fixed voltage, then it stays at 1.252v in load – also if I set +0.040v offset.

Do you know how could I solve this?
 

·
Registered
Joined
·
200 Posts
I almost solved using offset mode (+0.050v).

However, the voltage goes rarely to 1.241v as with the fixed voltage, then it stays at 1.252v in load – also if I set +0.040v offset.

Do you know how could I solve this?
Too much Vdroop, possibly defaulting to the standard VF curve. Drop Vcore back down a bit and go up an LLC notch. If that doesn't work, keep the new LLC setting but also nudge Vcore up step by step.
 

·
Registered
Joined
·
278 Posts
Too much Vdroop, drop Vcore back down a bit and go up an LLC notch. If that doesn't work, keep the new LLC setting but nudge Vcore up step by step.
I think, I solved the problem, thank you very much for the effort.

For 1.243v (@5.0 GHz) adaptive voltage I must use:

LLC4, vcore 1.240v, offset +0.040v. On Windows the max vcore is 1.323v

OR

LLC5, vcore 1.240v, offset +0.015v. On Windows the max vcore is 1.279v. So there is much less drop as with LLC4. For info: with a offset +0.010v I have as vcore sometime 1.234v and sometime 1.243v while stressing. So, I must set this "0.05v" more if I want to stay fix at 1.243v while stressing.

I would say, that I must stay with the LLC5 option, right?
 

·
Registered
Joined
·
200 Posts
I think, I solved the problem, thank you very much for the effort.

For 1.243v adaptive voltage I must use:

LLC4, vcore 1.240v, offset +0.040v. On Windows the max vcore is 1.323v

OR

LLC5, vcore 1.240v, offset +0.015v. On Windows the max vcore is 1.279v. So there is much less drop as with LLC4. For info: with a offset +0.010v I have as vcore sometime 1.234v and sometime 1.243v while stressing. So, I must set this "0.05v" more if I want to stay fix at 1.243v while stressing.

I would say, that I must stay with the LLC5 option, right?
It does look like your LLC5 setting is the best, provided that it remains stable. Glad that worked out for you!
 

·
Registered
Joined
·
251 Posts
The system is completely stable at:

56x4 - 55x6 - 54x8 - 53x10 (+2Boost-OCTVB)


Now I'm testing

57x2 - 56x4 - 55x6 - 54x8 - 53x10 (+2Boost-OCTVB)

2489429
2489433


2489438
 
4041 - 4060 of 4151 Posts
Top