Just in case 
Overclocking "Raptor Lake refresh" - 14900k,14700k, 14600k etc results, bins and discussion
31 reading
Nizzen
18,941 - 18,960 of 23,546 Posts
Joined
·
1,262 Posts
The problem is that gaming can be a heavy workload during shader compilation. Make sure to pair your settings with a power limit that is just above what you would get during gameplay only. I am pretty certain that there are plenty of reports of degradation at "only" 1.35 V.
Joined
·
225 Posts
Joined
·
3,291 Posts
To be safe we could run the CPU at it's base speed, like 3200 MHz for a 14900K, then it should be very safe. But we didn't pay for that kind of performance. One thing to disable 6 Ghz boost, but I refuse to gimp my CPU too much. I don't expect I will be able to sell my 14900K when that day comes, so maybe I will just "use it up".
Joined
·
332 Posts
Much appreciated! As far as I can tell, it is a change in 1402, but if your TVB is behaving differently that would be very interesting to know! (The tvb down-bins showed up both in the bios and in XTU when I had +2 TVB enabled, but I also had a positive temp offset so perhaps that caused the TVB bin settings to change. Interesting if so.
Edit: Ok, I just flipped back to TVB+2 with no positive offset and it has the bins set to drop 1 bin at each checkpoint. I don’t know why it was different before. Will keep testing.
Edit 2: Ok, so after some testing I think it was just that after updating my bios to 1402 the TVB menu was holding on to previous/old settings, After clearing and re-entering them it seems fixed. Interesting.
Joined
·
1,335 Posts
Even on my SP96 14900k 5.9Ht/off it needed around 1.34v in game. If you can keep it cool enough and downclock at high temps when loading, I thought it was still too insignificant to cause degradation 130w-200w in game is still alright and loading shaders has the limits.
The 14900k 6ghz I have sits at 1.25-1.27v in game atm..
The 14900k 6ghz I have sits at 1.25-1.27v in game atm..
Joined
·
332 Posts
Wow that’s a huge difference!
6ghz all core? With how much current in bios?
Mine (14700kf 5.6 all core) with 350A has 1.24V during bench and 1.31V in gaming. I had 1.25V in game and 1.2V in cb23, it was super stable, but the current was 307A and it was cutting the freq to 5.3ghz during cb23. So I put 350A and now it has no limitation but way more voltage (it's similar to the nvidia cards with PL).
The only way to decrease the voltage and keep the stability is limiting the current that cause less voltage and absorption during all scenarios but bench/cb23 (and I could get 1.20V but it was fake, the real frequency was struggling for all core to go over 5.4GHz because of the current limit).
Joined
·
1,995 Posts
I feel like people think a degraded CPU is a paperweight, when in reality you need a very substantial amount of degradation to lose -200Mhz, which means a 1-3% reduction to your framerate (in the few games you aren't GPU bound)
I have "destroyed" many chips, all of them are still running a decent OC and being used daily in rigs of friends. It's not that serious.
Regarding "AC_LL=DC_LL=LLC, CEP on": Am I actually destroying my 13600KF with LLC set to extreme and 0.01/0.01 AC/DC LL on a z790 gigabyte mainboard?? I have adaptive voltages with offset -100mv and under CB23 load it reaches 1.28v VR VOUT Vcore and also VID, but never above ~1.310v, slightly oced to 52p/40e (just to see if it works at all). But this is also with power limits on "auto" (pl1=pl2=4095w, 511.75A ICCMAX) - package temps around 80°C under load, 29-32°C idle. Ive tried SO many different settings, literally going crazy about all this AC/DC LLC stuff, but this really seems to be the only setting which gives me VR VOUT Vcore = VID under load and doesnt clock down due to running into iccmax limitation. Are there things happening "behind the curtains" that will damage the CPU rather quick?
Agree with others, we paid for these
I refuse to downclock my chip. I will OC it to the limits just being mindful with MAX vCore and temp.
For those who still afraid, I owned
K chips (14900ks / 14900k / 14700k / etc) and we should use them at their full potential, having fun OCing it. Don't afraid of early degradation.I refuse to downclock my chip. I will OC it to the limits just being mindful with MAX vCore and temp.
For those who still afraid, I owned
13900k (that early batch that potentially had oxidation issues) it worked at 360w limit for a year with 1.37v full load voltage (1.45 light load). No issues at all. Temp limit was at 90 and MAX light load voltage ~ 1.45.I honestly can't tell you. I guess it should be okay. Personally, I'm not a fan of Extreme LLC Settings because they might have a chance to raise the Voltage above the VID. But it looks okay from what you describe.
LLC8 (0.01) has a lot of undershoot/overshoot like BZ shows in this video when testing the Apex. (LLC8 shown below)
Thank you for the quick reply - now this is what I do not understand. Do I see this over/undershooting as a spike in the "max" column in HWInfo or only possible with oscilloscope? I cant watch this video rn, as I am at work
but will do once I get home.And even if this over/undershooting appears, will it be much of a problem with -100mv offset?
I tried to read much of falkentynes/buildzoids explanations but it seems to be really complicated..
Joined
·
3,291 Posts
@Jwick I did some OCCT testing. Relevant info :
The pcores ran at 53x and ecores at 43x. 277A current max, 313W max, vmin 1.128v and 92C max temp. With 400A limit removed I got 100 MHz more, but also 96C max coretemp, so a hard limit of my cooling.
Edit : In CB24 the pcores bounces between 5.5 and 5.7 even though the current peaks at 247A (400A Iccmax). Kinda strange.
![Font Technology Pattern Electric blue Symmetry]()
Code:
SVID : Auto
Sync All Cores 57x (E44, R45), HT On
Auto vcore, LLC5
ACLL=0.15, DCLL=0.75
PL1=PL2=320W
Iccmax : 400AEdit : In CB24 the pcores bounces between 5.5 and 5.7 even though the current peaks at 247A (400A Iccmax). Kinda strange.
You can't monitor them because they are too fast. There might be a chance to see one by coincidence, but you can't really monitor them. I Simply won't use such a high LLC. It is asking for troubleDo I see this over/undershooting as a spike in the "max" column in HWInfo or only possible with oscilloscope? I cant watch this video rn, as I am at work
And even if this over/undershooting appears, will it be much of a problem with -100mv offset?
I tried to read much of falkentynes/buildzoids explanations but it seems to be really complicated..
No you'll never see this in HWINFO. Have to use an oscilloscope.Thank you for the quick reply - now this is what I do not understand. Do I see this over/undershooting as a spike in the "max" column in HWInfo or only possible with oscilloscope? I cant watch this video rn, as I am at work
And even if this over/undershooting appears, will it be much of a problem with -100mv offset?
I tried to read much of falkentynes/buildzoids explanations but it seems to be really complicated..
Look at the max value under vMAX and min value under vMin and compare that to the average value. The 2 highlighted rows are most relevant.
The large droop theoretically will require a higher vMin (value in HWINFO). I've never tested this to see how much of a difference it makes though.
Nevertheless almost nobody uses or recommends using a flat LLC for this reason.
Joined
·
1,597 Posts
high ripple is no good, personally i would not like to set LLC higher than Level 6 with an Asus mobo.
would like to know why DLVR has been fused off as shamino stated, i have my assumptions but i am not sure about it.
Joined
·
1,597 Posts
now we know why Asus Z690 Apex/Extreme include true voltician.
Do you happen to know what the different LLC modes on current Z790 gigabyte boards actually mean or do you know what would be the equivalent on gigabyte boards?
18,941 - 18,960 of 23,546 Posts
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
