Overclock.net banner
8,981 - 9,000 of 23,546 Posts
both cinabench and ycruncher vt3(dependent on memory speeds but still does not compare to small) draws less current than OCCT SMALL AVX 2. You must run it for atleast 1hr tho. R23 is too tame of a benchmark and it can never be considered to be a "useful" stress test.
pulling 400w on OCCT small requires >1.25v vmin at 5.6/4.4 all cores ht enabled. Current will be slightly over 300amps. This is coolable(<100c) but barely with a strong custom loop at 20c liquid temps without delid/direct die. By running this, you are violating intel electrical specifications.
Really 400W at stock!? Did you test the same with 253W PL too to see how far down the core clocks go? I ask because I have not run OCCT with HT on yet. Only HT off.
 
Been pushing testing ht off and 63/48/52 requires 1.35vmin (1.42v llc7) in order to pass y-c and r15.

Seems to be the edge of efficiency on gaming loads.
 
  • Rep+
Reactions: Neo_Morpheus
Really 400W at stock!? Did you test the same with 253W PL too to see how far down the core clocks go? I ask because I have not run OCCT with HT on yet. Only HT off.
my KS pulled 370w vrout (no throttling) stock with MCE enabled. No, i have not. it will prob go <5ghz lol. i will try it. FYI my KS is avg with a 5.8 vid of 1.40xv and 6ghz vid at 1.472v.
Image

Image


One might ask " wait a min, intel stock violates intel electrical spec ???????? " techincally 5.6 all core is not the spec for the 13900KS. At 5.4 it wont be needing this much voltage.
 
One might ask " wait a min, intel stock violates intel electrical spec ???????? " techincally 5.6 all core is not the spec for the 13900KS. At 5.4 it wont be needing this much voltage.
Yes 5.4 with 253W PL would be it for the KS. Turning HT off kinda fixes it though.
 
Does anyone brave man except @Falkentyne run his chip LLC4 at 1.5V+ and did not degrade?

I am observing better stability with LLC4 over LLC6 on both of my chips and different mobos , those transient spikes are definitely there causing random instability.
 
Does anyone brave man except @Falkentyne run his chip LLC4 at 1.5V+ and did not degrade?

I am observing better stability with LLC4 over LLC6 on both of my chips and different mobos , those transient spikes are definitely there causing random instability.
You could use voltage suspension to clamp idle voltage. Voltage suspension is buggy on the white apex. didn't test it on the encore yet. You could just use LLC5.
 
Does anyone brave man except @Falkentyne run his chip LLC4 at 1.5V+ and did not degrade?

I am observing better stability with LLC4 over LLC6 on both of my chips and different mobos , those transient spikes are definitely there causing random instability.
My chip auto runs 1.5+ like I'm single core loads it hits as high as 1.518V die sense. This chip has 1.458V VID at 6GHz.
 
You could use voltage suspension to clamp idle voltage. Voltage suspension is buggy on the white apex. didn't test it on the encore yet. You could just use LLC5.
I have seen this within bios but never played with it , is it functional on a Static Oc and fixed Frequency ? From what i understand you can set a floor and a ceiling voltage so i could use a lower ceiling voltage and still use llc4 , right? But the question is how do i set these within bios?

My chip auto runs 1.5+ like I'm single core loads it hits as high as 1.518V die sense. This chip has 1.458V VID at 6GHz.
We have the same 6Ghz Vid , nice to hear that you are ok with this kind of voltage but is this voltage idle as well or you let the cpu downclock and reduce the idle voltage ?
 
The KS has a 320W extreme power profile.
The 13900K and 13900KS are the same architecture and design, so why would the KS be able to handle (in regards to degradation and longevity) a 320W powerdraw any better than the 13900K is another question... Intel could have set the same "Extreme power profile" of 320W for the 14900K aswell since the "14th gen" is even newer silicon.
 
  • Rep+
Reactions: logikju and yzonker
I have seen this within bios but never played with it , is it functional on a Static Oc and fixed Frequency ? From what i understand you can set a floor and a ceiling voltage so i could use a lower ceiling voltage and still use llc4 , right? But the question is how do i set these within bios?



We have the same 6Ghz Vid , nice to hear that you are ok with this kind of voltage but is this voltage idle as well or you let the cpu downclock and reduce the idle voltage ?
Normally I run 5.7/4.5 at 1.25llc7. but if I were to run stock I'd just let it idle at <1V with cstates and stuff. I always disable hyperthreading
 
The 13900K and 13900KS are the same architecture and design, so why would the KS be able to handle (in regards to degradation and longevity) a 320W powerdraw any better than the 13900K is another question... Intel could have set the same "Extreme power profile" of 320W for the 14900K aswell since the "14th gen" is even newer silicon.
Good question! From the article that i took the excerpt from and from intel thesmelves, have mentioned that power limits are configurable by the motherboard vendor and does not void warranty. Well, the warranty part is not explicitly mentioned by intel but tom's hardware obtained that information from official intel representatives. So, by the looks of it intel raised the power limit (on select motherboards that support the extreme power profile) for the KS skew to show meaningful performance uplift from I9 K on sustained heavy workloads when following intel's default guidlines. However, as you are aware, most of the Z series boards remove these limits out of the box and intel does not seem to care. Well, if they did, they wouldn't allow motherboard vendors to do that in the first place. Electrically there is virtually no difference between the KS and K skew, so the K skew is just as capable to pull 320w. KS is just factory OCed along with better/tighter silicon quality.
 
I have seen this within bios but never played with it , is it functional on a Static Oc and fixed Frequency ? From what i understand you can set a floor and a ceiling voltage so i could use a lower ceiling voltage and still use llc4 , right? But the question is how do i set these within bios?
it is "supposed" to function with static voltage too. Your assumption is indeed accurate. well, ceiling means it wont exceed that voltage under any circumstance. So, if you need a vmin of lets say 1.25v to pass all the workloads that you need the cpu to run, you can set a ceiling of 1.25v or slightly above. This would eliminate the unecessary idle voltage that must be present for vdroop on loads.
 
My chip auto runs 1.5+ like I'm single core loads it hits as high as 1.518V die sense. This chip has 1.458V VID at 6GHz.
I don't understand what is different between our systems. I never see even 1.4v at default clocks on my 14900k p-sp 107. Not at idle or light load. It regularly sees 6 Ghz according to HWINFO to. So boost is working.

This is bios defaults on the white Apex.
 
However, as you are aware, most of the Z series boards remove these limits out of the box and intel does not seem to care. Well, if they did, they wouldn't allow motherboards vendors to do that in the first place. Electrically there is virtually no difference between the KS and K skew, so the K skew is just as capable to pull 320w. KS is just factory OCed along with better/tighter silicon quality.
Maybe thermal throttling is what Intel considers the safeguard for most people, so that temps will be a problem for most users long before demage can be done from high powerdraw and current. Most people use either an air cooler or AIO, so makes sense that way. That and the fact the people tend to buy a new CPU before degradation becomes a big problem.
 
  • Rep+
Reactions: Jwick
Maybe thermal throttling is what Intel considers the safeguard for most people, so that temps will be a problem for most users long before demage can be done from high powerdraw and current. Most people use either an air cooler or AIO, so makes sense that way. That and the fact the people tend to buy a new CPU before degradation becomes a big problem.
well said.
 
Image


That was a very rich lm application and in just 2weeks the iceman sucked and dried a ton of lm for a 2nd time . Something is seriously wrong with their nickel plate as I have owned a few nickel plated waterblocks and none of them did this!

No wonder why I started needing more voltage. And I don’t like having to reapply lm every few weeks.
 
  • Wow
Reactions: owikh84
8,981 - 9,000 of 23,546 Posts