• 0 Posts
  • 14 Comments
Joined 1 year ago
cake
Cake day: October 26th, 2023

help-circle
  • my 13600k reach 125watts (vcore 1.077v) without any issue/ performance loss,.

    and this 125-130watts is max powerdraw in cbr23,. if i play games its take 30-55watts ,. some cases 75watts (like cyberpunk),.

    full performance with 125-130watts - 24200 cbr23 score;
    reduced performance with 95watts - 22800 cbr23 score https://imgur.com/a/owNYMrO (i use both hwinfo and hwmonitor :) );

    this 95watts is power limit over 125watts undervolt; so its not affect gaming performance and gaming power draws.







  • So, APO was utilizing an e-core from every e-core group, which enables cpu to reach a bigger cache, which leads an effect like AMDs 3d cache. It’s awesome. Has anyone tried to do this manually for 13th gen?

    like in my 13600k, i disable 2-4&6-8 e cores with process lasso,. (1 e core per group),.
    nothing changed,. same fps (521fps) in cs2 (cpu bottlenecked/ gpu75%) ,.

    until now ,. heterogeneous policy 0 with e-cores , or disabling e-cores gives better performance (1%lows),. but those not increase any fps like APO,.




  • i think you cant completely disable all p-cores ,. atleast 1 p-core have to run,.

    if we assume e-core (like i5 6400 , 4 cores)
    8 e cores give- single 902/ multi 6460 cbr23 score
    4 p cores give- single 1722/ multi 7540 cbr23 score
    1 e core give -multi 807 score ; 1 p core give - multi 1885 score;
    like 1p core =2.3 e cores,.
    in total,.
    12cores give- single 1722/ multi 14000 cbr23 score.
    (note , all values are just assumptions ,…not tested) ,. in my 13th gen cpu , 1p core =2.5 to 3 e cores,.



  • your 12700 have 8 p cores (16threads) and 4 e cores (4 threads),.
    see graphs,. those last 4 graphs are 4 e cores,.

    it mean ,. this work carry by E-Cores ,. and your p-cores carry forground works,.
    if you click current task ,. that task use p-cores mostly,.
    you can manually switch cores with taskmanager (set affinity) or you can use process lasso,.

    personally i dont like those methods ,.
    so i edited my balanced powerplan ,.heterogeneous policy 4 to heterogeneous policy 0 ,. which forcefully use all cores with hyperthreading,. and this method give best 1% and 0.1% lows in games


  • no,.
    dont use those calculaters is not work in realtime…

    firstly, are you looking for specific task ? like for single game?

    if you join 14600kf + 3070 ;
    in most of game (cpu use 50-70% usage and gpu usage 99%);
    in some cpu intensive games (like cs2, city skyline , cyberpunk 2.0 ) (cpu use 90-100% and gpu use 40-100% depends upon the graphics settings);

    when you consider bottleneck issue is ,. your cpu is not enough to use 99% of gpu in many games ,…

    so considering bottleneck for this items is pointless