Here're the results (3DMark):
- Steel Nomad: from 6870 points to 7546 points – + 9.84%
- Portal Royal: from 18848 points to 20988 points – + 11.35%
- Time Spy Extreme: from 11147 points to 12581 points – + 12.86%
- Time Spy: from 23304 points to 26587 points – + 14.09%
- CPU Profile: from 7877 points to 9156 points – + 16.24%
So, in total 12.88% improvement for FREE.
ALL I changed will be shown above. CPU and RAM at BIOS and GPU at MSI Afterburner.
CPU = i5-12600KF
- P-cores: 5.2GHz
- E-cores: 4.2GHz
- Ring: 4.4GHz
- AVX Ratio: -2
- LLC: Mode 4
- Voltage: Offset +(By PWM) 0.05v
- CPU SA: 1.25v
- CPU VDDQ: 1.35v
RAM = G.Skill Trident Z RGB
- Frequency: 6400MHz
- CL: 32
- Extension Mode: Performance
- Voltage: 1.35v
GPU = MSI Vanguard SOC RTX 5070 Ti
- Core: +400MHz
- Mem: +3000MHz
- Power Limit: 116%
It's 100% stable. Tested with Prime95, Furmark2, AIDA64, Cinebench R23, OCCT and Games. Temps are GREAT as well, not reaching 100C at the CPU even at those stress tests(up to 240W) and keeping around 50-65C at gaming, GPU gets up to 60C at MAX(350W).
In case someone wonders about it, the CPU cooler is a 360mm AIO, the MOBO is an MSI Z690 Carbon and the PSU is a XPG Core Reactor 750W.
4 Comments
100% your 5070 Ti is error correcting its VRAM speeds during intensive gaming.
+250 core and +1000 mem and +116 pl gives me 30k in timespy, have you tried less aggressive values yet?
What was your real world gains when it comes to gaming and everyday tasks? I’ve noticed that pushing my OC settings higher to get better synthetic benchmarks didn’t translate to any real world performance gains tbh. I’ve noticed that sometimes when i scored higher in 3d benchmarks my actual gaming performance got worse 🤷🏻♂️
5.2 ghz and 4.2 ghz is pretty good on a 12600k tbh. You did add an avx -2 so stress tests will really run at 5ghz . Some games will also trigger an avx offset which is why I’ve always preferred to just occ with -0 on avx