Looking for feedback on your experience using PBO 2 with your Zen 4 CPU. Please post results.
Define light? not -30 all core?Very Light Optimization
I think Aorus has some Bios Issues
Multi is great, Single seems kinda ****ed for some reason, though.Heres my results.
7950X
Cooler Master MasterLiquid ML240L (Have EK-Quantum Magnitude to go onto it when I can work out mounting it)
Gigabyte X670 AORUS ELITE AX
G.SKILL Trident Z5 32GB ( 2 X 16GB) DDR5 6000Mhz CL36, 1.35V,36-36-36-96
IGPU useage since need to mount water block before I can fit GPU.
CPUZ:
Single-Thread
816
Multi-Thread (32T)
16654
AMD Ryzen 9 7950X @ 5426.7 MHz - CPU-Z VALIDATOR
[7g5119] Validated Dump by bmgjet (2022-10-02 03:45:38) - MB: Gigabyte X670 AORUS ELITE AX - RAM: 32768 MBvalid.x86.fr
Ram Over Clock:
XMP to 6000mhz.
Timings tightened 34-34-34 1.36v (spikes 1.38v with LLC, droops 1.35v)
PBO +200
Core Optimiser
#1 -10mv
#2 -12mv
#3/4 -15mv
#5/8 -18mv
rest -22mv
Stablility:
1 Hour Ram Tester
15 Min linpack
1 Hour core cycler
I scored 0 in CPU Profile
AMD Ryzen 9 7950X, AMD Radeon Graphics (Raphael) x 1, 32768 MB, 64-bit Windows 11}www.3dmark.com
View attachment 2574185
View attachment 2574186
So I setup P95 using the exact settings from your screenshot and ran it for over and hour and 20 minutes. I was starting to see what I believe to be the same iterations repeating but i'm not an expert. I'm willing to help further if this wasn't it but just let me know.// Edit
I have all the logs I need now, just leaving this here for reference.
This seems like a suitable thread for asking for some help.
I'm currently adding AVX-512 support to CoreCycler, and need a log file for a full Prime95 30.8 run to double check if I correctly identified all of the tested FFT sizes.
I do have a log file that was generated on an Intel X299 platform, but for Ryzen 7000 support I really wanted to have log files that were run on a Zen 4 CPU.
So hopefully someone can upload a Prime95 log file with the following settings:
Max FFT size to 65536 is important so that it runs through all possible FFT sizes.
/snip
Unfortunately there's no good way to know if a full iteration has been completed, the most likely indication is probably that an already tested FFT size is being repeated. On the very first iteration every FFT size should appear only once, but after that, the FFT sizes can repeat multiple times before all of the others have been tested again.
On the X299 platform the first iteration took 1:46h, but it hadn't completed a full second one after an additional 2:20h. Just to give you some ballpark numbers.
Straight to 200,And what setting did you use for Boost Clock Override? Did you start with 0 boost override and -8 offset and then you went all the way up like 200 and then you were figuring out the Curves?