Originally Posted by crakej
I would not be the right person to do that (even if I wanted to!), but there are a few that have done their own testing which indicates it's just not the case. The only spikes that happen, happen on VDDCR CPU and do not reach the CPU due to droop, and are not any higher than spike sprovided for XFR.
In all the time we spend monitoring our voltages we just don't see these spikes. Some would say the equipment being used to measure is not fast enough to see the spikes, but I feel the amount of time we're monitoring our systems is plenty enough for us to be able to see these spikes in HWIfo (Or Aida etc) - I see other voltage spikes elsewhere so why not here?
Anyway, I have put my trust in the evidence that's been provided on this thread on many occasions. I find it hard to believe LLC will damage anything - settings that have dangerous settings are highlighted for us in the bios using different colours to represent risky settings - yellow for some risk and purple for high risk.
I think that Anty, soureraser and Acidstorm replied for me. So I have nothing to add about that.
Anyway, do what you think is better for you
Originally Posted by mat9v
On the subject of voltage spikes as they are measured at the back of the CPU:
1. Spikes are on order of 0.06V as measured and shown by AMD engineers on one of their posts/videos (can't find the link now) - so since we know that AMD suggests keeping voltage below 1.425V for the long term OC-ing, then to be perfectly safe we should keep below 1.34V LLC5 to get the same.
2. Personally, I will be exchanging this CPU for ZEN2 or maybe even ZEN+ so I don't really care if I damage it a bit in the meantime. That said the real damage is done when running transistors with high voltage and temperature constantly for a long time, as in setting 1.425V and playing with 48H of OCCT to test stability, jumping power states and resulting spikes in voltages in the same frame of 1.425V for miliseconds will do the same damage after 20 years
On another topic, my CPU is unable to stabilize at 4.2Ghz with memory set to 2933 - I can run it at 4.2 but with memory at 2400 - so no deal for me.
The final score is 4.1Ghz at 1.368V with SoC at 1.025V and good timings for memory. I like this
So your final score with 4.1 ghz is 1.368v LLC3 or 5?
And ram speed, timings and voltage?
What stress test have you passed (like 2 hours OCCT, 6 hours blend/custom max ram, 1000% memtest?) can u post some screenshot?
I want to compare if there's really a gain with a newer cpu like yours 38week vs mine 32week