A very rambly video.
tl;dr
P-cores, E-cores and ring are on the same voltage. When aiming for that 5.8-6.0GHz the voltage requested by cores can go pretty high - 1.5V or more - and while cores may be tested and validated to run with that, the ring might not be. Also reiterates that known findings from other places in terms of affected CPU models - i9, some i7, practically no i5 - and voltage curves - he is using Igor's articles - point at highest frequencies and high voltages coming from that being the cause.
My CPU reports a separate cache voltage and I thought ring and cache were same thing voltage wise, I guess I learnt something then, and 1.5v is the sort of voltage TVB sets, ever seen a i7 go as high as 1.5v? Although I do recall my i7 going high voltage out of the box (high 1.3s if I remember right) before I updated to a newer bios with different load line configuration.
I think the TVB fix and baseline as stock rollout cant happen quick enough for intel, issue of course though is vast majority of boards wont be up to date.
Will comment again after I watch his video.
--
Ok I got about halfway with lots of skipping as its a poorly presented video very repetitive.
Of course CPU's all degrade although usually its not a visible thing on chips that run at stock, I have seen a top end AMD FX chip being no longer stable at stock, that was in AMD's dark days where they were pushing chips hard.
Interesting to see the Igor graphs, something I havent seen before, so i9's push very high on voltages, i7's it seems can also push fairly high above 1.4v, I personally have never been comfortable with any voltage above 1.35v, 1.3-1.35v I will tolerate though albeit unhappily. The igor data suggests there is rampant under volting going on whether its users or board vendor side, as the voltages reported by igor are way higher than what reddit users typically report, still trying to figure out whats going on with ring voltage, buildzoid says its shared with vcore. I know it used to be on older gen chips, but I noticed on my current board, the ring voltage is separated in the settings, it is also separated on XTU. However I dont see any tools that report the live ring voltage as an absolute value. Looked at hwinfo and XTU and it isnt there. However is the following.
L2 e-core voltage offset - this I think is e-core cache voltage offset
CLR (cbo/llc/ring) voltage offset - this I think is ring, l3 and p-core cache voltage offset
IA voltage offset - should be core voltage offset
Also hwinfo reports some type of clock/voltage throttling on my ring,
Ring max vr voltage icc max, pl4, is frequently getting triggered. Just using the web browser, desktop. Might well be related to my cache under volt.
The cache voltage offset, I reported ages ago on my system has a bigger impact than core voltage offset on heat and power consumption. So is under volted on my system, but interesting to see I cant read a absolute value for it.
According to this, ring voltage is LLC voltage not core voltage.
https://edc.intel.com/content/www/u...atasheet-volume-1-of-2/002/ring-interconnect/
