Originally Posted by Axaion
I just updated from 22.214.171.124 to 126.96.36.199ABB and the volts are the same for me.
I did lose 80 MHz across all cores though.
Lost a few points in CPU-z bench (528 Single core now, was 531-532 before)
But eh, rather have it not try and kill the chip, but all this boost madness is a major fail from AMD's side.
Synoxia, see above, so this user/CPU sample is behaving differently and below on differing model another see as I do.
Originally Posted by crakej
I have to agree with you. My 3900x also has this behaviour on 1003 - slightly higher volts for same, or slightly less ACBMHz.
I've still only ever seen a light load boosting properly - voltage and MHz drop the moment any real work starts....
Originally Posted by lordzed83
I know that when installing Lew chipset drivers on 2501 i was loosing performance compared to lunch day chipset drivers. But that is not the case with ABB bioses.
Yeah I thought stock CPU with OC RAM on UEFI 0002 performed well. As PBO+150MHz doesn't attain the same clocks as older AGESA UEFI doubt the performance will be there, will check. Must also test ratio OC as well, perhaps that will show gain of performance over older AGESA at same clocks.
Originally Posted by Enzarch
Been lurking here a long time, figured I'd say hi and ask a couple Qs
Why is 'auto' PLL voltage so high? (over 2V) Is there any risk? I do seem to be able to get a bit more clock on auto than at 1.8V, seems marginal effect on temps
Anyone else use AIDA as their primary monitoring app? and notice more cpu performance degradation with it, more-so than others such as HWINFO (roughly 150pt loss in CBr20 vs -70pt)
3900x @ 4.5 / 4.5 / 4.425 / 4.45Ghz @ ~1.325V (load) (~7950pt CBr20)
B-Die @ 3600Mts 14-15-14-28 (~65.5nS)
BIOS v0002 (excited to see CCX OC coming to bios maybe)
On [Auto] PLL has remained 1.8V for me, but I also set it manually on OC profiles. Perhaps try that and check if LN2 mode switch is not set to On.
Originally Posted by Synoxia
Source that 1002 was killing cpus at stock? I run 1002 and 3700x with -0.037 undervolt and voltages/temp doesnt look scary.
X570 aorus xtreme is reaching advertised (and more) clockspeed on 1003abb agesa... check hardwareun
boxed review, i think there's a lot to be sorted out
Ah btw folks, i wanted to point out that amd changed 2700x max boost on their website... from 4.350 to 4.3... and they also nerfed 2700x... prior to 3k bioses i was reaching 4.3 ghz in games, on 2501 bios 4225. (Pbo on)
I haven't seen a source. Odd post here or there on web that a CPU died and IIRC some at stock, so stats could just be with normal range.
Shamino has only given us an insight on reasoning for boost behaviour change, he hasn't clearly stated that AMD have given data or stated unequivocally that silicon deterioration had been observed.
To be honest yeah I like the boost on older AGESA with PBO, but if I look at it in terms of %, it's ~3.5% gain.
Now I have no source, but just my opinion
What if this boost gimping on new AGESA is to keep product segmentation. For me if I looked at it, by gaining ~4.3-4.34GHz ACB on my R5 3600 I wouldn't buy or suggest buying a R5 3600X.
Currently a 3700X is ~£330, I have seen it as ~£300, would I buy a 3800X for ~£370-£380 if PBO OC allowed a boost gain on 3700X?