SumsarTheThird

so i decided to toy around alot with it today, and holy bobby did i get smarter once i realized what was happening. i ended up with curve -20 PTT = 140 TDC = 100 EDC = 190 (lowering this any further, reguardless of whats done to the other values, have a direct inpact on max speeds) and it settled down on 2-5mhz below 4400mhz on all core. and 5.05 peak on singlecore. doing 4.85-4.9 in cinebench singlecore. (this is also peak, as it "builds" up to that speed before the next square starts, and it drops to almost 0 before building back upto those 4.85-4.9 again. although there is some squares where it never even reaches 4ghz still.) but the biggest difference was the temps. shaving off those 70-80 mhz in comparison to the picture above on allcore, lowered the temps by 10 degrees under all core load on air. so i swapped back to my kraken x63 and saw a cool 69 degrees under load. 74 in prime95 (in which i throttled before) both on aio and air, i saw idle temps of low to mid 30's respectively. so well im not gonna get my chip as high as i'd like, but i got significantly lower temps and a slight singlecore boost in comparison to what i saw before. i guess i cant ask for much more. win/win right ? Final cinebench R23 scores is: MC: 22261 SC: 1621 if someone could tell me how those scores look, i think i could just leave it where it is now.

i absolutely did. im doing a fresh install more or less everytime i change any hardware (except from coolers :D) also updated chipset driver, mobo firmware and windows prior to writing here. your message above have some encouragement tho. tyvm. im running a 3080, so i certainly dont hope that is limiting my chip added two pictures. yesterday i did the same as in picture one with edc at 200. but it wouldnt run them at the same speed for some reason i will today. if i limit edc to 180, one ccx goes to just shy of 4600mhz, while the other stays around 4100 mhz. Now you say its a basic board, and im wondering if i replace it with something higher end, would it then deliver power to the chip differently, that would allow me to see better temps/clock speeds ? if so, do you have any recommendations ? it just feels shite to sit and play a quite cpu intensive game, and having low frame rates, then go into ryzen master or hwinfo and see that the cpu is chilling at stock clocks more or less. (a little above, but still) and i dont understand why its not just pegging a core when its needed. my friend that plays the same game on an almost identical pc, he is just running a bone stock 9700k and our frames differ by upto 20 frames in that game. which makes a hell of a lot of difference when that takes you above 60 fps. and this is actually the root of why im trying to achieve some proper boost speeds.

what about when you game, if you do. that actually the thing that started all of this. whatever core is active while i play, its usually just hovering around 3.7 - 3.9 ghz, with a verry rare sudden spike to something higher, 4.3-4.6 ish. it doesnt maintain this for longer than i takes me to blink, and falls back to the 3.7 - 3.9 range again. and this just repeats. is yours pegged at a higher clock when gaming ?

what voltage do you see in SC and MC in cinebench ? just out of curiosity

reguardless of curve or whatever. i can finish a 30 min cinebench run without thermal throttling. but running prime 95, takes it to thermal throttle territory instantly. and makes it hit 90 within a few mins. also reguardless of cooler.

Right, so i've done a little more testing. so i can add upto a -25, at -30 one ccx is clocking 3-400 mhz lower than the other. lowering the curve value by 5 untill it didnt happen was 20. Then i added the limits 200/130/130, and the same happened again. one ccx is significantly lower than the other. lowered the curve even more setting the limits to 200/200/215 -15 is the only way i can keep both ccx at the same clock speeds setting the slightest limit below this, and one ccx is clocking lower than the other. reguardless of the curve value i have. i also kept a keen eye on the voltages, which honestly doesnt really seems to change reguardless of the curve setting. only thing that really changes with curve is more heat and tiny clock speed upgrade. also, checking tdie in hwinfo shows one of the ccx's running 6-12 degrees hotter than the other. yaaay. third bottom of the barrel ryzen cpu in a row. (basicly each gen) So my take from this is i got a chip with a ccx that barely classified. and when it comes to temps. @freeagent idk what you are cooling yours with, but just booting my pc takes me to 60-68 degrees. and idle is 40-50. and yes, i've checked my cooler mount countless times, and tried with 3 different coolers now. Dark rock pro 4, Kraken x63 and an NH-D15. all are pretty similar temps. so idk. nothing more to do i guess. ill probably just let this cpu cook with the "unlimited" limits, and seriously consider if im spending a dime more on ryzen.

a cross from edge to edge with a little dot in between the lines. also went and picked up a Dark rock pro 4 yesterday, just to see if it was maybe my aio that was funky. absolutely no change, other than it running a little bit hotter now. looking through hwinfo after a gaming session, while also keeping an eye on it during: highest speed recorded, 4.745 ghz. while that was the highest recorded, whenever i was glancing at it i only ever saw it hovering around 4ghz on the highest cores. while all other cores doing almost nothing. does your cpu actually sustain those 4.9ghz+ clocks while gaming ?

Ram: G skill trident z. 3200mhz cl14. Mobo: https://www.gigabyte.com/Motherboard/X570-AORUS-ELITE-rev-10#kf i tried toying a little with the pbo curve. i've been at -20. but im now thermal throttling during cinebench. only reaching 4.45 all core. again i feel like i lost the silicon lottery by miles, as people on air coolers are claiming 4.95ghz SC. which frustrates me. another question tho, how do you measure single core performance ? when i do single core cinebench, its umping between serval cores, and only showing me peak speeds of under 4 ghz

So i recently bought a 5900x and an accompanying x570 motherboard. and i i feel like im not getting the full effect of the processor. The frequencies seems low in comparison to what i read people are getting. I've added a picture from 3 different scenarios. with a little log from hwinfo as well. While gaming, i rarely see it go above 4.0 ghz, if it does, i spikes to somewhere around 4.5 and then back down again straight away. and since i play a rather cpu intensive game at the moment, it seems like there is quite a lot of performance left on the table. so i tried both PBO and auto overclocking. which both did absolutely nothing for the cinebench scores or the clock speeds. on the topic of the scores. it scores around 20300 (+/- 150) Motherboard the 5900X is placed in is: https://www.gigabyte.com/Motherboard/X570-AORUS-ELITE-rev-10#kf and its cooled by a kraken X63 AIO. pump speed is at 100% and this is even with fans almost cranked to the max as well. is there something im doing terribly wrong or did i absolutely lose the silicon lottery ? Cinebench all core.CSV Cinebench single core.CSV While Gaming.CSV

okay so i tried this on all my nvme's, one at a time. nothing changed.

I'll try to install windows on a drive I bought 2 weeks ago. And leave all other out of the system. I'll update soon

Manufacturers own tests. But I've had windows installed on 3 nvme drives and 2 ssds. From different manufacturers. Still get this.

a little question that i hope could help locate the issue. It seems that after it "freezes" some games have to get reinstalled or told where the gamefiles are. also some programs have had to reinstall modules on startup, like msi dragon center. Could this point to something ?

Alright. im grateful for your time man! really! i was afraid it was either motherboard or cpu. i mean, i have been running the cpu with PBO enabled all the time. So i wouldnt be able to rule out that it might have been damaged, as amd do warn users when enabling it. so i guess im going on the market for something new. amd 5000 series isnt far away so i might just let the system suffer untill reviews are out, to see if its gonna be an intel or amd system this time around. Once again, thanks a lot for your time!

All of them pass without issue and are in "good health"