Ive gone through almost 2 failing power supplys from am4/am5 and im here to ask a question that many people have avoided.ARE am5/late am4 cpus even going to last 10 years?Because what i have learned is that in the aggressive overclocking that AMD says is safe is that the CPU frequency on most amd motherboards CONSTANTLY flucates to absurdly high/low values within less than a second. putting a strain on the PSU and slowly causing it to fail..Another thing is that EARLY am5 cpus and later am5 cpus reach tempatures of 95C ..and yes they often reach higher than that...AMD claims that this is normal ... and the reason they hit these tempatures is BECAUSE of the extreme overclocking via precision overdrive that is done. AMD says that at least for my am5 cpu(AMD Ryzen 7 7700X) that it is rated for temps of(TJMAX) is ALSO 95C giving it just a 0C difference. A single hairdrop could cause it to reach over that limit, and let me tell you it CONSTANTLY does this...So again... HOW LONG are these cpus going to last realistically.
>>109607209i've been running my Ryzen 7700 (non-X) for like three years now and had no idea about the power draw thing being an issue. granted i happened upon an 800w gold psu and my gpu barely pulls 125w or some shit. my biggest gripe is despite testing fuck off large noctua and thermalright coolers the CPU refuses to idle under 50c and with programs open idles at about 60c on Linux. scared the shit out of me moving from 4th gen intel and even to this day kinda irks me to check any task manager. benchmarking on windows as a sanity test shows it runs well above average for this die but seeing temps that high is just so anxiety inducing.
>>109607209>ARE am5/late am4 cpus even going to last 10 years? Have been running a 3600x for about 10 years for my server.A 5600x for about 3-4 years as my main. Built a 2600 for my father-in-law, been using about 7 years. NAS in the garage is a 2700 that runs some docker services I've had easily since 2016.Basically never had an AM4 shit itself. Two dead motherboards, several dead HDDs but overall AM4 itself has been the most reliable CPU I've had.>t.millenium boomer
>>109607209>ARE am5/late am4 cpus even going to last 10 years?lol no
Never in my life I've had a PSU die on me, all EVGA btw.
>>109607878the one that failed on mine was a EVGA gold .
>>109607209These CPUs support limiting max temp to 75C. At least my 9700x does. There isn't that much performance in running at 95C since efficiency goes way down with temp.
>>109607209>I don't want it getting hot ever!So put a decent cooler on it and limit it to 65W.
>>109607209Stop cheaping out on shit PSUs. Get gold rated or go home
>>109607789Idk if you remember but I replied to you once about also getting those. I swapped my 5900X for a new 5800X3D and got two more of those.Replaced my motherboard and looking out for any more...
>>109607209>HOW LONG are these cpus going to last realistically.long enough to go obsolete, like any cpujust turn off PBO and it'll become a retrocomputer someday
>>109610366>long enough to go obsolete, like any cputhe joy of AM4 and AM5 is that if your CPU actually dies, you can just upgrade to a new model that's several generations better without having to swap the rest of your hardware>>109607209another data point here would be ASRock failing to stop 9000 series AMD cpus dying on their mobos
>>109607209Considering that AMD is more interested in hiring neckbeards like Gooners Nexus to shift all the blame to motherboard manufacturers than actually fixing their shitty products, I'd say no.Not like Shintel is any better with their degradation problems
>>109608313Hold on, let me guess, you're an American using cucked 120v outlets aren't ya?The PSU has to work 3 times as hard with such low voltages.
top run was bouncing off 95C tjmax for the entire run.middle run was after setting 85C tjmax.
Do you really need to overclock though? I just use expo and disable pbo.
>>109611957When I said overstock I meant pbo .. sorry bout that
>>109607209yeah it kind of got swept under the rug due to asrock being singled out and taking the public fall for dying processors but i've seen more than one report about the 9000 series am5s having a disconcertingly high failure rate as a whole.
>>109607641You've had no idea about the power supply thing being an issue because it isn't, OP is making up some schizophrenic shit that only exists in his mind and is divorced from reality.
>>109607209Stop buying piece of shit PSU and stop running them near máximum, diphat.
>>109611261Unless you’re using a 360v outlet, it’s slightly less than twice as much amperage, not 3 times. Also, twice the amperage does not mean twice the “work” the PSU has to do. It’s twice the work that the PFC alone has to do, which is rated to handle loads far higher than what a 120v input will produce. In reality, you’re getting a 1-3% loss in efficiency and nothing else by using 120v as opposed to something like 230v, there’s no evidence to suggest PSU’s don’t last as long with lower input voltages, as long as they meet the rated spec.
>>109608313If you weren't retarded you would have bought a Seasonic psu instead.
>>109612020People are probably overclocking them without knowing. Every time I say expo is an overclock, someone is surprised by this.Expo on zen 5 dials in very aggressive voltages.
>>109612275Yeah dude, I love spending $250 for the same or worse quality unit I could get from CWT for half the money.
>>109612046Whuch is kind of the problem. Gaming bioses follow what amd tells them which is to turn it on by default. Most of the performance claims of modern processors come from pbo... and whatever the intel version is.
>>109607209max overclocked models like the 9850x3d, 9950x3d and 9950x3d2 are the ones least likely to age well but more conservative parts like the 9700x will probably last a long time
NiggaJust put a bigger cooler on it, we get clones of very decent models. Replace the pasta after 4-5 years if you are worried.Also get a bigger PSU. Buy cheap buy twice, etc.
>>109612399>Replace the pasta after 4-5Or just use ptm 7950 and literally never have to detach your cooler ever again.
>>109612399>buy a bigger PSU>anon is running a r7 7700Unless he’s running a 5090 as well, going bigger is an absolute waste of money and will only lead to an increased power bill from putting your PSU far below its efficiency curve. Crazy how much terrible advice and misinformation is ITT.
>>109612432>pay Linus $100000 or get scammed on eBay Been there, done that. I just use paste again, PTM is too expensive to be worth it and you can’t even reliably source the stuff unless you’re willing to get bent over a barrel.
>>109607209>almost 2 failing power supplysthis post makes zero sense. Either your PSU shits itself and stops working, or it works. There's not supposed to be an inbetween.OP do you know anything about pcs other than what some bro-OC tryhard told you? a guy who doesn't care if he blows his one of ~10 test bench setups up with his OC bullshit?I stopped bothering with any OC stuff on AMD as you need to win the silicon lottery with your RAM and CPU to get a stable OC. The baseline settings for cpus these days is pretty great and you won't get more noticeable performance by OCing anyway, outside of your schizo delusions that is.
>>109612480Anon probably just had a dusty house, 2 failing PSU’s is hardly conclusive data. Dude got unlucky and everyone rushes to give this guy terrible advice about what he did “wrong”.
>>109612480there is absolutely a inbetween with PSU failure.Resistors gets slowly more unstable..https://pastebin.com/SDmU6uj9
>>109612399If you get a better cooler the CPU will just further overclock itself to continue having the same tempatures of 95C at least on early AM5/AM4.But also thats not the problem. the problem is that not just "are these tempature/power draws good long term" but is the constant flip flopping and unsmooth voltage changes SAFE for the CPU long term.
>>109611358i pulled the water block off and it was immediately apparent what is happening. the phasesheet has picrel very clearly imprinted on it.
>>109615702Low level programmer and SEMI electronics guy.can you explain what this means? I understand the basic concepts of electrical engineering. Ground, voltage,
>>109607641OP here... not that i haven't been replying beforehand... When i say stresses the PSU I don't mean that in a physical "Watt sense" or "Oh its sending alot of electricity that might overload it" I mean the constant changing of voltage that could damage it ... im no electrical expert but i have taken SOME things in repairing/eletrical knowledge.basically on older systems before we had PBO or expo and all of that most voltages on your motherboard to your PSU were sent via a constant smooth line. Sometimes it might change.. but it was not changing EVERY second. Modern Am4/AM5 does not work that way. CONSTANTLY a spot on your motherboard connected to the PUS goes from 1.40volts to then 1.30V to suddenly 1.10V and then even 0.9V and then back to 1.2V its not a constant still stream. Its constantly JERKING...What im saying is that this MAY stress PSU's out and may slowly be damaging them LONG term that we don't know about.its just a theory. and any electrical experts can disprove what im saying id love to hear it .
>>109616129are you retarded? it means the coverage isn't great and you need offset standoffs>Low level programmer and SEMI electronics guy.who gives a shit what you are? does that somehow make you incapable of understanding anything else?
>>109616155because i don't do physical electronics/eletrical stuff. Wait PHASESHEET... OOH.... ok ... im googling what that is... Im sorry ive never been a watercooler guy. Ive always been a fan guy because in my opinion your system will die from watercooling related issues far earlier than it would die from heat from a "not efficient enough" fan cooler. But just my thoughts
>>109616182so i don't know the whole world of phasesheets and pumps and all of that.
>>109616149The PSU does not provide 1.4V.The only voltages available are 3.3V, 12V, 5V and -12v.Anything else is created by the motherboard, or other devices.
>>109616203well you get my point.This is one of the failing points/data from my failing PSU...you can see how it constantly JUMPS AROUND and is nowhere near stable https://pastebin.com/SDmU6uj9(Normally under INTENSE LOAD the voltage level typically does not even really REACH past 1v... its in the mV
>>109616220This is not my GPU btw its my CPUS integrated GPU.. or near the CPU that isn't even used
>>109616220It's still not something the PSU is going to care about. Voltages beyond the rails changing up and down isn't anything new, that has always been a thing with computers. The PSU only cares to keep its own rails in spec. If the PSU is going to care about anything, it's how much power you're trying to draw from it and maybe there's an argument to be made that the PSU isn't going to love it when your power draw goes up really high and really fast - but this isn't something an AM4 or AM5 CPU is capable of. The nastiest examples only go up to like 230W at full load. A normal PSU shouldn't struggle to provide that much power.There's something far worse in many computers that is going to trash the PSU - nvidia graphics cards. Those are constantly spiking power draw to absurd levels like 600+W for very short periods of time while drawing like 250W on average. Nvidia has been doing this since at least the 3xxx series graphics cards so... I don't know how relevant that is. Nvidia card did break one of my PSUs, the spikes were too much and kept tripping OCP.
>>109616350>Nvidia card did break one of my PSUsMight you recall which card and which PSU?
>>109616983An old seasonic 660W platinum unit and a 3060 Ti.
>>109607789I've had those pop up only after they've changed some AMD mce stuff in the Kernel around 7.0 release. So I wonder if those are actually real and the years before that mce code change just hid them or if it's just false reports.It must have worked right in the past because I had real, low voltage caused errors for L3 cache with actual numbers instead of just a bunch of zeroes in those address fields (I forgot to set typical low idle current or whatever it's called in the BIOS after an update).OP: Your CPU gets its general supply voltage from the VRMs on your motherboard and then the CPU's internal voltage regulators provide those finely grained core voltage levels between ~0.8V to ~1.3V. The 12V rail supplying your board's VRMs can easily cope with the load shifts unless it's some dogshit quality product. And any decent board's VRMs will just lol, lmao at a 7700X.If you don't like the high temperatures then just enter the PBO settings and set a lower temperature limit. You will likely not lose any meaningful amount of performance if you set it to 80C or even 75C as I do for my 5800X. PBO will pump and pump and pump those frequencies with no care until it reaches that temperature limit and then just become a bit less aggressive, so that's why you can see junction temps reaching even 95C. That's by design because TSMC says those temperatures are still "fine".
>>109618538>old seasonic 660W platinumDo you recall the model number?>>109620581>PBO will pump and pump and pump those frequencies with no careDo you know if that affects lifespan?
>>109621997No. It was a first gen Prime or something I think. They only had one platinum PSU with that kind of rating in mid 2010s I'm pretty sure.
>>109620581Do they actually crash anything for you or are they seemingly harmless? It is/was the latter for me. >>109610336
>>109622267I don't think you're supposed to get those even if they seem harmless. I have an AM5 CPU with latest kernel and microcode and I've never seen any mces.
>>109621997>Do you know if that affects lifespan?According to the eggheads at TSMC it shouldn't, unless you increase the thermal and electrical limits beyond the defaults. AMD just doesn't guarantee a CPU with PBO actually runs stable.>>109622267It seems harmless for me. When I had that low voltage error happen a few times it would actually crash the PC some times but with these new ones it just keeps running fine for several more weeks until I actually reboot for a kernel update. I should also note these only happened when I was away from the PC and it was idling. Maybe this is related to wake up from C6.>>109622319What makes me >>109620581 think they're false warnings is that all the hex values are all zeroes. The aforementioned real, low voltage caused L3 cache errors would show some actual information (when the machine didn't actually crash right away).
I put a peerless assassin 140 on my 9800x3d and that bitch is like 70C in games, I didn't do any custom undervolting either but maybe the motherboard set it up. It's a mini-itx build too
>>109622160>one platinum PSU with that kind of rating in the mid 2010sDo you recall the amps for the +12 V rail? I checked their site, and the 660W one (Platinum Series) has 55 A on the +12 V rail. Does this picture and/or model number look familiar?>3060 TiDo you recall the AIB?
>>109607209that boost behavior you're talking about is present on both AMD and Intel systems, theres nothing wrong with it. just don't buy some cheap ass gigabyte-tier PSUs. "a strain on my PSU" nigga if the PSU cant handle 50% load for a few seconds without exploding its a dogshit PSU theres no other way to put it.
>>109622555Yes. Pretty sure it's that one. AIB is gigabyte.
>>109622651>gigabyteThe ones I found on their site: Aorus, Eagle, Elite, Gaming, Vision, and Windforce. Do any of them sound familiar?
>>109622970No. It's a dog shit cheap one though. So whichever one of those looks like cheap crap, that's the one.
Ignore meme temps from AM5, that's just one core boosting as hard as it can. These CPUs use a trivial amount of power. Whatever is killing your PSU is something about your home's specific electrical setup. I would put a UPS between your valuable electronics and your outlet ASAP.
>>109623196>It's a dog [...] cheap oneThe 2-fan ones (the rest: 3-fan): Eagle and Windforce. I'll try to post a followup (other image) reply. Do any of the boxes and/or cards look familiar? Feel free to say if you'd like a larger picture of either (or both) card.
>>109623196>>109623357>followup
>>109607209Those garbage AM5 motherboards and CPUs are not designed for long term use
>>109607209>AMD claims that this is normalyep totally normal for the 7000 series CPUs....which is why they completely changed the behaviour for 9000 seriesreminder, 95C is not running hot
>>109623404Yeah because it made retards panic and fucked with default fan settings. The temperature spikes (even though CPU power package barely changed) ticked normal users off.You can get around this by using 'fan control' software and setting a delay to how fast your fans ramp up and down but for the average user they just hear fans jump around constantly and get frustrated.
>>109623357I don't remember what it looks like, I haven't seen it in a year because I was very disgusted by it. It is most likely windforce though because eagle looks expensive.
>>109607209in all seriousness you need a fucking UPS ASAP because your power delivery is almost certainly garbage
>>109623434>windforceDo you recall if it was the regular version? Or the OC one? What about the revision (only OC: 1.0, 2.0)?
>>109607209>CPU frequency on most amd motherboards CONSTANTLY flucates to absurdly high/low values within less than a second. putting a strain on the PSU and slowly causing it to fail..holy retardyou should go grift x users with how plausible you make your shit sound
friend of mine fried his 7600X somehow, and is getting a 7700X3D to replace it. What advise should i give him? ECO mode?
>>109623515Not a chance. Regardless, I ran it at lower than normal power limit to reduce the ocp trips. It worked but not perfectly.
>>109622417Yeah mine mostly happened when idle. I tried running with typical current idle for a while and still got them. And the fatal ones are always 0x0000000000000000s.
My 9900X died after 1.5 years on an Asrock motherboard. I sent it in for RMA, and AMD provided a replacement. And I believe, AMD is responsible for this failure not the motherboard.
>>109615702mmmmmmm fucking k'.i got an old one of these that didn't come with picrel but a non adjustable mount.
>>109612432Isn't PTM7950 meant for GPU ram chip?Just use paste
>>109613710>resistors wear outkill yourself>>109612480yeah its a bot posting on behalf of israel, we saw LOTS of threads like this in 2019 when zen 2 came out, intel always pays the same shadow marketing team
Overclocking has been dead for at least 10 years. It’s all about undervolting now.
>>109607209Laptop CPUs have been running at thermal throttling 95°C for like 15 years now with no issues
>>109610154The efficiency rating has nothing to do with build quality.
>>109626092You can use it on anything that gets above 60°C at least a few times>>109612461Get some Upsiren PCM-1 form Ali Or if you don't trust that, get Würth 402150101020 or Laird Tpcm 7250Most "fake" PTM7590 is likely PCM8500 which is also alright
>>109607209AM5 CPUs and DDR5 ram seem a lot more finicky and prone to failure than am4 and ddr4.
>>109626623DDR5 RAM is easy to brick if they have LEDs, the LEDs can result in SPD ROM getting corrupted. DDR5 RAM without LEDs is much less prone to failure.
>>109607209>ARE am5/late am4 cpus even going to last 10 years?My 3700x is about 7 years old now and I don't see any reason why it wouldn't last 3 more.
>>109607209My 5800x3d has been sitting for months because i dont know what the problem is. The problem is either the MOBO or the PSU or god forbid the CPU.
>>109607209If you are worried about the CPU running 95C then disable PBO. It will 60-70C that way for 10-20% perf drop on account of not using turbo clocks any longer.My AM4 build lasted long enough for the motherboard to get faults before the CPU got any. Luckily I made the switch to AM5 before the RAM insanity, and that's probably the only reason it was worth it.
>>109623445>UPS ASAPOP here..Perhaps i was wrong about Core boosting CONSTANTLY changing the voltage levels at a unsteady level that can slowly drain the lifespan of PSUS..This seems like a better explanation. My house was built in the 1950s... However I DO have my PSU on a power strip. >>109626103I assure you im not a bot. I upgraded from Ivy bridge to AM4 in 2023. and then later in 2026 to AM5. I just thought that the extreme core boosting of modern CPUS and the "Up, down, up, down" behavior causing the voltages to erratically change alot could cause damage to PSUs over long periods of time. Just a theory i had about my PSU failure....>>109628445>>109626623>>109610336Interesting how many of you guys ALSO come in to say that you've had a failing am4/am5 motherboard/CPU or ETC similar to me. Im not blaming AMD... But is it so absurd to think that in the race to be the "fastest" cpu they have set the PBO values higher than what can be reasonably done on a VOLTAGE level...I mean for fucks sake on AM5 my cpu goes from 4.85 ghz down to 3 ghz and then 2ghz and then you get the point, it CONSTANTLY changes at a extreme rate ... Im not saying that its changing it to be TO HIGH, but rather its changing the voltages to MUCH.. idkIm not electrician of course.
>>109626103>yeah its a bot posting on behalf of israel, we saw LOTS of threads like this in 2019 when zen 2 came out, intel always pays the same shadow marketing teamAGAIN the same thing could be happening with intel to.. im not here to shit on AMD..
>>109628415Im not talking about those CPUS im talking about LATE am4 and early AM5... And when i say late AM4 i mean the higher end AM4 that was trying to squeeze every ounce of performance out.The CPU you have is exempt from this discussion and it will likely last a HELL of a long time.
>>109607209>Ive gone through almost 2 failing power supplysYou suck at picking PSUs and suck at English, I did not read the rest of your shit.GTFO.
>>109629767they were all EVGA though.
>>109612245the energy lost as heat spikes exponentially with current so doubling the amperage results in four times more heat generation.
>>109629785You're a 120v cuck who burns through the PSU FETs because of APFC strain.120v drastically increases switching losses and stress on the main switching FETs, anyone who tells you otherwise is lying to you.
>>109629724desu the 5800x3d and x3d ships in general are capped so they should actually last longer than normal cpus that can be pushed indefinitely. 9000 series x3d are different tho you can push those indefinitely too
>9850x3D reaching 95C very fast during stress test>Go to bios >Set undervolt of 30mV>Now reaches only 76C with no stability issues or performance dropWhy are Ryzens like this and why don't they default to slightly lower volts?
>>109629854ok so imagine your in a hospital ward and they are serving you dead fetsuses..You then later learn you could had just asked for crackers and they would had given it to you. Would it be your fault for not knowing that you could just request the crackers?
>>109629854for me, at least one core wont do less than -15.
>>109629950silicon lottery
>>109629854ryzen gets the shit reject bins from epycyou might get lucky and get a decent one but they tune the stock settings so that they can sell you the worst silicon possible if they have to
