Thread 105825111

>ryzen 3000 cpus degrade ftom excessive voltages and suffer instability and high power consumption
>techtubers are silent, amd denies wrongdoing
>intel rocketlake cpus degrade for similar reasons, acknowledges problem
>techtubers make endless videos about it
Hmm, definitely no bias here...

>>105825111 (OP)
> amd cpu
>use it for years no issues
>becomes unstable after 4-5 years,

>intel cpu,
>broken voltage algorithm
>lmao 1.7v on air
>dies in 4-6months of playing minecraft.

>>105825111 (OP)
Intel exec, posts on 4chan, finds no sympathy.
It’s over, man.

>>105825111 (OP)
just buy ryzen 5000. what are you? poor?

>>105825111 (OP)
My china only 3500x worked just fine.

>>105825111 (OP)
how bout you post sauce, cocksucker

>no sauce
there never fucking is

die of crotch rot, opisacumsnotingfag

>>105825219
>>105825260
the source is your own /pbcg/ threads

>>105825111 (OP)
Don't use the stock cooler.

>>105825280
>still no link
ok, op sucks cum as expected

>>105825306
not your personal spoonfeeder

>>105825437
your spoon is full of cum and exclusively enters your mouth

>>105825111 (OP)
So do Zen 3s. Not an issue, just buy another one.

>>105825280
>your
You mean or? Fucking plebbitor

>>105825610
that's why they silently replaced the 5800x sku with 5700x

Only retards overclock these days so it's a non-issue. Massive gains from overclocking are long gone.

>>105825641
what the fuck?
>run cpu on higher voltage than was designed for
>burns out
>OH THE HUMANITY

is that what cumsnorter op is saying?
does he only know that oveclocking is literally running your gear beyond design specifications?
>gains
i saw people gaining 2-5% on r3000
and i heard all it amo unts to is to keep your cpu in perma-boost
what kind of retard would void their guarantee over that?

theres an interesting thing to mention about overclocking here
overclocking losing viability is a result of superior engineering/manufacturing process
as an engineer your goal is to create a component that does its job with the least materials used
if your component has performance left on the table, thats a job that could be done better
so its only natural that overclocking doesnt make sense anymore (unless you change the variables. like running under LN)

>>105825111 (OP)
>ryzen 3000 cpus degrade
2 more weeks

I had my 3700X for nearly 5 years and before upgrading to a 5700X3D I ran some new Cinebench numbers to get current performance for the upgrade and it was within margin of the numbers that were saved from when I had built the PC.
I don't know what to tell you.

>>105825641
>>105825697
No. They degrade at factory settings because amd didn't bin them correctly

>>105825775
post sauce already, cumsnorter
all i found was issues when overclocking

>>105825111 (OP)
proof?

>>105825111 (OP)
Still maining 2200G no probs

>>105825775
>because amd didn't bin them correctly
That and also because they then allow motherboards to pump 1.488V into them on automatic settings and let them reach 150W on all core workloads, assuning they can scrape past 4 GHz without crashing.

>>105825827
>still no sauce
so i can only assume that comes from r*ddit
and r*ddit is always wrong

>>105825729
You gain literally 20% extra performance going from stock default to tuned CPU+ram on 9800x3d.

>>105825858
post sauce
lets look into where the numbers come from

>sauce
>sauce
>sauce
>sauce
the bot needs a better script

>>105825633
they replaced it with the high clocked 5800xt , my mongoloid mouth breather

>>105825880
5800x3d got replaced by the 5700x3d
5950x got replaced by the 5900xt
5800xt is the scam edition cpu to screw buyers with unstable silicon in 12 months

>>105825872
>ram latency
thats not overall performance you fukken retard
its only visible if youre gonna do alot of fragmented reads. which youll do your best to avoid in the code
please go back to /v/

>>105825877
>have assertions
>get asked for sauce
>must be a bot
no, but you inject freebased cum into your anal vein

>>105825900
You can't read? You don't know what range or mean is?

>>105825900
>>105825913
Yeah the stupid ESL can't read text or charts go figure

>>105825913
oh my god
have you been lobotomized?
are you one of the cocomelon kids?
>ram latency graph
>>b-but its actually an average

how does your reply make any sense, mongoloid?

>>105825900
>>105825920
>>105825929
>>105825872
>source is discord and reddit
hard pass

>>105825920
>group support therapy
lamao
post bussy while youre at it

>>105825933
its always the case when something retarded pops up

>>105825111 (OP)
proof?. Because I'm still yet to see any

>>105825762
That's not what degradation means.

>>105825742
>>105825531
>>105825306
>>105825260
>>105825219
Don't know about how new you are, but I distinctly remember Ryzen degredation being a hot topic on forums at the time.

>>105825900
>>105825933

are you brow perchance? do you have at least 80IQ to read something and understand it?

>>105825900
ok, now you post sauce for OPs claims

cant be reddit or discord like you said

>>105826138
>5 year old nothingburger that every techtuber was aware of and spreading awareness years ago
Let me go ahead and start INTEL RUSTS posts out of thin air you raging homosexual, that's been an issue everyone knows about and avoids.

Holy shit, what's up with corporate shilling on this board, nowhere else do threads like this pop up but on /g/.

>>105826138
>manual tuning/oc
>manual tuning/oc
>manual tuning/oc
>manual tuning/oc
...
>>105826185
>sad homonem
is the refuge of the btfod

stay mad
>>105826192
now that was unhinged
why would i proof ops claims after asking he does it?
also go back, redd*tspacer
youre clearly mentally unwell

>>105826207
>manual tuning/oc
How is it manual tuning/OC when you have Ryzen CPUs running 1.4-1.5V out of the box?

>>105826202
homofags have domination fetish
and since theyre turbo losers forte ++
the only w they can get in life is by sucking up to whom they identify as powerful: big corpo

thats also why most consumer/shill threads are lgbt-coded
why lgbt gives such a strong support to big corporations
and why they do their best to achieve managerial positions like management or hr

using lgbtgroids as corporate bootlickers is a modern, holistic approach to a complex problem all marketing departments face int the modern world
dont let your business lag behind,
call /g/anon's solutions right now and receive a 15% off your first contract!

>>105826240
if you werent a complete analphabete you would have noticed that problems arise when the voltage is static
ryzen is designed to have transient high loads, but theyre supposed to be TRANSIENT
if you maintain your cpu at higher voltages obviously its gonna burn you imbecile
>but muh 5 fps out of 200
*complete imbecile

FUD thread

>>105826252
>TRANSIENT
Cope
If you actually bother to monitor the voltage you will realize that "transient" time takes up most of the operational service life of the CPU

>>105826298
ok
post sauce

>>105826185
I saw the Reddit and Discord logos and closed it, no need to read further. Now go back

>>105825219
https://linustechtips.com/topic/1226380-ryzen-degradation-is-pretty-bad/
https://www.techpowerup.com/forums/threads/ryzen-3000-degradation-with-static-voltage-and-static-frequency.276380/
https://www.reddit.com/r/Amd/comments/8cppib/zen_cpu_degradation/
https://www.google.com/search?q=ryzen+degradation
https://community.amd.com/t5/processors/ryzen-3000-safe-voltages-and-degradation/td-p/317194

>>105826372
>https://linustechtips.com/topic/1226380-ryzen-degradation-is-pretty-bad/
>click first link
>overclocking fuckup
do i really need to check the rest?

>>105826372
>first link: "I have been overclocking"
>second link: someone asking if 4.3ghz and 1.25V static will degrade, not a degradation report
>third link: someone OCing at 1.4V
>fourth link is just a link to google
>fifth link: I also managed to tune all cores to 4.425ghz @1.325V.
GOOD
MORNING
INTJEET
SAAR

>>105826381
lol out of curiosity i also checked one of those links

>>105826388
even for jeets, comparing them to this fucking retard would be insulting

>>105826396
it's obvious it's some sort of jeet or maybe a bolivian child, instructed to just post random links off the first page of google and hope no one checks them

>>105826394
yeah, the problem is that redd*t cargo cult ran their cpus at static voltages when the components are designed to regulate themselves
and then they wanted to turn this whole debacle into and outrage to force concessions

>>105825641
>hurr it's an overclocking only issue
People have been talking about this since 2019 and 2020 here. If you do the math, you can expect Ryzen 3000 CPUs running at stock voltages to lose anywhere from 100MHz to several hundred after 3-6 years. Go ahead and see how many Ryzen 3000 CPUs can still hit their advertised clock speeds (especially on all cores) and at what voltages in the current year. For instance, if you cannot achieve 4.1 GHz all core on Ryzen 7 CPUs from this generation without issues, it has degraded.

Because surprise surprise, AMD always has a history of factory overclocking their CPUs. So in a way, yes it is an overclocking issue because AMD always factory overclocks their CPUs. Intel did not do this for the longest time, that's why their CPUs became reliable workhorses that lasted decades until recent years with no performance loss. And they did not require voltages in excess to 1.2, 1.3, or 1.4 at all.
https://desuarchive.org/g/search/text/ryzen%20degradation/page/4/
https://desuarchive.org/g/search/text/ryzen%20degredation/

>>105826402
thats typical trany behaviour tho
and nowhere does it say that a trany cannot be bolivian kek

>>105826415
>ran their cpus at static voltages when the components are designed to regulate themselves
Maybe because they noticed the CPUs kept holding themselves at nearly 1.5V in their monitoring software and wanted to correct that?

>>105826418
ok
were 5 years later
post proof since its supposed to be that plentiful, cumsnorter

>>105826429
post proof since its supposed to be that plentiful
bc all i saw was how static voltages improve vidyaslop performanx

>fud thread
intel-aviv is at again boys, move on.

>>105826418
>cites his own posts from the archive
hahahahahaha
same links and everything

we found the shill boys

>>105826443
we dont have enough public for intelaviv to bother
i think its the same bunch of 5 pisscord tranies who post various lgbt-coded shit (like shilling for intel)

>>105826444
checked
also
what a fucking retard
also also
this looks like autistic behaviour

>>105826444
>this looks like autistic behaviour
complete lack of creativity.
and marketing people tend to be creative, remarkably so, even
i think its autistic trash

>>105826444
i think the vast majority of shilling comes from autistic trash, actually
tranified, lgbtfied nocodeshitter tranoids who ritualpost advertisement like the worthless retards they are

>>105826444
>cites his own posts from the archive
Source?

>>105826503
>span of 5 years
>exact same links
nta, but we 100% gottem, autismo

>>105826510
>exact same links
Literally anyone can go into the archives and cite past posts and discussions on the matter to support the conversation. The only autistic here is you. Case in point, wishful thinking >>105826460
>>105826472

>>105826442
Newfag

>>105826534
>>105826526
lol
seething autsimo retards

>>105826551
Yes, asking for proof of something widespread that happened practically yesterday makes you a newfag

>>105825111 (OP)
i have a core 2 duo
a phenom x4
a 1st gen i5
a 2nd gen i5
a 3rd gen i5
a zen 3
and a zen 4

they all work

i think i even have an old pentium 4 lying around here that still works too

the first pcs i personally owned were athlon and athlon xp and they worked fine too

fuck you gay brandfag kill yourself

>>105826560
except it didnt happen
and youre still snorting cum, this cant be healthy
>>105826561
op is a cocksucker from redd*t

>>105825111 (OP)
>>105826418
>>105826372

the fundemental dissertation of your retardation, is that ryzen has on die voltage regulation, that steps down the vrm votage for individually for each core, aka seeing 1.5v isoftware reading in stock and pbo operation does not reflect the actual core voltage

when you set a static oc, it puts the LDO regulators into bypass mode, allowing you to damage your cpu like the giga tards that you are, all you want.

>>105826591
op managed to create a good argument for locking down bios settings

>>105826591
>>105826606
(corr.)
or rather that redd*t should exclusively be given fisher-price tier computers

>sauce
>sauce
>sauce
>sauce
>im not eating that sauce
lol

>>105826591
>oc oc oc oc oc
You realize that a lot of people set all core ratios and voltages to LOWER power consumption and voltage while keeping clock speeds in specification? Never mind the presence of "AMD overclocking' in the BIOS.

>>105826561
oh i have a haswell-E too that works

>>105826622
>no you
>mental age: 5.5
>mind the .5 or he will throw a tantrum

>>105826624
operational range of ryzen is 0.1-1.5v (0.2 according to chud gpt)
did your people set their voltages to 0.2v?

>>105826658
i ran my 5600x at 1.025 with a stock cooler

>>105826682
and? wtf do you even mean?
wtf is it about the word TRANSIENT that you dont understand?

>>105826737
transient deez nutz

>>105826856
i got a better proposition:
i will insert my cock in a transient manner in your mothers holes

>running a 3700x
>cinebench 2024 score is almost 100 points lower in mt than it was a year ago running defaults in msi bios
bros...

>>105825111 (OP)
I recently rented a server with a 3700X and it was unstable when running zfs with avx enabled, so that wasn't just an isolated issue huh?

>>105827427
There are multiple instances of someone with a 3700X having instability in /pcbg/ despite no OC

works literally same for me since day one
cope

>>105826418
>AMD always had a history of factory overclocking their CPUs
Do you know how far back?
>>105828000
Do you have links?

>>105827427
it really isn't isolated like people in here like to pretend it is
i've seen lots of anons reporting unstable/degraded 3600s and 3700x's over the years

>>105828207
post prime95 and cinebench scores

>>105828320
>NOOOO IT IS NOT RELEVANT WHAT YOU JUST TOLD I AM NOT GOING TO BELIEVE UNTIL YOU POST BENCMARK RESULTS AND THAN I WILL FIND CPU THAT SCORES HIGHER TO PROVE MYSELF RIGHT
how about you buy a new CPU instead of wasting life on 7700k ranjeet?

>>105825148
meanwhile people who have overclocked their 6700k's to 4.8ghz at 1.45v still have them running exactly the same way 8 years later

>>105828352
Haswell through Skylake was indestructible

>gets btfo
>starts posting about fairy-tales
yep im out
intelaviv is not sending good marketing team

>>105828545
sauce?

>>105828429
Do you know why?

>>105826591
Hey, Reddit knows more than AMD, chud.

>>105828337
I didn't say it wasn't relevant, I asked for cinebench and prime59 results to confirm there was no deg

>>105826624
buddy using the curve optimier lets you set an offset for each voltage regulator, if you want to save power , use that instead, you'll save more power and get better performance.

>>105828656
It doesn't matter why

>>105828352
doesnt matter, different process node, you could run 1.6v safely on conroe (65nm) , go back far enough and >3v is a common operating voltage.

>>105827427
Here's a github issue btw, where multiple people have the same issue with ryzens.
https://github.com/openzfs/zfs/issues/16452#issuecomment-3046738453

>>105832599
>some guy's 9950x also had the issue
>rma'd and the new one is much better with fewer chksum errors but they're still happening
yikes
amd and their shit binning strikes again

>>105828352
Specific builds and cpu-mobo combo required?

The AMD defense force is incredible. My own 3600 died last year at stock, and you can find plenty of even earlier, similar reports of random BSODs from people who may not even know how to OC at all.
I still just bought a newer Ryzen, because why the fuck would Intel be any better these days when absolutely everything is going to shit. There's no "good guy" here, you just have to try and avoid getting fucked too hard.

>>105825111 (OP)
>ITT A BUNCH OF AMDRONES SEETHING AND HAVING MENTAL BREAKDOWNS BECAUSE SOMEONE HAD THE UTTER AUDACITY TO SAY
amd has problems too

>>105825280
>>105825437
When YOU make a claim, YOU are supposed to post the source to support your claim, you shit eating butt hurt retard! That's how the burden of proof works!

Next time just post the source or just admit that you were lying!

>>105825111 (OP)
It's called planned obselence. They're taking a page out of the SSD/HD playbook where it degrades after 5 years or 500TB reads for no reason so you can buy another cpu the way you have to buy another hd.

>>105833990
i had a 1800x that fried itself after 8 months, i had a 1600x that died after 4 months, i had two 3900x that died. first one was its pci-express lane to the gpu that stopped working, the second was just horrible instability issues.
then i had not one, but three 5950x's that died.
meanwhile my shitty intel 6700k still going strong being used as a file server since 2016 and my 9900k has been going strong overclocked to 4.8ghz since 2019.

i now have a 14900k in my main pc but i have gone through three of them due to the intel frying bug. this third replacement has been fine though. only ever been ran on the firmware fixes / bandaids.

honestly, no one makes stable hardware. but it is frustrating how the amd community acts like lisa su shits out gold and rainbows. amd shit stinks like everyone elses shit.

AHEM, AMDRONES

>>105825111 (OP)
Yeah it was a real issue, as for why e-grifters barely talked about it or offered trash half-advice videos on it idk. The 3600Xs with their tiny included coolers, famously killed one of the two CCDs over time if you left them on stock settings.
Overall this is a non-issue on Zen 2 if you just turn off PBO (no benefit) and set vcore offsets (varies depending on SKU). You can lop off a whopping 10C on 3900Xs with a -0.075V offset + no PBO and only incur like 5% performance loss.
Trusting the stock settings from ANY board partner or CPU brand is an express ticket to fucking your shit up.
>t. Zen 2, 3 and 4 enjoyer
/thread

>>105834072
oh forgot to mention, my ryzen 2700x is still going strong since 2018 as well. my only ryzen to not kill itself.

>>105828656
My armchair hypothesis is because they are quad cores. Easier to bin. Less stuff to go wrong. You start scaling more cores, you increase chance for defects. Couple that with smaller die sizes due to smaller manufacturing process, increases the chance for defects and voltage related issues (hotspots and leakage).

>>105834077
OH NONONONONO. NOT MY RYZENINO!

>>105834082
And just to be clear, the implementations for toggling PBO were broken as fuck and often two layered or obtuse because of the board partners.
Many people still had it on, even if they thought they turned it off in one place, so yeah it's AMD's fault for prematurely shipping PBO and the board partners faults for shit BIOS implementations for controlling it.

sar, our processors work, sar. no wrong with our ryzens, sar. please sar, do not redeem intel, sar. very much redeem, ryzen, sar.

>smear campaign
So the next gen intcel cpus will be completely shit, again.

>>105834155
>any criticism of amd is a psyop and a grand conspiracy against amd

>>105834082
>famously killed one of the two CCDs over time if you left them on stock settings
****CCXs (leaving you with 3 cores)
typo

>>105834170
Don't bother fixing typos in your smear campaign. We all know what you are doing InTel Aviv.

>>105834165
Look, all I know is that software has specific warnings about Intel CPUs causing crashing and degraded performance while not warning about AMD CPUs.

>>105834186
everyone knows intel has problems. but you amdrones refuse to admit your shit sucks too

>>105826138
what subhuman language is this?

>>105834170
A CCX is not a core.

>>105834195
Perhaps because it doesn't suck as bad.

>>105834180
Yeah for sure bro, I'm totally like the rest of the mentally ill freaks/bots in this thread.
>explicitly state that I'm an AMD user in my post
>give actionable advice to avoid AMD's faults
I'm surprised you didn't call me an AMDrone, anyway I hope my post helps someone make sense of this

>>105834200
>A CCX is not a core
no shit
>picrel

>>105834220
>no shit
>>picrel
I see four cores in that CCX.
How does one CCX dying leave you with three cores?

>>105834072
>then i had not one, but three 5950x's that died.

what on earth were you doing to cause that.

>He doesn't suppress his CPU's power level to a reasonable figure
yeah I sure love doubling my power usage and screeching noise level as well as shortening my hardware's lifespan for some 11% theoretical benchmark performance score boost I won't even notice.

>>105834082
>was
Still is. Stock EXPO settings were pumping 1.3v through my modest 7600x3d on a 2025 ASUS board with updated bios. I don't think it would have lasted long being at 50° idle.

>Ryzen 3600
>4200 MHz
>1.4V
>90w
>95C
Problem?

>>105834072
Cpu erratas aside, the last, reliable hardware has got to be the 9900K to 11700K series of cpus.
I wouldn't trust anything bleeding edge, factory overvolted or AMD cpus with fancy packaging tech (chiplets, stacked dies etc.) to stay reliable after 4 years

>>105834853
How are chiplets, "fancy packaging tech"?
It's hardly different to how Intel did Core 2 Quad.

>>105834103
>>105828656
ability degraded on dual core pentiums and core 2 duo/quads as well, from early conroe onwards. maybe even pentium d penryns (I skipped them).
the overclock would be unstable after a couple months if youre very unlucky and maybe last a few years.
it's just that today's intel and nvidia are especially dogshit. not just the powerhog effect

>>105825111 (OP)
womm

>>105834085
My 1700 is fine, but the memory sucked, so I just kept the chip.

When you go past stock, don't pikachuface if you destroy your chips. It's the risk you run.

I don't want to comment on how you niggers are burning through cpus judging by these posts
Learn how to use a fucking computer

>>105834116
I can confirm that PBO is still dogshit today with unexplained behavior on my 9800x3d+asrock b650 steel legend
It really makes me want to go back to Intel if not for their lack of avx512 and shitty e-cores nobody asked for.

My 13600kf is still at 1.5v, still works :)

>>105834870
>ability degraded on dual core pentiums and core 2 duo/quads as well, from early conroe onwards. maybe even pentium d penryns
Do you know if that applies to only overclocked ones?

>>105826138
I remember people talking about whether the high voltages it would use were harmful, because they could be if run constantly with enough load, and the explanation that the algorithm only boosts that high briefly on particular cores and yadda yadda
I don't recall degradation actually happening or actual reports of it outside of overclocking though, and that screenshot proves nothing except some hysteria on kiddie forums. Certainly it wasn't widespread like the Intel degradation if it even was an issue in the first place

>>105835125
Intel degradation wasn't widespread either. It barely affected a sub 5% group of buyers but everyone thinks it happens on every single 13th and 14th gen system.

>>105834934
the guy is full of shit, i ran a x5450 at 1.45v for 6 years, never lost a mhz, conroe is even more voltage tolerant,

the killer for core 2 was the VTT aka fsb termination voltage, if you ran 1.5v vtt on a 45nm core 2 it would die quickly, same with 65nm running 1.6v vtt would kill, and people used to do that alot since fsb frequency was more important than cpu clock for games, lots of people would run 500mhz fsb with really high vtt and mch voltage, , if you were smart you would run a lower fsb and vtt and tighten up the Trd instead.

>>105825111 (OP)
I have multiple Ryzen 3000 CPUs that have been running for several years without any issue at all.
The Ryzen 3000 degradation thing was a forced meme peddled by Intel shills.

>>105835163
the 13th and 14th gen degredation is wide spread, it affected desktop and mobile chips, because of how the voltage algorithm was bugged it mainly killed chips running light workloads at high temps, every chip will die without the microcode patch, its just a matter of when.

>>105836271
its mostly bad binned i7s and i9 chips with extremely good cooling (360mm aio or better, i.e. not prebuilts) that are able to 1-2T boost endlessly without throttling that degraded

>>105836397
its all of them, intel borked the boost algorithm, the original microcode allows the 2t boost of 6ghz up to 90c , when it shouldve been 70c , and the adaptive voltage mechanism to combat vdroop, overcorrected under light loads, allowing spikes up to 1.7v to touch the cpu periodically. , aswell several mobo vendors set their values to undervolt cpus to get higher boost ,its why it took so many different patches to fix, because it was multiple issues contributing to the instability

>>105825111 (OP)
I don't care, 14900K has traumatized me from Intel, months of paranoia before i got my refund and everytime it crashed i investigated the reason with the crashdumps the reason and it was always something bonkers insane like the CPU incrementing its instruction pointer off by one resulting in obvious program flow corruption if not an instacrash since it was misaligned.
About 8 weeks in it couldn't even install windows 11 consistently since the decompression process crashed too...
Also not having AVX-512 support on their consumer chips in 2025 despite having it on your 11th gen chips is some of the greediest shit Intel could do.
>>105836397
my 14900K(amikaze) had a good aircooler on it and it degraded the fastest when playing games that put load on few cores. Overheating constantly in multicore loads is what actually made it take 1 month before crashes happened.

>>105834853
>the last, reliable hardware has got to be the 9900K to 11700K series of CPUs
Do you know why?

ryzen shills scare me

>>105834853
>to 11700K
rocket lake had high failure rates according to puget. 10th gen was the last stable consumer processor.
>>105838192
no cuck core chiplet bullshit. tried and true 14nm.