Overclocking stability

[Somerandomtechyboi]

7 hours ago, RONOTHAN## said:

There is some well binned Micron 16Gb A die that does boot and run some benchmarks at 5800, so it's not impossible. It's unlikely, but possible. The kit is also Kingston, and Kingston kits are usually either Micron or Hynix, so odds are it's more likely to be a really good Micron kit than it is to be a really bad Hynix kit, or even a really bad Samsung kit wouldn't make sense as even the worst Samsung kits do 6000MT/s. 

so i guess theres binned ones that are still more trash than samsungs but not so trash that its obvious

 

2 hours ago, ArushM said:

So I just have bad dies

yes

 

if thaiphoon isnt accurate just plug in some generic samsung hynix and micron settings you can find off the internet, whichever one works/stable means thats the ic you have

 

considering the slow clocks id assume temps arent gonna be that much of a hurdle so youll probably be fine pushing 1.5v if neccesary, and theres always the option of a ram fan if you wanna go over that

[ArushM]

6 minutes ago, Somerandomtechyboi said:

fine pushing 1.5v

What do you mean fine, I would rather keep my ram in one piece, temps are fine but my concern is more about the fact that my PC is shutting down/restarting randomly not the OC, that I can probably get on my own.

[Somerandomtechyboi]

25 minutes ago, ArushM said:

What do you mean fine, I would rather keep my ram in one piece, temps are fine but my concern is more about the fact that my PC is shutting down/restarting randomly not the OC, that I can probably get on my own.

turned all the overclocks off even turning xmp off and reset bios to default settings still has issues?

 

send pic of the psu and its wattage table, could be an underpowered psu that simply cant power your components

 

if the psu is fine then replace board cause strange issues that arent caused by unstable ram is almost always the mobo

[Levent]

37 minutes ago, Somerandomtechyboi said:

so i guess theres binned ones that are still more trash than samsungs but not so trash that its obvious

 

yes

 

if thaiphoon isnt accurate just plug in some generic samsung hynix and micron settings you can find off the internet, whichever one works/stable means thats the ic you have

 

considering the slow clocks id assume temps arent gonna be that much of a hurdle so youll probably be fine pushing 1.5v if neccesary, and theres always the option of a ram fan if you wanna go over that

I dont know what planet you live on but 1.5v is not fine unless you actively cool your rams. Stop recommending people Extreme OC. We been through this many many times.

[ArushM]

45 minutes ago, Somerandomtechyboi said:

turned all the overclocks off even turning xmp off and reset bios to default settings still has issues?

Yeah

45 minutes ago, Somerandomtechyboi said:

send pic of the psu and its wattage table, could be an underpowered psu that simply cant power your components

its a 1000W g6 by EVGA and I have no GPU, is that really it

45 minutes ago, Somerandomtechyboi said:

if the psu is fine then replace board cause strange issues that arent caused by unstable ram is almost always the mobo

Its been fine for the better part of a year, what would cause it to change all of a sudden

43 minutes ago, Levent said:

I dont know what planet you live on but 1.5v is not fine unless you actively cool your rams. Stop recommending people Extreme OC. We been through this many many times.

Thats what I thought but thanks for the second opinion

 

If it helps it only crashes when I proform an action(clicking on a youtube video, going to an instagram post etc) and never when its idle, but its also fine doing any stress test

 

Sorry if I dont respond for the next little bit but I gotta go to sleep tomorrows the last day of school

[Somerandomtechyboi]

18 minutes ago, Levent said:

I dont know what planet you live on but 1.5v is not fine unless you actively cool your rams. Stop recommending people Extreme OC. We been through this many many times.

extreme oc? thats a very low bar for extreme oc, its encroaching on high oc territory (<1.7v) but nowhere near extreme as you can still daily it given active cooling if thats even neccesary

 

if we were talking extreme oc itd be closer to 2v not a measly 1.5v and you might need to consider running subzero assuming the rams scale with cold, not exactly an oc thats dailyable even if you wanted to daily it

 

i dont even do actual extreme ocs myself as i dont have phase change or any form of subzero cooling, maybe some frequency runs past ddr3 3200 but thats about it for anything remotely close to extreme oc cause it isnt stable enough to be dailyed and im still on aircooling

 

now that ive gotten bored of frequency runs cause this w3680 only manages a rather dissapointing 3400 max (already reached with a bloomfield w3503) im just toying with high frequency stability, so far ive only managed to get 2832c10 and 3000c11 somewhat stable for triple and dual channel, not exactly extreme is it when these are technically dailyable once stabilized, doesnt even require >1.55v vtt and id only start getting uncomfortable past 1.6v though i think i should optimize vdimm for that 2832 cause 2.3v might be abit overkill for that frequency

 

6 minutes ago, ArushM said:

its a 1000W g6 by EVGA and I have no CPU, is that really it

 

6 minutes ago, ArushM said:

Its been fine for the better part of a year, what would cause it to change all of a sudden

any spare psu? maybe try rma psu since that does have a 10 year warranty

 

if psu is fine swap the board, most error prone component in a build so any unexplainable issues that arent caused by other components means its probably the board, and ive seen my fair share of issues after 50 ish broken boards that ive gone through and tried to fix from boards that refuse to turn on to random resets for no reason and also some other boards that just spontaneously combust (burning vrms) which ive had two so far

[Levent]

5 minutes ago, Somerandomtechyboi said:

extreme oc? thats a very low bar for extreme oc, its encroaching on high oc territory (<1.7v) but nowhere near extreme as you can still daily it given active cooling if thats even neccesary

 

if we were talking extreme oc itd be closer to 2v not a measly 1.5v and you might need to consider running subzero assuming the rams scale with cold, not exactly an oc thats dailyable even if you wanted to daily it

 

i dont even do actual extreme ocs myself as i dont have phase change or any form of subzero cooling, maybe some frequency runs past ddr3 3200 but thats about it for anything remotely close to extreme oc cause it isnt stable enough to be dailyed and im still on aircooling

 

now that ive gotten bored of frequency runs cause this w3680 only manages a rather dissapointing 3400 max (already reached with a bloomfield w3503) im just toying with high frequency stability, so far ive only managed to get 2832c10 and 3000c11 somewhat stable for triple and dual channel, not exactly extreme is it when these are technically dailyable once stabilized, doesnt even require >1.55v vtt and id only start getting uncomfortable past 1.6v though i think i should optimize vdimm for that 2832 cause 2.3v might be abit overkill for that frequency

Every time I call you on your bullshit you respond with an essay. Your 15 year old CPU and equally old platform does not translate into today. Stop acting like god-tier overclocker.

 

1.5v on a chip that is designed to run 1.1v is absurd and its very clear you have no clue what you are talking about.

[ArushM]

16 minutes ago, Somerandomtechyboi said:

any spare psu? maybe try rma psu since that does have a 10 year warranty

Dont have a spare but could do that

17 minutes ago, Somerandomtechyboi said:

if psu is fine swap the board, most error prone component in a build so any unexplainable issues that arent caused by other components means its probably the board, and ive seen my fair share of issues after 50 ish broken boards that ive gone through and tried to fix from boards that refuse to turn on to random resets for no reason and also some other boards that just spontaneously combust (burning vrms) which ive had two so far

I mean why would the board all of a sudden just die, I have had it for close to a year now with no issues

12 minutes ago, Levent said:

run 1.1v

Wait what? its running at 1.35V stock

[Levent]

1 minute ago, ArushM said:

Dont have a spare but could do that

I mean why would the board all of a sudden just die, I have had it for close to a year now with no issues

Wait what? its running at 1.35V stock

DDR5 DRAM chips are designed to run at 1.1v, they are binned for higher clocks. Yours is fine. AFAIK most DDR5 kits come with 1.25v XMP profiles, considering you are only using 10% more voltage than factory rated them.

[ArushM]

Right now though havent had any crashes for about 3-ish hours at 5800CL36 No other OC, IDK if that means stable or how I test this but I just want to fix the issue not the sypmtoms

[ArushM]

1 minute ago, Levent said:

DDR5 DRAM chips are designed to run at 1.1v, they are binned for higher clocks. Yours is fine. AFAIK most DDR5 kits come with 1.25v XMP profiles, considering you are only using 10% more voltage than factory rated them.

Ok thanks

[RONOTHAN##]

49 minutes ago, Levent said:

I dont know what planet you live on but 1.5v is not fine unless you actively cool your rams. Stop recommending people Extreme OC. We been through this many many times.

You could argue that 1.55V is OK for some DDR5 since Klevv ships some kits with 1.55V XMP with a lifetime warranty (specifically KD5AGUA80-80R380S and KD5KGUD80-80R380S). I don't know why they do, their kits are using the same settings as G.Skill and others use with 1.45V, but it does at least indicate that the memory shouldn't degrade at that voltage (given that Klevv is a subsidiary of SK Hynix, I figure they probably know the safe voltage values of the memory kits they make).

 

I do agree though that going much above 1.4V-1.45V on passive cooling makes no sense, you tend to scale inversely with higher voltage, and for memory chips like Samsung that OP has that you can't even find in speed bins above 1.35V, I would stick to 1.35V. 

 

32 minutes ago, ArushM said:

its a 1000W g6 by EVGA and I have no CPU, is that really it

Unless it was dying, no. The 1000W G6 is a very good PSU. Besides, even if the PSU was dying, it's more common for random shutoffs to be the issue, not crashes. 

 

1 minute ago, ArushM said:

Right now though havent had any crashes for about 3-ish hours at 5800CL36 No other OC, IDK if that means stable or how I test this but I just want to fix the issue not the sypmtoms

So wait, when are you having crashes again? Are you or are you not having them at stock settings? 

 

If you weren't crashing at stock, you're fine right now with the memory OC but have crashed before with the same/lower OC, there is something called retrain instability on Intel, where it works just fine for one reboot but crashes the next. It's very likely that's what you're running into and need to lower the memory speed to get it to be stable across reboots. 

[Somerandomtechyboi]

16 minutes ago, ArushM said:

Dont have a spare but could do that

do that first before outright swapping the board

 

16 minutes ago, ArushM said:

I mean why would the board all of a sudden just die, I have had it for close to a year now with no issues

boards are just like that, seen my fair share of boards up and dying for no reason, even just completely random shelf death after a few months where even deep cleaning wouldnt recover it

 

16 minutes ago, ArushM said:

Wait what? its running at 1.35V stock

i think you should read the context before replying otherwise you might get abit confused

 

29 minutes ago, Levent said:

Your 15 year old CPU and equally old platform does not translate into today. Stop acting like god-tier overclocker.

i dont exactly consider myself a good overclocker as ive only managed to get high frequencies, only just getting into benching shenanigans with spi for actual ram performance tuning, and i still have no experience with ddr4 nor ddr5

 

and i have no real reason to suspect that you are just projecting on me as the lack of common overclocking knowledge would suggest you arent really an overclocker

 

if i was a god tier overclocker id have some sort of subzero cooling system available and id also be able to buy a bunch of actually good hardware and not stuck on an entry level x58a ud3r and an average bin cpu, then id have tested tons and tons of ram ics obscure or not obscure, and also experience across ddr1-5, and whaddaya know i have none of that cause im broke asf and i have no interest in ddr1 atm

 

29 minutes ago, Levent said:

1.5v on a chip that is designed to run 1.1v is absurd and its very clear you have no clue what you are talking about.

uh hello its overclocking obviously youll be running higher voltages for overclocking, if you cant handle high voltage dont even bother

 

and you cant use ddr5 standard voltage as a reference for overclocking and any half decent overclocker knows that, proof is simply by testing different rams, and if you dont do the trial by fire yourself like me then look online and im sure there has to be someone thats done the trial by fire for you running a high overclock for extended periods of time

5 minutes ago, RONOTHAN## said:

You could argue that 1.55V is OK for some DDR5 since Klevv ships some kits with 1.55V XMP with a lifetime warranty (specifically KD5AGUA80-80R380S and KD5KGUD80-80R380S)

you could also just look for high rated xmp kits as a reference assuming you have the same ic as the xmp kit you ars looking at in this case hynix a die, i mean they arent gonna give lifetime warranties if the ics are gonna degrade at that voltage so if you wanna be on the safe side then these are for you not that anyone trying to be safe would even run above mild (<1.4v) voltages

 

for trial by fire on rams

my crappy hynix cfr starts degrading at >2v which is about where it stops scaling, my samsung gdie used for the above overclocks tops out at 2.3v for scaling and stability as going higher destabilizes it completely and i have not seen any degradation whatsoever even without a fan, then my shitty tinfoil samsung f die stick has no issues running at 2.46v or 2.6v and scaling to that voltage but its utterly useless as that stick wont even clock past 2600 at 2.6v, ran the thing at 2.46v for 8 hours prime95 when i had a p6t deluxe again no degradation

 

and whats the common thing with all the ics above? theyre all rated for 1.5v stock cause thats the ddr3 standard voltage aka theyre "designed" for 1.5v yet they have very different characteristics as to how they scale with voltage and how they degrade with voltage (if they even degrade at all that is)

 

this is common knowledge that any half decent overclocker would know, even novice overclockers should know this

[Levent]

5 minutes ago, Somerandomtechyboi said:

i dont exactly consider myself a good overclocker as ive only managed to get high frequencies, only just getting into benching shenanigans with spi for actual ram performance tuning, and i still have no experience with ddr4 nor ddr5

 

and i have no real reason to suspect that you are just projecting on me as the lack of common overclocking knowledge would suggest you arent really an overclocker

 

if i was a god tier overclocker id have some sort of subzero cooling system available and id also be able to buy a bunch of actually good hardware and not stuck on an entry level x58a ud3r and an average bin cpu, then id have tested tons and tons of ram ics obscure or not obscure, and also experience across ddr1-5, and whaddaya know i have none of that cause im broke asf and i have no interest in ddr1 atm

 

uh hello its overclocking obviously youll be running higher voltages for overclocking, if you cant handle high voltage dont even bother

 

You sure talk alot about things you never even once touched or experimented on.

 

We arent just talking about degradation. A chip that is designed to run at 1.1v nominal voltage or 1.25 OC profile, will come with an appropriate cooling for that.

 

You experience with EOL hardware, not current hardware. Stop talking like you have hands on experience with everything under the sun.

[Somerandomtechyboi]

Just now, Levent said:

We arent just talking about degradation. A chip that is designed to run at 1.1v nominal voltage or 1.25 OC profile, will come with an appropriate cooling for that.

actually thats a pretty good point if we are talking cooling, i mean its also common knowledge that youll need to run a fan over the rams above a certain voltage, that also varies depending on the ram ics as they all have different temperature sensitivities

 

afaik ddr4 its around the 1.45-1.55v mark and ddr5 somewhere around 1.4-1.45v but thats from reading about it

 

ddr3 is so varied and has rather limited info so i have no clue if theres even an average range for most ics like ddr4 or ddr5 cause ive already done 2 stick 2800c11 stable 2.1v with no ram fan before i sold 2 of my cfr sticks, and ive only started running a fan over my rams recently cause i simply forgot that i might need a ram fan and i point two fans at myself to get rid of mosquitos and not to sweat like mad when toying around with my testbench so most of my 2v+ 2800+ ddr3 oc shenanigans were done with no ram fan

 

temp sensitivity also scales with frequency so higher freq = more temp sensitive

 

11 minutes ago, Levent said:

You experience with EOL hardware, not current hardware. Stop talking like you have hands on experience with everything under the sun.

fair enough i usually have a disclaimer that i only have general info on ddr4 and especially ddr5 cause i dont really read up on ddr5 where ddr4 i went down the rabbit hole of looking at a bunch of ics forum posts and buildzoid videos

 

only ddr2/3 knowledge that i have which transfers to newer generations is more general stuff like cpu imcs not giving a shit about vdimm or ram ic degradation tendancies (alot of them dont really degrade and only have a hard voltage ceiling in the form of how far the ic itself scales with voltage, if they do degrade its probably when it starts diminishing in voltage scaling or long past the point of max voltage scaling)

[ArushM]

5 hours ago, RONOTHAN## said:

I do agree though that going much above 1.4V-1.45V on passive cooling makes no sense, you tend to scale inversely with higher voltage, and for memory chips like Samsung that OP has that you can't even find in speed bins above 1.35V, I would stick to 1.35V. 

Ok thanks

5 hours ago, RONOTHAN## said:

So wait, when are you having crashes again? Are you or are you not having them at stock settings? 

 

If you weren't crashing at stock, you're fine right now with the memory OC but have crashed before with the same/lower OC, there is something called retrain instability on Intel, where it works just fine for one reboot but crashes the next. It's very likely that's what you're running into and need to lower the memory speed to get it to be stable across reboots. 

I crashed once with complete stock settings, now im running 5800CL36 though and nothing else and its fine but thats relative and it is still probably prone to crashing. Its never been an issue on boots, only when actually using the computer. Seems to do it the most while scrolling though instagram and watching youtube. Its also throwing a kernal power critical error in event viewer everytime it happenes (go back to the bottom of the last page if you want to see the whole thing)

 

@Somerandomtechyboi I am not going to touch voltages because although my ram isnt running the frequency it should be, my issue is instability and crashes, not how to push ram on an unstable board

@Levent thanks for your insight, I wont raise voltages

[Somerandomtechyboi]

12 minutes ago, ArushM said:

I crashed once with complete stock settings

thats pretty worrying

you sure its completely stock? like no overclocks at all not even xmp

 

assuming the psu is fine id try detuning the cpu next see if its the cpu, just manually set all p cores to 4ghz all ecores to 3ghz and vcore to 1.1v, thats such a conservative setting that basically every cpu can run it

 

if crashes still happen then yea replace board

[ArushM]

1 minute ago, Somerandomtechyboi said:

thats pretty worrying

you sure its completely stock? like no overclocks at all not even xmp

 

assuming the psu is fine id try detuning the cpu next see if its the cpu, just manually set all p cores to 4ghz all ecores to 3ghz and vcore to 1.1v, thats such a conservative setting that basically every cpu can run it

 

if crashes still happen then yea replace board

I can run 5.5/4.5 at stock voltage I doubt its that. Im like 90% sure all settings were stock but I could be wrong Im still human.

[Somerandomtechyboi]

1 minute ago, ArushM said:

I can run 5.5/4.5 at stock voltage I doubt its that. Im like 90% sure all settings were stock but I could be wrong Im still human.

id still try it anyways just to make sure it isnt the cpu being unstable

[ArushM]

24 minutes ago, Somerandomtechyboi said:

id still try it anyways just to make sure it isnt the cpu being unstable

Ok later though, im at school, anything else?

[Somerandomtechyboi]

10 minutes ago, ArushM said:

Ok later though, im at school

welp im done with 10th grade and now i get to enjoy break for a few weeks 

alone left to contemplate my life choices

 

welp enjoy it while you still can as ive seen my elementary years go by just like that then middle school went by in a flash due to the pandemic, 2 years left till im in collage, then i probably go work a job like everyone else, and i even get to see the loved ones and family around me die off ontop of that, and i dont even have any direction in life as ive lost all of that 4 years ago, so yeah getting old is no fun =p

[ArushM]

6 minutes ago, Somerandomtechyboi said:

welp im done with 10th grade and now i get to enjoy break for a few weeks 

alone left to contemplate my life choices

 

welp enjoy it while you still can as ive seen my elementary years go by just like that then middle school went by in a flash due to the pandemic, 2 years left till im in collage, then i probably go work a job like everyone else, and i even get to see the loved ones and family around me die off ontop of that, and i dont even have any direction in life as ive lost all of that 4 years ago, so yeah getting old is no fun =p

Anything else I can do though, I really doubt any hardware is broken. More likely to be some setting just need to know which. But also crashing on stock settings isnt great so maybe it is hardware. Maybe smth broke when I was OCing. What about warranties, arent those voided when I started OC'ing?

Sorry if I sound a little off, just worried I screwed over a pretty expensive computer.

[Somerandomtechyboi]

11 minutes ago, ArushM said:

Anything else I can do though, I really doubt any hardware is broken. More likely to be some setting just need to know which. But also crashing on stock settings isnt great so maybe it is hardware. Maybe smth broke when I was OCing. What about warranties, arent those voided when I started OC'ing?

Sorry if I sound a little off, just worried I screwed over a pretty expensive computer.

no you cant really do anything to your hardware when ocing cause youll be limited by temps or the thing topping out with voltage scaling before you even start seeing degradation most of the time anyways

 

cpus will destabilize with more voltage above a certain frequency if temps are too high (inverse/negative scaling) same thing with rams and some just stop scaling after a certain voltage

 

so you dont really have an incentive to push voltages to degradation territory as youd have seen diminishing returns or outright inverse scaling first

 

there is some fiasco with intel cpus degrading at stock settings though, something to do with too high power limits or something at stock settings so chances are you may be affected by this, havent read much on it but thats about what i can remember, just look up intel cpus degrading at stock settings

 

also they have no real proof that you overclocked your hardware so voiding warranties isnt a concern cause it cant be enforced in the first place, just make sure you dont tell em that you overclocked your hardware and you should still be able to claim your warranty, besides your overclocks are so mild that nothing should have happened anyways especially on the rams cause frequency plays no part in degradation and only voltage + heat do unless frequency happens to make things run significantly hotter but that only applies to me as i play with 45nm and 32nm x58 where frequency alone is known to heat the cpus up abit

[ArushM]

20 minutes ago, Somerandomtechyboi said:

no you cant really do anything to your hardware when ocing cause youll be limited by temps or the thing topping out with voltage scaling before you even start seeing degradation most of the time anyways

 

cpus will destabilize with more voltage above a certain frequency if temps are too high (inverse/negative scaling) same thing with rams and some just stop scaling after a certain voltage

 

so you dont really have an incentive to push voltages to degradation territory as youd have seen diminishing returns or outright inverse scaling first

 

there is some fiasco with intel cpus degrading at stock settings though, something to do with too high power limits or something at stock settings so chances are you may be affected by this, havent read much on it but thats about what i can remember, just look up intel cpus degrading at stock settings

 

also they have no real proof that you overclocked your hardware so voiding warranties isnt a concern cause it cant be enforced in the first place, just make sure you dont tell em that you overclocked your hardware and you should still be able to claim your warranty, besides your overclocks are so mild that nothing should have happened anyways especially on the rams cause frequency plays no part in degradation and only voltage + heat do unless frequency happens to make things run significantly hotter but that only applies to me as i play with 45nm and 32nm x58 where frequency alone is known to heat the cpus up abit

So you think the CPU committed suicide and thats the root of all my issues, cause thats really fucking annoying. How can I test it?

[Somerandomtechyboi]

Just now, ArushM said:

So you think the CPU committed suicide and thats the root of all my issues, cause thats really fucking annoying. How can I test it?

as i mentioned before detuning the cpu

 

you wont notice the diff when browsing and stuff, games maybe on the more cpu intensive ones, and when you are away just run prime95 smallest ffts

 

crashes when doing stuff or errors/crash on prime95 even after detuning the cpu thatd lead me to belive its the rams being unstable cause afaik mobo issues shouldnt be able to cause rams or cpu to become unstable, can be further confirmed with prime95 largeffts cause if its unstable rams itll error out pretty quickly

 

if no errors on any of those then id put a final test with folding @ home cause it is surprisingly good at detecting unstable hardware, usually bsods if the hardwares unstable afaik atleast thats ive seen running f@h on overclocked setups