#overclocking

1 messages · Page 85 of 1

lavish tundra
#

no wait

#

I give up

#

idk what the die is

#

it could be mjr or ajr

#

I think its mjr

modern walrus
#

good control to mjr tom

lavish tundra
south sky
#

It's obviously ajr

#

i've said this so many times now

lavish tundra
lavish tundra
modern walrus
#

for the 3800 18-22-22-42?

south sky
#

that one is MJR

modern walrus
#

cuz I've been confusing everyone asking about two different sets of memory

#

okay got it

south sky
#

the 3800 one is MJR and the other is AJR

modern walrus
#

I'm saving this to Notepad rn

lavish tundra
#

I see

south sky
#

someone is saying c-die is ajr?

#

how

modern walrus
#

well

#

quite frankly

#

I have no clue what I'm talking about

south sky
#

ajr is c-die*

lavish tundra
modern walrus
#

Well I've done the old make an ass out of you and me sorry

south sky
#

I don't know how you'd get to that conclusion but it happens

modern walrus
#

wow no bot censor? nice

south sky
#

like thaiphoon would say hynix

modern walrus
#

BIOS says hynix

lavish tundra
#

idk what else it could mean

south sky
#

yeah because it is hynix

#

if thaiphoon reads the manufacturer it is written

modern walrus
#

I think leaving dynamic on, ignored my input

lavish tundra
#

alatron is ajr very similar to c die?

#

xmp looks about the same

south sky
#

no

#

nothing alike

#

one is decent and the other is solid

lavish tundra
south sky
#

decent

#

ajr is solid

#

very solid

#

it's basically big djr

lavish tundra
#

Ahhh nice

modern walrus
#

I'm gonna change ring ratio to 48 & see if this HWiNFO craziness goes away

#

so does that mean I should try to tighten timings?

#

or bump to 1.4v?

#

or both? lol

#

or something else?

#

nope. ring ratio at 48 still makes HWiNFO show my core clock at 30,775.7mhz

#

it has to be windows 11

#

has to be

#

So we've settled on the 3200 is C+ or B- die?

south sky
#

3200 is what?

modern walrus
#

ajr

#

i.e. better than C die but not as good as B die?

south sky
#

yes

modern walrus
#

so I can probably tighten 18-20-20-40? I doubt it'll go down to 16-18-18-38 again. Is 17-19-19-39 even a thing?

#

& what would voltage limits look like? 1.4? 1.45?

zenith palm
modern walrus
#

okay cool thank you

#

I can reduce by a factor of 1? it doesn't have to be 2?

#

or an even number

zenith palm
modern walrus
#

do you think voltage 1.4+ would help, hurt, or have no effect

zenith palm
#

Uneven cas just doesn't work on ryzen

#

Depends on the die

#

C die, don't go past 1.35

modern walrus
#

well the committee decided that it's not c die but something between c & b called ajr

#

I'd love to tell you more but I understand maybe 5% of it

#

okay well set is on intel

#

so I'll try some odd numbers

#

lol

#

right now I'm trying to figure out what's messing with HWiNFO

#

I think it's Hyper-V on windows 11

#

even tho I disabled VT-d in BIOS & uninstalled the Hyper-V windows feature, it still runs with all sorts of crap running on it

#

it's looking more & more like I can't do anything about it

zenith palm
#

Idk what the voltage scaling is like on hynix dies so I'd wait for arshia or alatron

modern walrus
#

will do thanks you

modern walrus
#

is it normal that whenever I change a setting the computer restarts then revs its fan for about 2 minutes? lol

zenith palm
#

Probs retraining the memory

modern walrus
#

it sounds like it's having a pep talk with the RAM before it runs out on the field

#

Is there a brand of RAM that generally has the most success for overclocking?

#

or does totally vary based on model & you have to meticulously search for RAM with b dies?

#

do A dies exist?

#

oh is there a pinned thing? my bad

zenith palm
modern walrus
#

Does G.Skill ever have good dies?

#

nvm that's too broad a questoin

#

This is another one of those things I wish I knew something about before I threw money at it lol

#

wow I'm looking at the Zen RAM overclocking sheet

#

it doesn't look like overclocking 32gb sticks is a very popular pass-time

#

looks like 0 people have done it

#

I assume that contraption is a computer (pinned) but what the heck lol

#

at 1.210 volts my 10900K could probably hit 50 all core. 51 I need 1.305. Not sure how tha translates exactly

#

I'm rambling

#

I need a nap

#

thanks for your help

#

Erin go... potatoes

zenith palm
modern walrus
#

I'll keep that in mind when I mess up one of these sets lol

zenith palm
#

Lol

sterile flame
#

it literally died at 3266 at 16 20 20 20 40

#

on 2T aswell

#

53 minutes in (i went to bed and let it run)

#

very clear that it wouldnt run at higher freqs if it cant run at that

#

and i cant imagine cl 18 3400 is gonna be worth it over 3200

#

ima see if it can run through stable back at 1.4 on 3200mhz

#

aka what i had it on for almost a year

#

if not, somethings wrong

sudden torrent
#

If it's still having issues at that point you need a bios reset

sterile flame
#

at what point

#

at 3266?

#

so go into bios and reset the settings?

sudden torrent
sudden torrent
sterile flame
#

oh shoot

#

it started spitting errors 30 minutes into 3200

#

something is most definietly wrong.

sterile flame
#

i just reset them, so do i start over clocking again?

#

or literally just leave it as is

sudden torrent
#

Reset to default is usually enough to force it to retrain
You can try setting the options you had before that were working and go from there.

zenith palm
zenith palm
zenith palm
sterile flame
#

id have to log into an old discord im not in anymore to see exactly what they told me to put

#

cause i def dont remember

#

other than 1.45/1.4 on the dram voltage

#

left everything on auto and am running 1.4

sterile flame
#

3466?, like the non rounded ones?

zenith palm
zenith palm
#

Some boards especially with ryzen don't like it cuz of the infinityclock

sterile flame
#

i bet he posted this in every channel

zenith palm
#

100%

sterile flame
#

id mention a mod but ion wanna annoy anyone

#

oop he gon

zenith palm
#

Jp came to the rescue

sterile flame
#

ive been running a god damn test

#

for an hour

#

and it was on 2133

#

fml

zenith palm
#

Rip

#

@sterile flame what kit of ram did you have btw ?

sterile flame
#

MFR Hynix

#

the actual name is adata xpg z1

#

2400 2x8

zenith palm
#

mjr?

sterile flame
#

nah mfr

#

yeeees its prolly going to work

#

made it to 50 minutes on 3200 16

#

it hasnt even gone over using 1.376 on the ram even though i have it set to 1.45

zenith palm
#

Uhhhhh

sterile flame
#

ayo hol up aint 1.145 soc hella high

zenith palm
#

Are you sure it's not vdroop from zentimings

zenith palm
sterile flame
#

its automatically setting it to that

zenith palm
#

Uhh

#

1.1v

#

Should do

#

1.125v if not stable enough

sterile flame
zenith palm
#

Try

sterile flame
#

ah, its at 1t but it manually enabled gdm

#

thats probably how this is working

sterile flame
#

like, anything that says vdroop

zenith palm
#

Voltage shows up as less sometimes

sterile flame
#

ok where do i look for it im confused

#

the test finished successfully.

zenith palm
sterile flame
#

it set it to 1.360 after i had it set differently

#

it keeps going back to 1.360

dull ginkgo
#

it just means actual voltage 'droops' from what you manually set it to at the VRM side

sterile flame
#

its not even just drooping it wont keep what i set at all

zenith palm
#

I hope you're not trying to set that as soc voltage

sterile flame
#

absolutely not

#

im not that slow

zenith palm
#

Just making sure cuz you were talking bout soc voltage a min ago

dull ginkgo
#

I might just point at vdroop if you set it to 1.4 and it reports as 1.36, but anyways

sterile flame
#

yeah dimmv wont stay to what i set it to

#

now its at 1.392 even though i set it to 1.38

zenith palm
sterile flame
#

when i restarted into bios twice it was 1.36

zenith palm
#

Should you be even pushing those voltages on your die

sterile flame
#

dimm volts at 1.38 is more than fine

dull ginkgo
#

trying not to add even more factors, but psu voltreg can also affect things
you know what, that shouldn't matter

sterile flame
#

set soc to 1.1

#

dim is 1.38

#

i think

#

its running at 1.392 but w/e

#

lets see if this works

dull ginkgo
#

what exact sticks do you have again..?

sterile flame
#

hynix mfr adata xpg z1 2x8gb

dull ginkgo
#

sad

sterile flame
#

chips were free

#

16gb of ram that can comfortably oc to atleast 3200, for free

#

is a blessing

sterile flame
#

passed the test with 1.39v 1.1soc

#

3200 16 18 18 18 18 36

sudden torrent
#

Not bad for a free 2400 kit

sterile flame
#

seems as if its passing the sse test too

#

i could run anta but i see no point

#

its clearly stable kekw

#

not to mention i did all this alr like a year ago for 3200

zenith palm
sterile flame
#

2 occt tests came up stable

#

no reason anta which ive alr run at this speed and cl before wouldnt run fine too

zenith palm
#

Anta better at detecting errors imo but up to you

modern walrus
#

still at it

unknown.png
sterile flame
#

wth is ur dram frequency

#

how did that boot

#

wth is going on in this screenshot

#

how is ur ram not dead

#

what

south sky
modern walrus
#

shows correct on Task manager tho

#

will validate now one second

#

file corrupted

south sky
#

tm just reads targetted

modern walrus
#

strange... it's at 20

#

I increased clock to 3700

#

16-20-20-38 stable at 3600 after 8 hours of TM5

sterile flame
#

8 hours 0_0

#

you mean you ran several tests?
or it took 8 hours?

modern walrus
#

I left it running while I slept

sterile flame
#

is redoing my cpu oc going to effect my ram oc

#

like could that make it unstable

#

what is the highest recommended voltage for zen 1

modern walrus
#

I think you're supposed to do that before RAM but I'd wait for someone more knowledgeable

sterile flame
#

4.025ghz spat out nothing but whea errors all the way up to 1.3875v

#

idk if this is meant to be on 1.4 so i didnt test that

sterile flame
#

now its freezing like it was before

#

uuuughhhh

lavish tundra
#

Which cpu u have

sterile flame
#

3000G

#

zen 1

#

@lavish tundra

sudden torrent
# sterile flame 3000G

Zen 1 is good for 24/7 overclock at 1.35v according to AMD. Plenty of people go a little higher than that, around 1.4v or more, with no problems or obvious degradation.
Zen 1 is 14nm, so it handles voltage better than the newer, smaller processes.

sterile flame
#

so go for 1.4 then?

#

i plan to mine crypto ( with the gpu doe)

#

i shouldve gotten my cpu oc stable before ram

#

i think lowering ram made the cpu oc stable

#

even though before i reoc'd the cpu the ram oc was perfectly stable

tall pelican
#

if you're gpu mining, you dont need to oc a cpu

sterile flame
#

i still play to play games just mining part time

#

and also software compilation when im not mining

#

but i def gotta mine cause im too broke to aquire the money i need for a new cpu on my own 😔

#

3000mhz cl 16 1.38 with cpu 4.0 1.375 is stable

#

I SPOKE 5 FLIPPING MINUTES TOO SOON

#

THE SYSTEM CRASHED AT 5 MINUTES LEFT

#

im heated.

sudden torrent
#

1.4v is probably OK, at the very least it'll last until you can upgrade

sterile flame
#

if it crashed at 5 minutes left thats not stable at all no?

sudden torrent
#

If it crashed at all it's not stable

sterile flame
#

5 minute early celebration 😭

#

i have cpu vddp at 1.050

#

should i up that

sudden torrent
#

Nah

#

Keep in mind also, the voltage you set in bios isn't the same as the voltage under load. It droops a bit.

sterile flame
#

ik

sudden torrent
#

Do you have load line calibration on?

sterile flame
#

what what

#

do i have what

#

i just finished booting back into windows smh

#

but what do u mean

sudden torrent
#

LLC keeps the voltage higher under load, helps with stability

#

I keep mine on level 3 or 4, too high can hurt more than it helps

sterile flame
#

should i restart and enable it

proven canopy
#

There's always some level of LLC. The "levels" will be different depending on board/brand.

sudden torrent
#

Here comes an expert lol

proven canopy
#

If you're running an 3000g, no reason to fiddle with llc

sudden torrent
#

Not even when approaching 1.4v?

proven canopy
#

Idk, see how voltage droops between idle and load. Though not every board has an accurate sensor, asus die sense or gigabyte's vr_out work well.

modern walrus
#

heck. I don't know if 2 27" monitors is gonna fit here now that I actually have them

#

Does it make sense that I can clock 3200mhz memory up to 3700 by changing timings from 16-18-18-38 to 16-20-20-40?

#

cuz it's an hour into Testmem5 & still error-free

#

then again I still don't know how much of an overall improvement 3700mhz at 16-20-20-40 is compared to 3200 at 16-18-18-38

#

I guess I'll see what the clowns at reddit have said lol

lost crow
#

I just upgraded to a ryzen 5 5600x anyone know the voltage max is before premature failure could happen?

modern walrus
#

The jury is still out on that but the general sense I've gotten from this discord is 1.2-1.3 under load

#

you'll see 1.4+ at idle but don't worry about it

#

Are you overclocking in BIOS or Ryzen Master or what?

lost crow
#

Normally ryzen master but msi has a good bios just gotta learn it

modern walrus
#

Plus that also begs the question of what is "under load?" Jury is out on that too. For example, I'll hit 1.281-1.294 during CB R23 but on Prime95 I won't see voltages higher than about 1.2

lost crow
#

And also cool I've been over volting

modern walrus
#

I dunno if MSI is the same as my ASUS, but my ASUS has basically redundant Overclocking features

#

The motherboard version does not include Curve Optimizer

#

Curve Optimizer is the way to go for sure over all core for most things

lost crow
#

Msi has auto overclock I just don't trust it after what was investigated with all the manufacturers

modern walrus
#

because core optimizer will set all core clocks as high as it can while still maintaining your ability to boost 5000+ on single core

lost crow
#

What?

modern walrus
#

Ya plus you won't see the same level of performance from Auto Overclock

lost crow
#

Dang

modern walrus
#

for mine, I have to go to under the advanced tab then scroll down & I'll see "AMD Overclocking"

#

it'll make you go through a disclaimer then after you hit accept you want to go down to Precision Boost Overdrive (PBO)

lost crow
#

Sounds familiar

modern walrus
#

Set it to advanced then you can either play with TDC/EDC/PPT on your own or just hit motherboard

#

in my case, motherboard meant really high amp levels

#

which ultimately, the higher the amps the better for overall gains, but that also means more heat

#

do you have HWiNFO?

#

you should look there to determine core preference

lost crow
#

I don't think so

modern walrus
#

okay well download HWiNFO (free easy to use nice to have)

#

then look for this:

#

You see at the top where it says Core [x] Clock then in parenthesis pref #[x]?

#

pref #1/1 is your "best" core

lavish tundra
lost crow
#

Ok it makes sense

lost crow
modern walrus
#

Once you're in BIOS & after you set the TDC/EDC/PPT limits (or just left them to "motherboard") go to curve optimizer below it & set your top 3 (since 5600x is 6 core) to -25 and the other three to -15

lavish tundra
#

The max safe voltage is different for each cpu

modern walrus
#

Leave scalar to auto or 1x

lavish tundra
#

there is no such limit thats global to all cpus

modern walrus
#

don't mess with the thermal limits either

lavish tundra
lost crow
#

I'm aware 1.374v was max for the 2600

#

Ah ok

modern walrus
#

and once you've done that restart your computer & use Cinnebench R23 or something to test your success

#

btw you may want to get HWiNFO & CB R23 and do a baseline test to see your improvements

lost crow
#

Obviously

lavish tundra
modern walrus
#

once you've seen your improvements you can go back to Curve Optimizer and increase the -25s or -15s

#

more negative the better for higher clocks

#

i.e. -20 is "better" than -15

lost crow
#

Ok

modern walrus
#

oh sorry one more thing

lost crow
#

Hmmm?

lavish tundra
#

Uhh ggood ur method is needlessly long

modern walrus
#

another option in BIOS will be to set the boost increase. You can start at +200 but that'll take away a bit from your All core clocks

lavish tundra
#

fit test the cpu and go from there

modern walrus
#

then I defer to you because you are more knowledgeable

lavish tundra
#

so you have baseline values

#

the curve optimiser part is good

#

just time required can be cut by establishing a baseline

lost crow
#

I used to mess with numbers and get different results but different architecture didn't want to blowup a new cpu

modern walrus
#

Well don't ask me how to do things timely with this because I've spend at least like 50 hours on it by now lol

lavish tundra
#

set everything to auto in bios and download prime95 and hwinfo

#

run prime95 small ffts while observing voltage in hwinfo

#

wait for like 15-20 mins for it to stabilise

lost crow
#

I'd rather more time in it so less things can go wrong

lavish tundra
#

actually fit testing and moving from there is safer

lost crow
#

Ok

lost crow
lavish tundra
#

asrock bios might be better than msi for OC

lost crow
#

Not when it's lock off by asrock

modern walrus
#

how long do I have to testmem5 before I can call it good enough?

#

not for permanent but before I go back to BIOS & tighten timings again?

lavish tundra
#

should take 1.5hrs or so

modern walrus
#

oh it's at 1:30 & still moving

lavish tundra
#

Ive had like no errors and then 5-10 errors last 5 mins

modern walrus
#

When I set it while I slept earlier it went 6 hours

lavish tundra
#

What speed is ur ram running?

modern walrus
#

3700mhz rn

lavish tundra
#

I doubt that if extreme anta is taking that long

modern walrus
#

was 3600 when it went for 6

lavish tundra
#

open up cpuz or something and verify

modern walrus
#

It says Customize: Extreme1 @wooden thicketa777

lavish tundra
lavish tundra
modern walrus
#

my HWiNFO is all messed up. The only place I can get an actual correct read of memory speed is task manager

#

I've been all over the Aorus discord trying to get an answer

#

and the clowns from microsoft ran sfc and dism then told me to go to the community

#

like bro what do they pay you for

lavish tundra
#

did you try a reinstall?

modern walrus
#

oh ya

lavish tundra
#

or a bios update?

modern walrus
#

I've reinstalled & tried 3 different versions

#

I updated from F2 to F4

#

CPU-Z shows the same nonsense

lavish tundra
#

what mobo u got actually?

modern walrus
#

actually the only thing besides task manager that shows the correct speeds is TM5 which says 3696 mhz

#

Aorus Z590 Ultra

#

Yes it's the latest on the website

#

and this problem existed before I started messing with RAM with just XMP on

lavish tundra
#

try gigabyte siv

modern walrus
#

the problem is I don't know if the issue started with the windows 11 update to build 22000 or the BIOS update because I did both within like an hour of each other

#

Well that's another strange thing

#

when I open App Center and double click SIV nothing happens

#

I've reinstalled that too

lavish tundra
#

Hmmm

#

can you roll back the windows update?

modern walrus
#

the little disc looking thing spins next to my curser but nothing opens

#

hmm ya I guess I didn't try that yet

#

I'll see now

#

other notable thing is when I open task manager, HWiNFO keeps popping up then going away ever 2-3 seconds

#

oh ya that and the Intel Driver and Support thing is broken. It'll say error when I try to run it in Edge or Edge Canary

#

It says under Recovery Options "Go back This option is no longer available on this PC"

#

so I don't think I can revert to an earlier build

#

it doesn't really affect usability much it's just annoying really

#

I might just hold out until the next windows 11 build is released & see if that helps

lavish tundra
#

seems like you need to do a fresh windows install

#

not just update

modern walrus
#

As in system reset or as in make a bootable flash drive?

#

I don't really mind re-installing. it doesn't matter much since everything is on OneDrive. Installing the programs again & the time involved are really the only issues

lavish tundra
#

just a heads up btw

modern walrus
#

I'm not sure you can even make a bootable windows 11 install thing but maybe I should just run 10 until the actual release happens

#

roll back overclock settings/ lol

lavish tundra
#

RAM OC can cause severe windows corruptions

modern walrus
#

oh ya I rememer you saying to dedicate a drive to testing

lavish tundra
#

Yeah

#

If you have only one ssd you can clone it before ram OC and then restore after ur done

modern walrus
#

corruptions sfc & dism wouldn't detect?

lavish tundra
#

sfc and dism isnt foolproof

#

its actually very flawed

modern walrus
#

the only reason I'm slightly hesitant to go through the reinstall mission is because the issue was present prior to RAM oc

#

I mean it'll probably fix it I'm sure

#

but I'm hoping it can be fixed with less drastic measures

#

then again I haven't reinstalled in awhile

lavish tundra
modern walrus
#

I didn't even reinstall when I swapped motherboards

lavish tundra
modern walrus
#

that was in like late May & without incident until now I guess

lavish tundra
#

well windows is wierd

modern walrus
#

ya it's annoying. core clocks at 30k? llc at 29k? memory at 9k? HWiNFO is basically useless now except to tell me timings and temperatures

#

& voltage...

#

okay it's not useless but ya, again, it's just annoying

#

I'm thinking it has something to do with Windows 11 using Hyper-V for so much stuff

#

Maybe I should start with loading optimized defaults on the motherboard & go from there

#

to like isolate the problem

#

see if it's something on the motherboard or a windows issue

#

or maybe can run a bootable version of linux to check

#

like see if HWiNFO gives the same nonsense on there... does HWiNFO work on Linux? lol

#

Looking through everything on HWiNFO the only bad readings are anything in mhz

#

well not including GPU

lavish tundra
modern walrus
#

heck

#

It has to be some sort of error reading the system bus...at least I think I dunno nvm I'm probably just making stuff up

#

I'll play with some memory timings. That should break it more

#

I'm gonna drop tRAS from 39 to 38 & see what blows up

#

well we have POST that's something

#

pretty sure my base clock is reading 603mhz now

#

bus clock rather

sterile flame
#

sigh i dont really know

#

how long this is going to take but im getting irritated

#

im boutta set game boost and xmp and call it a day

#

3 days of just letting my pc sit and stress test almost

modern walrus
#

lol boy do I know that feeling

#

save your user profile in bios then revert back to something you know works fine

#

and get back to it when you're not sick of its crap anymore

#

I don't care what anyone says that's funny

#

Holy crap I did it

#

bus is 100 again

#

hyper-v has returned to the darkness from which it emerged

#

my clocks make sense

#

joy to the world

#

Zotac

#

oh cool now TM5 won't run fantastic lol

eager obsidian
#

how do i overclock my gpu and my intel cpu?

zenith palm
#

Your gpu you can use msi afterburner and heaven benchmark or timespy to stress test, for an intel cpu you need to make sure you have a cpu with a k at the end a z series board

#

Although you can remove power limits if you have a locked cpu which is kinda like overclocking it

#

With intel XTU

tall pelican
#

please never use xtu unless you're on a laptop that doesnt support throttlestop

zenith palm
#

Noted, just thought xtu was a tad more user friendly from what I've heard

zenith palm
tall pelican
#

it has access to full VID table

#

say you set 5.0 and forget to set voltage because its in a different menu, you're looking at 1.5v+

south sky
#

XTU is great, just not as good as bios

south sky
#

Or anything else

tall pelican
#

xtu makes it easy by putting apply right there

south sky
#

All software oc things do

#

The vid tables usually don't go that high either

#

Most go to 1.6-1.65 I think under full load 100C with spec ac_ll? At least on high end chips

hard goblet
# modern walrus

my bus clock also like that lower than 100 xD .. in mobo its set to 100 tho .... even tho cpu spectrum spread is off still same.. other reddit and forum post says about hyper v.. and svm ,,, im lazy to uninstall my wsl2.. xD

modern walrus
modern walrus
hard goblet
modern walrus
#

I have two & confuse myself constantly because of it

#

I'll check to see if Hyper-v is on on this computer

hard goblet
#

leme know if u make 100mhz stable

modern walrus
#

okay I will. it says hyper-v is on in run>msinfo32

hard goblet
#

disable svm too

modern walrus
#

Hyper-V was somehow messing up HWiNFO on my 10900K computer

#

even tho it wasn't even a windows feature I supposedly had installed

#

I was able to make it stop running with this command: bcdedit /set hypervisorlaunchtype off

#

dunno if I'd recommend doing it tho. I haven't done it on this computer. Isn't virtualization necessary for programs & security features?

hard goblet
#

or the windows sandbox

#

wsl

modern walrus
#

oh okay well even with the box unchecked in "Turn additional windows features on and off," it was still running

#

as a matter of fact, even with VT-d turned off in BIOS it was still running

#

but that command in elevated cmd finally made it stop

#

I just checked on this computer and Hyper-V was enabled so I'm removing it

#

When that's done I'll restart then enter the command & let you know progress

modern walrus
# hard goblet nope unless u want vm

okay good and possibly bad news. Good news: my bus clock now shows 100mhz on HWiNFO. Potentially bad news: in msinfo32, it says virtualization-based security not enabled.

hard goblet
#

nice\

modern walrus
hard goblet
#

can u try if i can leave svm on xD

modern walrus
#

SVM is on in BIOS

#

I didn't touch it

hard goblet
#

cool cool

#

leme try

#

so just this right bcdedit /set hypervisorlaunchtype off

#

?

modern walrus
#

you have to restart for it to take effect

#

yes in command prompt ran as admin

#

I assume "bcdedit /set hypervisorlaunchtype on" would put it back on anyway right? lol

hard goblet
#

aey

#

100mhz now

modern walrus
#

nice

#

I dunno at what cost

#

or to what gain exactly but it's there

hard goblet
#

i will put that as a note xD

if bus speed decreases like 98-99.8 even tho cpu spread is off or on ,,, might be related to hyper v

try to disable it via cmd as admin "bcdedit /set hypervisorlaunchtype off"

if u want it back just do "bcdedit /set hypervisorlaunchtype auto"

hard goblet
modern walrus
#

oh wow my first time actually contributing information instead of taking it

#

lol

hard goblet
modern walrus
#

That's what I was reading

#

but if you try to increase it in BIOS

#

you might get away with like .1, .2, .3

#

but this guy was telling me it affects PCIe connections somehow so I just left it alone

hard goblet
#

check this out

3600.png
#

my 3600mhz is 3570

modern walrus
#

hey I have that exact same motherboard on this computer

#

soo did that command break it?

hard goblet
#

i will retest aida now

hard goblet
#

xD

modern walrus
#

what does HWiNFO say?

#

mine shows memory clock at 1900mhz which x2 is 3800 like it's supposed to be

hard goblet
#

the security part is for core isolation

#

for randsomware

#

related

modern walrus
#

I see. Windows Security still shows Core Isolation tho

#

if you click on Core isolation under Device Security, do you have Memory integrity enabled?

#

That can slow down your RAM supposedly

hard goblet
#

enabled but not working i guess

#

core-isolation is needed for windows sandbox to work

modern walrus
#

When you click on "Core isolation details"

hard goblet
#

its a prebuilt vm u can just pull off to test certain app

#

once u close it it destroys data

#

so its always fresh

#

brb benchmarking aida

modern walrus
#

That's the thing with all this that worries me. Like how much am I compromising security by disabling hyper-v & Memory integrity?

zenith palm
modern walrus
#

well it's not the full 100 but I doubt it makes much difference

#

mine was down to 98.9 the other day

#

From what I read (take that with a grain of salt) frequency = bus clock x ratio

hard goblet
#

see the ram

#

the clock of the ram

modern walrus
#

so if you have a static ratio you'd have a lower frequency with a lower bus clock

hard goblet
#

it affects all the bus

modern walrus
#

hey 35mhz ain't bad haha

hard goblet
#

sheesh xD

#

pcie

#

cpu

#

ram

hard goblet
#

sometimes it goes even lower like 95mhz

modern walrus
#

Yes & for all I know it's been lower, I just haven't paid attention until recently

zenith palm
hard goblet
#

if those number keeps flactuating..

#

it affects the cpu and ram speed too

#

🙂

modern walrus
#

which times 46 times however many cores or whatever is a fairly substantial difference

#

which is why I wanted to try to up bus clock but after trying 102 & having to CMOS reset I gave up that crusade

hard goblet
#

the lower it gets

#

the lower clock u gets

#

xD

#

pcie mem and cpu

zenith palm
#

so i should run bcdedit /set hypervisorlaunchtype on

modern walrus
#

I stumbled over like a 2006 forum where people were saying they upped bus clock for overclocks

zenith palm
modern walrus
#

but that's ancient

hard goblet
#

even if i stress it.. the maximum it can make me go is 99.2 xD and it even dips at 90

#

now we confirm is hyper v doing

#

im glad that so easy xD

zenith palm
modern walrus
#

did you run as admin?

zenith palm
#

yup

modern walrus
#

noo

#

it's off

#

not on

#

lol

#

I was saying that command might undo the changes

#

bcdedit /set hypervisorlaunchtype off

zenith palm
#

what so tomie pinned wrong command :o

hard goblet
#

"bcdedit /set hypervisorlaunchtype off" is good because u dont need to disable hyperv

zenith palm
#

that worked

hard goblet
#

u just disabling it

#

so if u want to use vm stuff

#

u can enable it

#

but if u are overclocking

#

disable it

modern walrus
#

I haven't tried enabling it for the record haha

zenith palm
#

still

hard goblet
#

so fsb is 100mhz

modern walrus
#

restart

zenith palm
#

do i needa restart or something?

#

ahh

#

Lemme try restart then

hard goblet
#

when u disable it

#

i google this.. this should look like this

download.png
zenith palm
modern walrus
#

That's what mine said the operation completed successfully

hard goblet
zenith palm
#

what that

hard goblet
#

i cant explain xD its for emi stuff

zenith palm
#

might not have applied cuz my windows boot is borked tho needa reinstall

hard goblet
#

when everything is fine

hard goblet
# zenith palm what that 
Spread spectrum clocking works by continuously modulating the clock signal around a particular frequency. This “spreads out” the power output and “flattens” the spikes of signal waveform, keeping them below the FCC limit. The MCLK Spread Spectrum BIOS feature controls spread spectrum clocking of the memory bus.
#

so it does mess with bus speed too

modern walrus
#

I wish BIOS had a "disable all power saving stuff" option

hard goblet
#

the only thing i disable in my mobo is

#

fast boot

#

because that messes with driver state

modern walrus
#

it does? gotta disable that then

zenith palm
#

I see

modern walrus
#

is that why my darn Thermaltake fan controllers don't always connect correctly?

#

or a possibility of why

#

at least

hard goblet
#

yeah also it makes my dram led turns on xD maybe its my ram thing xD against that option

#

i have to remove 1 ram and disable that xD

#

then plug the ram again

#

tadah it works

modern walrus
#

lol I'm gonna go disable fast boot

hard goblet
#

if u dont have that problem u can leave it i guess xD

#

Fast Boot and Fast Start Up is auto disable for me xD idont want them

#

i want my full mem test at boot and complete shutdown

modern walrus
#

Where is fast boot?

#

you have the same board so it's easier to ask you than keep looking lol

hard goblet
#

its on your mobo

modern walrus
#

Yes I'm here in bios

#

Slow typing on my phone

hard goblet
#

in some cases ( not all ) Fast Boot causes Black screen at login Screen.. unable to access bios menu.. dram problem

#

but again if u have no problem leave as it is xD

#

its on boot section

modern walrus
#

okay I disabled it

hard goblet
#

after making bus speed stable

#

latency is much lower now

modern walrus
#

well I'm glad I was able to be useful

modern walrus
hard goblet
#

heres the real one

unknown.png
#

not so noticeable

#

but my cpu clock is hitting higher now

modern walrus
#

that read speed is way up

#

amazing .8mhz makes such a difference

hard goblet
#

cant really really on ram speed xD sometimes they have margin error xD

#

where they change each test

hard goblet
modern walrus
#

I didn't see an increase in clocks running CB but the score was higher

hard goblet
#

maybe a few points increase xD

modern walrus
#

I think 21731 v. 22395

hard goblet
#

nice

modern walrus
#

which is fairly substantial. I still can't beat my all core 47 of 23193 with any settings on PBO but oh well

hard goblet
#

my old score is low 8749 xD r23 benchmark multi thread

modern walrus
#

If you use Curve Optimizer, did you put any voltage offset in BIOS?

#

Why is it so low?

hard goblet
#

well i have r5 only

#

r5 3600

modern walrus
#

I use R23 always

#

oh derp

#

I'm like I didn't know there was a Cinebench R5

#

sorry I'm still a noob here

#

well here & in general lol

zenith palm
modern walrus
#

@zenith palm is right it should be "off" not "on"

unknown.png
clever epoch
#

Curious, did you try offset 0? Wonder if that would change the behavior

clever epoch
#

Good info about spread spectrum and hyper-V 👍

hard goblet
#

I edit it

zenith palm
#

Thank

zenith palm
hard goblet
#

My cpu spread is on auto so i dont know if its on or off. But my bus speed dont change so probably off

hard goblet
zenith palm
hard goblet
#

I see

modern walrus
#

oh good luck

#

I'm nearing completion of downloading all my former crap & installing all the programs

modern walrus
#

tho I guess I just relocated knowledge I mooched off of someone else didn't I?

#

oh man...Outlook... I forgot about Outlook. That's gonna be another mission

sudden torrent
clever epoch
#

f

faint tangle
modern walrus
#

Haha true

#

This stuff sure is discouraging. Just crashed out of nowhere. Aren't there different levels of idle, C-states? Can those be disabled? Is that a bad idea?

sterile flame
#

Bro
Is it even possible to do an overclock on a Ryzen chip
And not have it be an all-core OC
Like I'm trying to overclock my r5 3600 (pointless I know but cinebench go brrrrrr)
I can get it to run all cores at 4.6 without hitting the temperature max and without overvolting it
The chip actually runs at 4.1 base 4.2 boost on its own because of how Ryzen CPUs overclock themselves if there's headroom
But I'm trying to make it run at 4.0 base 4.5 boost and for some reason I can only apply an OC to all cores at all times

sterile flame
#

If it can run at 4.6ghz all cores
And it runs at 4.1ghz base on its own
Why in the world can't I get it to run at 4.0 base 4.5 boost

clever epoch
#

uhhh... what voltage are you using to run 4.6 all core on a 3600?

#

because you either have probably the best bin in the world, or you're going to kill it in a month of load

#

Also you can't really set a specific base or boost, you just have to let the PBO algorithm handle it. You can nudge it by setting PPT/EDC/TDC

#

unfortunately no curve optimizer for zen 2

sterile flame
#

It was peaking at around 1.45v which I know is a bit high and I have since brought it back to stock
But at stock it still runs at 4.1 base

clever epoch
#

"a bit high"

#

bro, 1.45V is going to kill your 3600 if you load it

#

PBO can load up to 1.5V because it will do it for light, bursty loads. Setting above 1.3V on an AMD N7 chip is going to degrade it

sterile flame
#

That was at peak under 100% load during a 45 minute aida64 stress test run. Average was 1.37

clever epoch
#

yeah that's way too much

sterile flame
#

Everything I've seen said keep it under 1.35v it runs much lower at stock

clever epoch
#

what, you set 4600 all core and left voltage on auto?

#

1.35V with a super droopy LLC may work if you don't care about keeping the chip more than a couple years

sterile flame
#

I set it to 1.35 but I'm assuming it overruled the limit I set in favor of frequency

clever epoch
#

I wouldn't go higher than 1.3V set, with moderate LLC

sterile flame
#

I think my motherboards bios is just broken because I've had other little issues like that too

#

Like I set the frequency target to 4.6 and the voltage to 1.35, thinking it would aim for a frequency of 4.6 but lock the voltage at 1.35, but instead it locked the frequency it 4.6 and aimed for a voltage of 1.35.

#

This is the first version of my motherboard too. I wonder if that's why there's a second version

clever epoch
#

I think the first thing you need to do is reset your expectations

#

The only way you're really going to OC your 3600 is either using advanced PBO, or setting static clock/voltage (much less than 4.6)

sterile flame
#

Would you mind giving me a tl;dr on pbo?
Cause all I know about it is that I can't seem to figure out what it does. I tried literally every setting while I was messing with the overclock trying to get it to actually boost and it ran exactly the same every time.
Also I don't see why it should be unrealistic to run a chip at 4.0 base 4.5 boost at 1.35v when it runs at 4.1 base 4.2 boost at 1.2v under load

clever epoch
#

It's unrealistic because you don't have that much control over the boost algorithm.

#

You have control over 3 parameters for PBO, and the chip will go for maximum clock speed while staying within those 3 parameters (as well as other internal parameters that you can't really control)

#

you have PPT - Total Power
EDC - Time averaged current
TDC - Peak Current

#

You want to set PPT to the maximum amount you can cool without running into the thermal limit (90C); that will depend on your cooling setup. For a 3600 that will probably be about 130W.

Doesn't matter too much for the 3600 since it's highly unlikely you'll actually run into the power/cooling limit unless you have terrible cooling

#

EDC will also depend a bit on cooling. You want to increase EDC and TDC by 10A (setting both to the same value) and benchmark on each step, to see what TDC/EDC gives you the best performance

sterile flame
#

That's stupid
Ryzen is stupid
Can you set base and boost frequencies on Intel?

#

Lol

clever epoch
#

once you find the optimal EDC, you then mess around with TDC. You can increase TDC to see if you get more perf

#

no, you cannot

#

You can either set a power limit (like Ryzen but much more limited) or set static clock/voltage

#

I'm not really sure why you're so upset by this anyways. PBO by default already extracts like 95% of your processor's performance.

#

You can write your own software that sets the CPU to 4 GHz static under load, and then switch to PBO (where you get higher boost clocks) for low loads

sterile flame
#

Just because it's annoying to me that I have temperature wattage and voltage headroom, but I only gain 100mhz boosting, cause it may as well not be boosting at all if it's at 4.1 base

clever epoch
#

PBO gives you control over the temperature and wattage headroom

#

TDC/EDC will also help you extract a bit of the voltage headroom

#

really, all of the voltage headroom

#

the SMU has a better idea of degradation than you do, and will adjust the voltage accordingly

sterile flame
#

Sorry, I'm not quite understanding, will tdc/edc just raise the clocks until it hits the target voltage? Cause I can just do that the easy way, it runs at 4.3ish ghz all cores at 1.35

#

Cause if tdc/edc can make it run an OC and still boost then I'm about learn it very intimately

clever epoch
#

You don't want 1.35V. you want 1.3V under load, max

#

but with PBO, you don't have to worry about it

#

and TDC/EDC adjusts PBO parameters, and PBO does automatic boosting

#

I recommend re-reading what I posted about the PBO parameters

south sky
#

1.3 is too high

short blade
#

@lavish tundra @dull ginkgo fastest tm5 run yet

Untitled.png
lavish tundra
#

I call hacks

short blade
#

i've decided that my pc rebooting with no errors at idle is probably curve optimizer and not ram

dull ginkgo
short blade
#

nah

#

previous fastest was 59:57

dull ginkgo
#

lol

short blade
#

i've decided to try decreasing all of my core offsets by 1 except core 0 cause it's already at +3

#

let's see if my pc still reboots while idling overnight

#

didn't get any errors that could tell me which core

modern walrus
sterile flame
#

@clever epoch you still have to worry about it with pbo, cause if you don't set voltage and just set pbo it'll let your cpu pull up to like 1.5v stock if your chip wants it
Haven't encountered a single am4 bios that defaults lower than 1.45,
My brother bought his 3600 when they came out and has been running a 4.2ghz all core auto OC at 1.47ish with no noticeable degradation whatsoever

clever epoch
# sterile flame <@111605868138655744> you still have to worry about it with pbo, cause if you do...

  1. PBO will boost to 1.5V for low loads, because degradation is a combined factor of voltage, temperature, current, and time.
    If other factor(s) are low, high voltage is fine (up to a limit, then you have to worry about oxide breakdown which doesn't care about current). PBO is safe as long as you don't adjust the safety limits, as it is controlled by the SMU on the chip itself, which has access to much more information in real time than you do.

  2. Your brother is lucky, and he hasn't loaded his 3600 for very long.

1.47V static will absolutely degrade any Ryzen chip, even Intel chips. Unless he's left his computer off or idling pretty much the whole time, I can guarantee you there's degradation after 2 years, he just hasn't tested.

short blade
proven canopy
#

On 2) - every chip is different, some will tolerate higher voltages/currents for longer. Luck of the draw.

short blade
#

will any 7nm chip actually tolerate 1.47v for 2 years at any level of load above roblox though

#

the highest FIT voltage I've seen was 1.30 in all the 20+ 7nm ryzen CPUs I've owned

#

perhaps it was a sample that could do 4.5-4.6ghz at 1.47v before and he never noticed degradation because he only set it at 4.2ghz

#

could be unstable at stock

#

my voltage doesn't get anywhere near that high in demanding games

clever epoch
sterile flame
#

@clever epoch ok but when I told you I ran my CPU for an hour on an Aida 64 stress test and the peak voltage hit 1.4+ you told me that was too high, but now you're saying 1.47 peak is fine
So which is it? Is 1.47 a safe peak voltage or not?

clever epoch
#

1.47 voltage under low loads is fine

#

1.4+ static is not

#

hell, 1.3+ static is not

#

Also, I was commenting on the fact that your average voltage under load was above 1.35 or so, don't remember the number

modern walrus
short blade
#

I'm not willing to compromise on power saving

#

high/ultimate performance mode prevents your cpu from ever going idle

#

and I run my PC 24/7 to mine on GPU overnight

#

and also spend more time studying/working than gaming

#

so it's important that I can idle

#

I just lowered all my core offsets by 1 and it hasn't crashed yet

#

gonna test raising back individual cores to identify which one was the culprit

sacred kernel
#

anyone know if I can set tRCDWR lower than tRCDRD delay?

zenith palm
#

Yeah they don't gotta be the same

sacred kernel
#

ok

#

will it cause any instabilities?

#

@zenith palm

zenith palm
#

depends

#

i cant tell you idk what die you have etc

#

try it and test with tm5

#

extreme anta

sacred kernel
#

hynix cjr

zenith palm
#

depends on your other stuff

#

with every timing you change

#

test every time

sacred kernel
#

ok

short blade
#

I have tRCDRD 18 and tRCDWR 9

sacred kernel
#

20-8 no errors so far

#

nvm one error

zenith palm
#

One error = fail btw

sacred kernel
#

yes so I'll try 9

#

or

#

are there any alternatives

#

I've seen people adjusting some ohm values

#

and fixing single errors like that

sacred kernel
#

no errors with 9

sacred kernel
#

so

#

limit reaches these

#

but frequency is max 4525

#

why is that

twin widget
dull ginkgo
#

Early sample chips, when they came out, had much worse silicon

twin widget
#

True

zenith palm
#

i only get 3.9 all core with tuned pbo on my 3600

#

4.05 single

clever epoch
# sacred kernel

My best guess: that's the frequency of the clock gen for the most recent polling interval. The cores can run AT MOST at the clock speed of the clock gen, usually most cores will be running less than that speed.

sterile flame
#

@twin widget yeah so did mine. He has a little tower cooler on his and I have a 240mm aio though
But yeah, the frequency wasn't what I was noting though, the core voltage was

midnight lynx
#

can anyone tell how to overclock pentium e5700

modern walrus
short blade
#

what's your cpu running at with nothing open

#

package power in hwinfo

#

I idle at around 14 watts with balanced mode, c-states on, and 1.05 vsoc

modern walrus
#

let me check

modern walrus
#

54 watts lol

#

Is it better to do GPU mem clock or core clock first & does it matter?

south sky
#

And yes it matters

#

same for cpus, you oc the system ram before the cpu

#

Just because mem can easily destabilise gpu but not the other way around

modern walrus
#

ah okay

#

thank you

modern walrus
#

Do you want your GPU OC not to hit any of the limits in HWiNFO?

#

i.e. power, thermal, reliability voltage, max operating voltage, utilization, & SLI (doesn't apply to me but just putting all of them)

zenith palm
modern walrus
#

I'm using X1 but I got em all cranked up

zenith palm
#

Yeah same difference

modern walrus
#

darn

#

9216 on Port Royal. Best w/ my setup is 9396

#

so close

#

average is 8288

modern walrus
#

I don't really get scalar. Here are my results comparing 2x and 3x

#

zz...zzZ...zzZ...Z...z...

unknown.png
#

5900x PBO +150mhz btw

short blade
#

alright my PC finally survived idle overnight without rebooting

modern walrus
#

If I connect 4 monitors to my 3070 it doesn't wanna play ball

#

it won't support 1440 on one monitor

#

and won't run HDR on the two HDR compatible ones

#

I guess 4 monitors, 2x 2560x1440, 1x 3840x2160 HDR, & 2560x1080 HDR is too much

#

heck

sterile flame
#

PBO is trash
Manual overclock for better frequencies, better temps, and better voltages at the same wattage

#

PBO made my 3600 boost to 4.2 at nearly 1.5v and 90w
Manual overclocking I got it to 4.5 all cores at 1.3v at the same 90w

short blade
#

lol

#

whatever you say man

#

my 5800x does 4.3 manual and 5.05 pbo

#

also... somehow I doubt you have actually checked if 1.3v static is safe for your chip

sterile flame
#

It isn't 1.3v static, it's 1.3v peak.
TSMC says 1.3v is about the top end for their 7nm process
If you scrolled up a little before opening your mouth you'd see that yeah, I definitely "checked"
Idk why youd choose to talk down to people just because they don't like AMDs crappy auto-overclocking, but that's a really pathetic way to live your life.

#

I know damn well you didn't check if the 1.45v that PBO is running your cpu at under load is safe for your chip, because if you had you'd know that it isn't.
Remember that big controversy where amd and motherboard manufacturers were passing blame back and forth, because Ryzen chips were overvolted at all stock and amd claimed mobo AIBs were the ones doing it, where as AIBs basically turned around and said "bullshit were running the exact specifications amd outlined and letting the chip sip the power it wants on its own"
And then Ryzen 5000 and Radeon 6000 came out and everybody stopped caring

zenith palm
sterile flame
#

So yes, I spent 3 days looking into safe voltages before I even touched the thing, whereas you enabled PBO and said "that'll be good enough" apparently without knowing that less than a year ago there was talk of lawsuits over PBO killing people's chips from running them at too high of a voltage

proven canopy
#

This reads like a copypasta

sterile flame
#

@zenith palm it wasn't at idle, I'd watch it pull 90w and 1.47v in tandem.

zenith palm
#

Tandem?

zenith palm
sterile flame
#

Like, during the same polling period. It would show me 90w at 1.47v. not sustained, but definitely above 1.4v sustained

zenith palm
#

Having high voltage under small bursts or idle is perfectly fine

sterile flame
#

Having consistently over 1.4v under load on a chip where 1.3v is technically a bit high is not perfectly fine

zenith palm
#

My 3600 doesn't go near that mate

sterile flame
#

With pbo it would very rarely dip below 1.4v under load

#

Maybe my board is crap, I don't know, it probably is, there's been a revision
All I know is that 1.3v peak at 4.4 is better than 1.5v peak at 4.2

zenith palm
#

Currently idling at 1.45v but drops when it goes under load lol

sterile flame
#

How high does it go under load?
I can show you my cinebench scores with PBO running 4.2ghz at 1.4v+ and with the manual overclock running at 4.4ghz 1.3v if you'd like

zenith palm
#

Sure lemme pull up cb 1 sec

sterile flame
#

I use r23 if you're looking to compare

zenith palm
#

1.288-1.31v

proven canopy
#

r23 might be a lighter load than whatever you actually use the chip for

sterile flame
#

Oh, well that's good. What motherboard do you use, and what frequency does the chip run at?

#

@proven canopy I know, I use Aida64 to test stability.
Cinebench scores go brrrrrr tho

proven canopy
#

aida fpu only is pretty intense IME

sterile flame
#

Apparently I never screen capped the cinebench score I got with PBO but it was around 9300

#

Not sure why it says it was at 3.6 but

zenith palm
zenith palm
sterile flame
#

Could also be your motherboard to be fair

#

I know but usually it'll show my real frequency on cinebench
Like here's the run I did at 4.6

#

It actually says 4.6

#

But the 4.4 run says it was 3.6
Actually it's possible that I did save the PBO one and just didn't name it, cause they all say 3.6 except for the 4.6

#

How annoying

#

But yeah, 3.9mhz all cores is about all you can expect if you don't tweak it. It probably boosts individual cores to 4.2 still, right?

zenith palm
sterile flame
#

Ah yes, I've heard that.
Id still be willing to bet you could squeeze higher frequencies out of a board with a better VRM design, though.

zenith palm
#

Cuz i have early production chip

sterile flame
#

Although buying a whole new motherboard for a couple hundred more mhz is pointless anyway

zenith palm
#

But it is known that early production 3600s only go up to about 4.05-4.1ghz single core

#

And also my pbo is tuned btw

#

Although i can't exactly remember what the values i have it turned to are

sterile flame
#

What kind of temperatures do you get at that frequency

zenith palm
#

Around 70c with an id cooling se 224xt

#

All core ofc

clever epoch
# sterile flame So yes, I spent 3 days looking into safe voltages before I even touched the thin...

I'm sure your 3 days of online research led you to more accurate conclusions than 3 years of work, testing, and communication with TSMC's nodes, after however many years of education and experience in designing chips and analyzing degradation/failure modes.

Certainly, AMD's PBO is so dumb that for 90% of scenarios you will have equal or greater performance using PBO than you will with a manual OC; PBO which is controlled by the chip's SMU which has realtime access to voltage, current, and dozens (hundreds?) of temperature sensors throughout the chip.

#

I have yet to meet a single person who degraded their CPU with PBO

#

I've certainly met at least a couple using manual OC, and at least 3 using CTR

sterile flame
#

Ooooh I see. Mine gets to like 72 on a 240mm aio, but it only gets that high cause I swapped the radiator fans for ones that matched my case fans because I had the temperature headroom. It was running around 67-68 after an hour on aida64 before I did

zenith palm
#

Right well idk what you doing but mine is a 30$ cooler....

sterile flame
#

@clever epoch
Lmao. Ok
with PBO enabled my CPU ran at voltages under load that you sat here and argued with me for like two hours saying they were too high.
I don't know how many more ways I can explain this to you. The setup I was using yesterday, when you told me my voltages were too high under load, was when I was using PBO. Enabling PBO was overriding the voltage limits I was setting.
I feel like you've just kind of decided you don't like me and assume therefore that no matter what I do it's going to be wrong
You told me to enable PBO and keep it under 1.3v under load
Well for some reason, those two things are mutually exclusive in my system
PBO was overruling the voltage limit I was setting and hitting voltages you said were too high
So I set the voltage and disabled PBO so it would actually abide by the rule I set, and now that's wrong too.
I'm not going to keep screwing with my overclock because some guy on the Newegg discord is determined to tell me I'm wrong. I found an overclock setting that is stable and performs almost 10% better than PBO did, and at a safer voltage. So whatever. Have fun, guy

clever epoch
#

You're misunderstanding what I've been telling you

faint tangle
#

If you disagree with someone then just ask someone for a second opinion

sterile flame
#

No you're misunderstanding because with pbo enabled it would pull 90w at 1.47v sometimes.

clever epoch
#

That's fine

faint tangle
#

What cpu do you have

sterile flame
#

At full load and 4.2ghz

#

Ryzen 3600

clever epoch
#

if it's PBO doing it, and you didn't change any of the safety limits, it's safe

faint tangle
#

You sure?

#

3600s shouldn't exceed 1.35

#

Even 1.35 isn't good

cold jolt
#

static isnt the same as what PBO is doing

#

the voltage isnt really 1.47v

clever epoch
#

^

faint tangle
#

Ah

cold jolt
#

monitoring software isnt good enough at capturing the frequent and constant changes in voltage

#

it hit 1.47 for a few milliseconds at best

#

then dropped to 1.2, then back up and down and up and down

#

thats why PBO is almost always better than a daily overclock

clever epoch
#

even a "fast" polling interval of half a second won't realistically capture the rapid changes in clocks and voltage

faint tangle
#

Pbo doesn't make any difference for my chip

#

I got an awful bin

cold jolt
#

because its ability to dynamically adjust speed AND voltage depending on load and temperature is better than any manual oc could be

faint tangle
#

Like maybe 3pts on cinebench

#

Haven't done a manual and won't try to

cold jolt
#

although, in some cases a manual oc could perform better

#

so if you have really damn good cooling, it may outperform PBO, but running the risk of silicon degradation over time

faint tangle
#

I've got the id cooling se-224xt

sterile flame
#

It shouldn't matter. If everything polls at the same time then the information it shows you on Ryzen master will be a snapshot of your system at that instant. If it shows 1.47v at the same time as 4.2ghz and 90w then those circumstances existed in tandem inside your system
And yes it's for a short period, but if it's happening often enough for it to consistently show when polled then that's bad. These things are cycling 4+ billion times a second and it's only showing you a value for one of the cycles

faint tangle
#

Wait

#

You're not overclocking through ryzen master are you?

sterile flame
#

No, of course not. It's just the software I can think of that would have the most accurate poll data for a Ryzen cpu

faint tangle
#

Ah

#

Hardware info shows voltage, no?

sterile flame
#

Yes

#

But with Ryzen master being made by amd I just figured it would be a better example

cold jolt
#

yeah, its fine for monitoring

#

PBO is just hard for software to monitor

#

due to how it operates

#

thankfully you dont really need to monitor PBO, because you just set it and go

#

then compare your results from that with whatever you can get from a static overclock

#

lol

#

anyway

#

PBO just makes sense, because it works well enough, and frees up the hours of your time that it would have taken to do a proper daily oc

barren ridge
#

yeah it's mainly just up to your mobo's vrms at this point

#

if you have good mobo and cooling, it would push as hard as it could

#

and stable cause it's adaptive

zenith palm
zenith palm
zenith palm
#

tFaw is okay at uneven values right?

sterile flame
#

You guys keep using PBO
I'll keep my manual oc

weary valley
#

okay

twilit hedge
#

Quick verification, 1.35v is fine for 3200MHz xmp, right?

zenith palm
hard goblet
#

still broken

unknown.png
#

awful

#

small task bar

#

welp back to big boi taskbar

zenith palm
hard goblet
#

it used to be pretty

#

now too much spacing

#

not align clock

hard goblet
#

but the answer is yes

#

i think u can us 1.25 on it too

zenith palm
clever epoch
prisma arch
#

i have 32gb corsair vengeance rgb pro 3200mhz 16-18-18-36
i have overclocked to 3266 15-17-18-35
do you think i can go 3333 cl15/16

clever epoch
prisma arch
#

ok

clever epoch
#

v 5.32

tall pelican
#

8gbit cjr, so should be able to do at least 3600

prisma arch
#

ok

tall pelican
#

you also have to 1.45v for memory voltage headroom

prisma arch
#

oh ok

#

i left it at auto :/