#overclocking
1 messages · Page 4 of 1
im pretty sure i have already lol
Can anyone help me so I kwasnt having any stablioty issues til i checked mark temp dependent vmin on in mpt but then i turned it off and i reinstalled my gpu drivers and everything and still having stablity issues since enabling that..? idk what it does… but I DDU and I restored my BIOS to default and still getting fps spikes and not smooth at all
oh
When I move my camera around in game it doesn’t feel like 165hz at 165 fps feels like 60
have you actually used something to measure your frametimes and fps and whatnot? like with riva statistics or something
Yea capframex
It was being bad even on stock settings with no oc
Should my minimum frequency be at 500? Or should I put it 100 below my max
Put it 100 below max and* bring your max down
If 2600 is causing all this stuttering then 2600 is just not stable
Or follow Forks' advice and not use adrenalin for this
Doesn’t it cause clock stretching though
My oc settings are fine I havent hsd any issues til I enabled that setting in mpt@
Having your min too high can cause clock stretching if there's something throttling
Like a heavier load (power limited)
can you run imlc/aida on this? i am curious LOL
Binning time?
No, I just spilled a tray of my good ram lol
Hello papis
Can anyone help me
okay
Can someone help me with uhhhh limiting my cpu wattage
since it stays at 88c idle
If your idle temperature is actually 88C then power draw is not the issue here
Are you sure you mounted your cooler correctly
Prebuilt from micro
I assume it's been like that since you got it?
I guess
Because I only noticed that because dallas looked at my temps
And told me
After me being in bios trying to enable resizeable bar
And the fans have always been spinning like crazy
Kinda
Yea, you should take it back to microcenter in that case. I assume it's still under warranty or the return period or whatever? Only been a week
Alr
Also for future reference and others BIOS is not idle, it does put a load on the CPU constantly just not a big one
(practically idle)
It doesn't allow sleep states either iirc so that'll increase the temp naturally
Don't think on a workload that low 88c is really normal or justifiable
You're correct, I was just saying that BIOS isn't indicative of actual idle temps
Thats what i was saying
Even with a load it shouldn't hit 88
yeah my bios temp is a few degrees higher than windows idle temp
like 30-33 vs 23-24
more than a few I guess
Sus
You should have a look at moreclocktool or overdriven as well
Instead of amd software?
Yes
With Mct I would always have to set the overlcocks though manually everytime
And? If forks says it's better then use it
Forks has some ridiculously good results with rdna 2 cards, I would take his advice over mine
Strongly suggest moreclocktool or overdriven
Oh bet
@proven canopy
What about cpu fan temps
And tower fans
What should they be at
Lol, hardly, also I've gotten pretty bad bin luck. My ref 6900xt seems good, but those are locked to 3000mhz...
You should ask @humble jay for advice 
Welp
24K on a 6800XT is more than enough proof for me

69XT OCF
I forget what sub the 24k was, the card was on ln2 though. Probably 8c8t 12900k @ ghz ecores off. I wish I ran once with all enabled for a higher total score.
?
?
?
?
?
^
^
ahh you needa turn off windows driver auto update for some reason they have an issue with amd atm
ooh
also ddu in safe mode
yea i did
thats what i dont get lol-
so do i pause it the auto update.. and then do a safe ddu in safe mode ?
how do ik when to turn the drivers back on
i just did that
you have an amd cpu? chipset drivers will go in there
try factory clean install with adrenalin instead?
the one in program files shows something else
i did
this the one i downloaded
Tried an older version?
that is
Straight up delete previous adrenalin version then
DDU did that
iirc ddu has options for deleting software too or just drivers
deleted from appdata too right
If you're deleting all the AMD related software, delete the AMD folder
Well idk if that AMD folder is my software though..
@dull ginkgo
yay to windows update settings
turn them back on whenever you need them
And how do Ik when that is lol-
Iv never had thee problems with my other pc I don’t get it
Windows updates don't matter much unless there's something big
for the time being just leave it off
So do this then
yea
People are telling me my gpu is running off windows drivers and not the amd gpu drivers like what
windows automatically grabs drivers when you reboot again after ddu and you're connected to the internet
Yes
Do that
Specifically do that
You can also pause windows updates in the settings and turn off your wifi/remove the ethernet
how am i gonna download the driver with no internet lol-
download the file and not install it
so this
It's not going to disappear from your downloads during ddu
when i download this though. it downloads the adrenalin software
i havent Ddued though
Just download it now then ddu
rn i have 22.5.1 instaled
Mhm
but when i try to launch it. it says this
I'm aware of the error
You can't ddu without downloading the drivers you install after ddu prior
Otherwise windows just auto opts for some driver it can grab in that time
So 1st step is this?
1st step
Download 22.5.1
Then
That
Then pause windows updates in settings (it'll resume after 7 days)
Or do that first
Do either
Your problem is it's updating itself and figuring itself out before you even get to finish the ddu process
You need to stop it from doing that
With 22.5.1* downloaded you can now also remove your ethernet and turn off your wifi
Then it's set as intended
Next, pause windows updates for 7 days
Then unplug ethernet and turn off wifi
Then do DDU again
In safe mode
Then install 22.5.1 as you've already downloaded it
unplug ethernet rn?
Yes
that shuts off the wifi right
my pc isnt wifi blutooth thing it needs a ethernet to work
oh bent
bet
How do I turn it off manually?
Like my pc box or my pc WiFi lmao
Bottom right corner
Yea I unplugged my Ethernet
Oh ok
Okay now DDU?
I mean unistall my bad
Clean and restart
Bet
Then when it boots to windows
Log in
If log in requires internet
Plug in ethernet
Then unplug when in
Then install 22.5.1
I shouldn't need to give any other instructions
It’s running
It's literally the official driver just run it
I would
did some testing on tWR scaling
AIDA cares a lot, IMLC doesn't care at all
interesting
Reboot
You should be done
Change this back
Unpause windows updates
Plug ethernet back in
Done done
Didn’t start with no internet in rn lol
Well plug it back in
I did just say this
Ya cos you listened
Now I shouldn't need to repeat myself nor anyone else when they say ddu
Not like this obviously
do i turn that off
Idk, had no effect on the ddu process
this problem i was having is happening to everyone
Again, wouldn't know about that
I never talked to you about that
You can just install 22.11.2
No need to ddu
You've cleaned it all up with ddu now, anything past this point is fine
Team wegg benching flash drive, password is wegg https://www.dropbox.com/s/dplsjxioy8z2r1q/Team Wegg benching flash drive 1.0.7z?dl=0
im trying to hepl my nephew with his pc and his wifi on his pc is really bad but when he goes to his phone the sped is better and faster he isnt hardwired either he is able to play without a cable but its showing slow speeds..? any advice
probably no wifi drivers installed or just a bad wifi module in the pc for future questions of this type please head over to #building-and-recc-chat though
@proven canopy is precision X1 better for 4090 curve tuning? I can only see up to 3ghz in afterburner
I don't mind working half blind
But curious
Not sure if you'd maybe heard something from someone else about it maybe

Afterburner needs a update for 4090s iirc
scalar do auto
Ok
everything is different
try doing -15 first
yes
ok
then try doing cinebench r23 single run
then imma run cinebench ye
to see if that has any significant changes
Bro what are those limits
Also why is there only one magnitude in curve optimizer?
idk
There should be a seperate one for each core
nice
now what abt my 6900xt
can i get it to be more unlocked
its just limiting me at 375w
MPT
what
You need to mod your vbios to get more power
For not much real world performance gain
Bricking your card if you screw up
well
you can do that
but this is more of a bandaid approach
Oh god, the ! guy found this channel
can you just go away lol
try going -20 on curve
to see if it makes any difference
i know it did for me
Ehh if youre just trying to go for benches then sure but if you’re looking for performance gains in gaming then don’t bother
whats mbt
ok, well i didnt buy a 6900xt ocf for just gaming
Click the link I sent
Card Bricking Simulator 2022
Yep
moving on since my 3090 is happy
since its not radeon
this is what i have
and 1000w bios
@narrow schooner did you die?
i still dont have a great way to measure power draw lol
I could buy a kilowatt, but that just measures total system power draw, and id buy an elmor pmd but ehhhh $50
Use hwinfo64 and maths init
That's not always accurate and running hwinfo in the background will lower my benchmark scores
You know you want that PMD
50 dollars though 🗿
But PMD tho
I would buy it myself but they don't have one for the nvidia 30 FE plug
Adapters 
For that to work I'd have to do like 4 adapters cascading lol
PSU > 12 pin cable > 12 to 3x8 > PMD > 3x8 to 12 > gpu
With gender change adapters after the PMD
Big titanium psu + kill-a-watt is most useful imo, I have a pmd and barely use it, creates a lot of clutter
@proven canopy @humble jay can I get some clarification here
6 ghz 2p 5.6 ghz 8p 4.4e
1.3v
This is good??
Ecores should do higher
What bench? Just try run some of the hwbot benches that have global points and see how you stack up
I'm seeing an sp104 chip do less that's why I'm asking
I haven't done any per core rules just all core and offset
I think one e core isn't happy past 4.4
I need to figure out which cluster
I can later. I'm running out of time atm
I keep getting a clock timeout windows event for processor 16
So I have to figure out which e core that is
It's consistently that one
This is all btw at under 310W
Spikes higher but meh
I'm also trying to learn how to run higher vcore but fixed vid
Wait does SP even mean jack anymore? Wasn't the best indicator bust was very useful
Does SP all go up a notch for 13th gen due to better frequencies?
he did on his shop, but it may be oos atm
Isn't 12vhpwr different from the 30 FE plug though
I'm pretty sure they're different sizes
pretty sure same size, FE just doesnt have the +4 pins
you can always ask elmor though
I'll double check it later when I'm home
Afaik they're the same just no sense pins on 30 fe
The auto OC thing? Only available to some GPUs
You can't hurt the pc dragging the clock or voltage slider in afterburner
so if i max it nothing but good things will happen
haha
Was hoping for a yes or no
You can max the voltage and power and it'll run hotter but clock higher. Everything else is trial and error.
What model GPU? We can give you a guess for a safe overclock
To clarify, you can't "blow pc up" by dragging said sliders in AB. Maybe if you lock fan to 0% and run it until something melts...
Asus 5500XT Dual evo OC 8gb
I don’t know much about overclocking but I wanna overclock so any help is appreciated
Maybe start with +75 core and +400 memory, and test to see what kind, if any, errors pop up
Instant crashes during tests means drop back on core, graphical weirdness means drop back memory
Nah it just would crash if you maxed out
Yep. I'd test core first, then mem
Run something like 3dmark time spy
Or play cod
Ye that’s what I need it for cod is so badly optimised
Valorant get like 300fps
Cod 90ish
2nd run of 13900KF tuning. 6.0 2p and 5.7 8P seemingly stable (need to tease a bit more voltage out of it I think)
Gonna try figuring out which is the bad e core cluster
Still 1.3/1.31V VID/vcore
6 is definitely doable at reasonable voltages
What I'm doing different now is messing with LLC, I'm doing LLC 4 this time (3 seems better but too hot)
Might play with the LLC a bit more

I have drawn some pretty terrible e cores
So far seems to be 4.5e stable
For 2 clusters
IT MIGHT BE 3 BAD REEEE
Nope, 4.6e is stable
I just spent like an hour tryna test limits of the e cores and then all of that just washed away and now it wants to do 4.6

meh, 8 core rankings are worth way more anyway
strong e cores are nice for TS/FS overall score though I guess
@proven canopy r8 it so far
6.0 2p 5.7 8p 4.6e 5.0r
I think I'll def be lapping, I think 6 is the P limit for sure but my 8p might be able to go a little higher
I'm 1.29vcore/1.315VID (Vid bumped up due to 5.0 ring)
But I think 1.32 is still *kinda safe
I haven't really locked in the stability yet tho but it's running comfortably

If I put the ring down to 4.9 then the vid goes back to 1.285
Omg if I had the temp headroom it might do 5.8p
Suprim does 3030MHz

Might try flashing the liquid bios to it for 600W and see what it does
kW bios next?
kW
As in a kilowatt
No KW one yet
4090?
There is
Send
If you have it send

I went looking through the KFA vbios stuff and MSI but if maybe asus had one then I didn't look through all of those
There's like 159 unverified vbios'
@proven canopy 1000W 4090 vbios?
What about it
If I say please will I get it 
I wouldn't flash that, might get the card stuck at x8
Stuck as in, no flash back to 16?
Don't really need it tbh anyway, especially for ambient
Ooh is that what happened to BZ?
yes
Ah ok
rip
Well my riser is 4.0
It didn't break my friend's suprim fwiw
So I will pass
Thanks for letting me know about it tho
Ok I'll try suprim liquid 600W
I don't think it has 500 graphics points in it but I still have ram to do which will probably be a lot of points
Wat
The 1.1?
Hit the load test
It changes based on bandwidth and drops speed to save power at idle
Oh ok
I did see it drop 2.0...1.1
Made me freak out 
Never witnessed that before
Oh yeah there's 4.0
If you hit the "?" there's a small render that comes up to load it hopefully enough to get it to max state
Yep, never stop learning
Good news is occt is working on my bench drive again
It genuinely just wouldn't launch for 2 days

Ima also ddu so I have clean drivers

Are the codes still the same I wonder
Oh I forgot about voltage slider unlock

Flash was successful
575W bay beeee
YEET
WTF
Oh ok afterburner was just dumb
It said 42C core and hwinfo saying 87C hotspot

It's actually 62 core
You made sure it was the legit copy of AB right? There's a virus version going around
Not the extremely convincing, afterburner-msi.com (don't click this one)
Ah man I need to do ram after this, it'll take forever
💀
My programs looking sketch af rn
4090 don't like high clock crash kekw
My card suddenly trying to do 3120 on the new vbios
It can do it
For a moment


3090 seems stable
Hey that's my old card!

Let's try 3105
Interesting that it's pulling the same power tho
That's what happens when you're voltage limited
I'm thinking the flash did nothing and this was the voltage slider
Yeah
Oh well, I have a backup of the vbios
So no harm done
epower time
Ok +210 seems stable
I can't really do a curve like this tho
If it wants 1.1 I couldn't help it
It got to the end of the test and died on 3105

The fan whine is the loudest thing on this card

NOOOO
I'M 17 POINTS SHORT OF TOP 100

That's run variance level, I say shut it down and let the heat that's soaked dissipate and run again
Ima bump the clock a touch again
Maybe I need more mem clock

My core clock seems to be about as good as it'll get
I'm 68th for my hardware total
Hmmmmmmm

Mmm ima try more mem
If I did the ram I would be well into top 100 reeeee
I need to do ram I think

Ok I'm at least happy af with the system
I BSOD
NOOOOOO

IT WAS LOADING THE RESULT
YESSSZ
YEEEESSSS
98TH BAY BEEE
Woo! Nicely done
It didn't upload to my acc

Ok not making that mistake twice
The score uploaded but not linked
Big sadge
Try again
YESSSS
YEETED MY ROOM TO COLD AND GOT A BUNCH OF POINTS
It uploaded correctly
93rd BABY WOOOOO
The 3DMark.com Overclocking Hall of Fame is the official home of 3DMark world record scores.
Def gonna lap the cpu next week, the 12900KF is waiting for it too
Sadge
600-700 from top 100
Idk if RAM would help here

Ram ram
Oh vram?

Well I think I was pretty close to the limit
+1200 I was seeing regression
+1150 was where I landed so I can try again another time ig
No time left for now
upload to hwbot
Cracked the paste before I turned on ht + mem oc, but could be worse. https://www.3dmark.com/spy/33345305
Did a 2475 core run for 16.8k gpu score
Jeeeeez
limit is 480W gpu

RAAAAAAM
Went better than expected, 2505 mhz gpupi https://hwbot.org/submission/5138999_
tbf mine thinks its a "530w" gpu.... it also thinks its pulling like 20w in furmark
Lol
The 3DMark.com Overclocking Hall of Fame is the official home of 3DMark world record scores.

This can def be better
I haven't tested above* 6GHz so this should be a pretty normal score

shunt mod go brrr?
soc simp
do you guys like my overclock
oh lol, i was just asking around if i could grav mount my trex to gpus with a riser
i only have access to dice tho, should still be fun
just need to wait for hot to get back in stock now

I don't think gravity mount is a good idea for dice. It is very easy to crack the paste on a gravity mount gpu, and you'll have to stir the dice around
If you can do without that, sure
can dice get cold enough to crack kpx?
i never stirred with my last benching either, just poured pellets
i dont think so
still shouldnt grav mount though lol
i dont actually do gpu xoc so i forget lol
maybe with enough voltage, but that's a problem on ln2 as well around -130 IME
Again, the problem is that you need to agitate the dice at some point
It will be difficult to do so without wrecking the mount
Yeah, I'd imagine small pellets wouldn't need much stirring
they were fairly large tbh
can see them in this
lilene said not to crush them down
I mean, if temps get worse over time, you know the answer
why
smth about splashing
you should crush your dry ice
it help sub 40c in that 2080ti bench
held*
alright, ill powder it down next time
what was your 7980xe running at on dice
we dont talk about that awful session
90c then?
nah, 67c
💀
got cpu to 4.7ghz avx512, mightve been phantom throttling from vccin starve tho
imc is too awful for me to be competitive
damn no 4kc12?
imma test my chip next week
gl
what
would have an error in 1B with maxmem
8gb maxmem, tried 25M as well, different error from not enough mem
can i overclock my gskill trident z royal ddr4 3200 to 3600 (on x570e)
sk hynix
probably youll have to change the timings though
@kind walrus hello you there
it says its running 5.01ghz
this is hwinfo
So is it running its highest clock speed already
To improve my vram, i've increased the memory clock offset to +550 in msi ab, is there anything else i can do to improve vram performance
3060 maxq
In games like Minecraft where Single threaded performance is more important would it be better to just OC the crap out of 1-2 cores versus trying to get a better all core OC?
You dk which cores it will hit and how
Stable all core seems better to me
Unless you are forcing the game on the overclocked cores with lasso I see no point
i heard mc is multicore now from some mc server recently, not sure how true that is tho
might've been referring to server side of things
You'd really just want it to increase your normal boost target because yeah, you can't pick what cores it uses efficiently
Minecraft with shaders or rtx probably would end up gpu bound first
i run heavily modded mc at like 500 fps on a 6 year old 7980xe, anything past haswell is probably fine surely
unless shaders, yeah
Gotcha.
Yes, that'd be server-side.
Heyo, so, i just got my new 3060ti, I'm trying to use msi afterburner to auto OC and it goes for 30 seconds, changes a few little things then says cancelled, any reason why that would be? Should I use a different program? Should I use the galax 1-OC software made for my card?
I've had OC scanner maybe work once for me and it was terrible
oc scanner blue screens for me lol
might be using power limits in it's oc forumla, so that might be breaking it
Afterburner is great but oc scanner reminds me of like, ryzen dram calculator 💀
Just terrible
You mean the calculator that told me my c-die could do 3600c18 with 1 simple trick?
Spoiler, it doesn't.
I mean you should be able to do 3600c18 with b die
Well yeah but not my old c-die kit rated for 3200c16
That kit's not budging up from that. All I could do was SCLs and slightly tighten tRFC
3266 would error within seconds
On a somewhat related note I got my 5800X to do 4775 all core in R20 on ambient. Stock max boost is 4700 so I had to brag just a little there lol.
I'm the #3 score on AIO now 
Did it again in R23. It's hilarious that the people below me might have 5050 but they're clearly clock stretching. Upon closer inspection their effective clock is 4525 lol.
Should i use 20 rather than 23?
Ive been using 23 for a while now
Ok lmao
R20 has global points, both have hardware points
The one time i used it it actually made a not horrible voltage curve

welp speeeeed
show me some fps gains
Hwinfo > task manager. Time for cinebench or w/e.
Nice, came in a shell
Hardware just shows 1 number
Wait
I don't mean that
Who's gone crazy w/ ram on a nonK 12th gen or done some stuff w/o touching the (vcc)SA? Was gonna aim close to 3600c16 or maybe around 3200c14 because gear 2 is poopoo.
Why not try 3600c14 if you're going to be on b-die like that
Ballistixs should have no trouble with that
But yeah non-k ram oc is pain, unless you're on ddr5 - it's more bearable there
Why not try get 3800 16 or 14
Tis 16gb rev e
0.95vccsa is the reason
Oh right missed the non k
Yea it ok. I'll try anything w/ 1.5 Vdimm or less. Btw it 8Gbit I think (1GBx16chips). Work tomorrow so Sunday is pc day. Gonna tune mem then play old games on the temp 660.
overclocking juice acquired
oh big sexy
M8E usually does really solid CL
great tRP too, poor tRCD
3600 12-17-12 might be attainable, c13 shouldnt bad that hard
tCL scales with vdimm, tRCD/tRP do not
sounds ambitious with my experience with m8e
12-17 anyways
14-18 should be easy, 13-18 might be doable depending on the bin
at 1.5v
i needed 1.7v to hit cl12 at 3800
trcd 18
It's says m16fe on your stick :/
I believe that's 16gb rev e
Yes that would be
3600c12 is the same absolute latency as 4000c14, which ive seen fairly often
(for tCL)
Don't mind if it's really "16gbit", but I just get mixed up when the dimms are 16GB capacity also. See I thought M8 & M16 mostly meant number of chips per dimm rather than specifically the 8gbit or not. what would the 64GB kit be that's 2x the capacity, lol. ---Maybe these oc easier than a 4x8 setup, but really I'm not concerned being such a ddr4 noob obviously.
Hey locked SA effect CL or really just the gear 1 mhz/MTs? Ah yea I found a Thaiphoon burner pic of how it helps me understand/tell M16 didn't mean 16gbit.

Mostly gear 1 stability
It's Dr 8gbit E
It's chip count
Wait wha
So m4fe would have 4 16gbit chips?
Isn't that 8gb?
If it's just a code I'm following but if those are the chip counts then I'm mega confusion
Is it not 8 16gbit chips SR??
Well M4 could be laptop ram perhaps, idk. I just indeed figured M16 = 16 chips and 16GB dimm / 16 = 1GB. (1GB = 8gbit chip)
---Yea I'll aim for around 3200c14 at 1.4v give or take and also mess around in gear 2 w/ high speeds a bit.
Try and see how high you can post loose timings but higher freq
If you get 3600 you're doing amazing
Then dial it in, you may need to kick freq. Down a notch
Gear 1 @3533, 3466 etc. Will be more worth it than anything you could get in gear 2
Cool, I've heard you need maybe around 4400+ to sort of match 3200 gear 1 or somethin like that. I do know "good" Rev E can hit way higher than my 12400 or b660 will allow, probably. Not a bad thing.
Yeah you'll 100% be SA limited by the chip
They cap around 3600
Some 3667, some much lower
Ah ignore the pic. Couldn't zoom in when I was looking for another. Mhm, I've read even 3600 might be a challenge indeed. That's ok. Not like it helps the 12400 a ton anyways.
So I think I might be on to something about the upgraded input filtering on the XTX resulting in much better clocks
How so, easy 3ish ghz?
Looking at the RDNA 2 vid over again from BZ he talks a lot how the ref card has pretty rubbish filtering, and the main difference between the aib's and the ref cards is exactly that really
OCF upgraded the hell out of that
On the 6900XT
So I'm wondering if that, because he says it talks about removing a lot of the noise from the voltage signal, has anything to do with them clocking higher
XTX has some insane power fluctuations
Makes me think Funny of the opposite of Nvidia in a sense how the FE cards seemed to be binned better. Interesting, this xtx sort of like the higher end 30 series in some ways. 3080 would test my 650w psus, lul.
Aib's are definitely pointing to better results so far
3.2 vs 2.9? Tho the actual results are a lot closer
Around 10% at 3ghz seems like a decent spread though. Still wild these newest cards are around 3ghz while previous gen was/is 2ish ghz. 
Yeah 5-10% looking possible
I'm not convinced about this apparent A0 vs. C8 thing yet tho
I don't feel like AMD would willingly release something when they know they can do better? But that's also why the possibilities of 7950 XTX and 7990 XTX exist
So we'll see ig
Moore's Law Is Dead agrees with you. His theory is they were having trouble getting 3ghz at below the 4080s power limit but if you can push 375W it's pretty easy really.
XFX barely pushed any higher power tho iirc?
I think it's just more a smoother and clearer signal = more better
The XFX and TUF can do the same clocks as the Sapphire while the sapphire has a much higher tdp
I'm not sure if more power is just plain better yet
Need the card myself to see that
I'd love to see igor do a power consumption graph of an aib card as well
Right the usual OC power limit is pretty low, but there's been some amazing results from modders
Nothing official or public so far, just a couple tweets
We'll see results from BZ soon
i really need to buy input caps
ive been salvaging them from my dead 7970 vapor x lol
with a tiny lil blowtorch at a distance lol, miracle the caps work
I have some but never used them kek
i need them for 260 comp
i bought pack 100 2.5v 820uF nippon caps for output ive been using for months now lol, never actually capped on the input side
probably should
so i need some help
70-75C is perfectly normal for a GPU
mhm i wouldn't worry about anything below 85c even than ur technically fine
Yeah up to 95 is technically fine, just not ideal
It'll throttle before that point
oh okay
any tips on gpu fan curve? dont really know what to set it to
or have it as
questrion
my fan speed is at 97%
but the higest my gpu fans taht i have them go to is 75..
?
Something is overriding your setting or you didn't apply the curve
As far as curves go you really just play around with it. You're trading noise level for heat and vice versa. You just have to find what works for you.
Gpu fan will not be affected by the bios fan curves
alright hm idk then
Some bios can read gpu fan speed but not control it
In GPU-Z? That just resets the monitoring info
If you're looking for a reset you want DDU
If you consider this fan behavior to be odd and unacceptable then it's worth a try
But personally I wouldn't care
Yes
Either of those will be able to tell you if it's currently at the expected speed and timings
Then if you want to be doubly sure you can run a stress test like TM5
Alright so it’s working then?
Yep on the surface at least. I don't think there'd be a need for a stress test unless you start having more problems like blue screens or crashes.
Hm Mk just asking bc I’m playing apex and when I alt tab my game runs like bad,
That's more a game problem
Figured
Some games run funky if you alt tab then alt tab back until pc restart. Sometimes it's also if we weren't fullscreen but idk. Part of why I like dual screens is somewhat less tabbing out. Plus, Apex is Apex (haven't played in 2 months but did for 700ish hrs).
this popped up again after alt tabing
yea i havent had this issue with my other ram though but that was 3200mhz
Interesting if ram does that sort of stuff. Not sure on the gfx driver, perhaps ddu & reinstall latest. My dad had an amd (ATI) gpu but I've only had Nvidia. I have driver issues too but can't remember when last was.
I got the last est
Idk if there’s something special I have to do bc it’s 3600mhz of ram
Since you're on Intel you can try changing Command Rate to 1T, I see it defaulted to 2T
how do i do that
3600 18-22-22-42, is a s8c bin
dont get what u mean
cdie is pretty much the worst dram for ddr4. tho there's still some oc potential, expecially on that 3600 18 22 22 42 1.35V bin
I should do this though
find "CR" or "Command Rate" in bios, just change "2T" to 1T
No it's far from the worst, but it's limited in potential and only scales up to around 1.35v
fine, worst ddr4 dram from sammy 
i have some ChangXin 3200 jedec sticks i tried to tune and they error out with anything beyond a tiny tune
E-Die would like a word
It's described as "C-die but sadder"
2400c14
Usually under the memory overclocking menu
isnt edie EOL tho
Ones of those
Advanced DRAM config
Command Rate, set to 1
There’s
1N
Alright snd that’s it?
For this memory fast boot THSTS on auto shoukd I mess with that or what about DRAM timing mode on link



