#overclocking
1 messages · Page 108 of 1
yes
Ok I will
tPPD
What does it do
Windows felt a lot more smooth
With the timings lowered
Starting testing now
Ideally what are we aiming forV
4400 cl17?
4400c16-16-16-36 would be good
4400c16-17-17-37 would be slightly less good
4400c15-16-16-36 is likely impossible under 1.6v DRAM
You think that’s possible for@me 16 16 36?
4400c17-17-17-37 is probably fairly easy
flat 16s is almost certainly doable but the question is whether it's doable without a fan
Ahhhh I see
Failed test
4min 7 errors
But I didn’t have power down mod disabled
at 4400c16?
just tRRD & tFAW?
Yeah
try tRRD_L at 6
I just did power down mode off
& leave the others
Power down mode off only
No errors so far
Haven’t gone into bios yet
To change to 6
just got an error
going to try 6
alright
I’m excited to try the cpu oc on this board too 5.1 or 5.2 like u said my board could’ve been bad
Wait
Was I supposed to have voltages set
Io sa
To 1.3???
are they not set at all? lol
ya you can definitely get errors from those cuz auto will likely put them too low
Oh wow
Should I set it back to 4666? Lol
4400 15min test 1error on the first test
well it kinda depends on whether you want higher frequency or tighter timings
cuz if you go up to 4666 you'll likely end up with 17s or 17-18-18-38
maybe even flat 18s
since your voltage is gonna have to be high to run tighter
when you got the error, did you see what the temperature was?
ya that's a lot of errors
& that's without tuning anything either
which is gonna make it hotter
especially SCLs
oh 42C isn't terrible but I think you need a little more voltage if you're getting the error
this is at 4/6/16 for tRDD & tFAW?
SA IO won’t change it?
Yeah
4400
In bios it said 1.4 and 1.29
For those
On auto
I think
is 1.4 for IO or SA?
I’m not sure but
I thought it was usually 1.0 tbh
So idk what ur said 1.4 and 1.29
It**
The top one was 1.4
Bottom 1.29
I had to head out for dinner but when I get home I’m going to continue
What all steps will I need to do
If you aren’t available
okay
set 1.3/1.3 & test again. 1.4 is no big deal on SA but not good for IO. If you can't get it to run without errors on TM5, you can raise voltage to 1.55-1.57 for DRAM but you might want to drop to 4200 or so instead cuz if you need 1.55-1.57v to run 4400c18, you're gonna likely going to need like 1.6-1.65v to run c17 & even more voltage to run c16 & without active cooling that's very unlikely to fly
ya I was able to get 4200c15 at 1.64v with 2x16gb so you might be able to get it with 1.6 or so but c16 is probably the better choice without active cooling
Any other timings also
bdie's achilles heel is heat
I'll be around but if I'm not for whatever reason, the guide is really helpful.
Ok ok
I'll be around tho
Thank you . (:
np
ggood never leaves.
lol I try
RAM oc never ends
back
thank you haha
trying 4533
I'm still here cuz uh I live her I guess
let's see how it goes then haha
oh wait
its supposed to be 4200 c16
lol
am i trying the timings?
16 16 16 36 4200?
or what timings is it supposed to be
ya that works
maybe try 4266 first
4400 for daily isn't nuts tho if you wanna try that first
try flat 18s-38 first
Ok I’ll wait another 10 then try that!
Temp 43 max
ahh 3 errors on number 4
i prob need more voltage
43 on that test isnt bad tho is it?
43temp
18 18 18 38
4400
ya more than likely a tad bit more voltage would help but 43C is getting a liiiitle hot
how much voltage are you running rn again?
it won't as like a direct result but it'll likely be hotter inside the case because of it so indirectly ya
same with GPU
try 1.55
kk
I think you can get away with that without active cooling but still recommend active cooling lol
im not the most tech savy guy with building etc
you think it would be hard to do the fan thing
he was talking about
cuz i have an extra fan
laying around
well there are a few ways to do it
Dominator Platinum DRAM already has generous overclocking headroom, but to push it to extreme performance levels, you need active cooling. The Dominator Platinum airflow RGB is custom designed for quiet operation and optimal cooling efficiency, with dual PWM-controlled fans that deliver up to 10....
What plug do I need for that? Usb?
motherboard fan header
And if I was to try to do the fan thing do you think it would take like 30min?
eh 30 minutes sounds about right
That’s not bad
the second one you have to attach the clips to it with four screws
then it just goes over the little snap in things that hold the RAM in
The first one is easier?
the second part is true for both I believe
I dunno if you have to assemble the first one
but it is way cheaper which is good
honestly I dunno how well it will work. I had decent resullts with the 2d one but I was also running 1.65v
the second one has two fans
it was getting to 48C then crashing for me
trying to run 4200c15-15-15-35
but that was also on 2x16gb RAM
which tends to require more voltage than 2x8
Ahhh ok ok I see
The me just doing the fan thing rn
It works ok?
So I don’t have to wait lol
Or no
ya you can
I think the simplest way is basically hang it with a zip tie
if you know someone with a 3D printer, they can make you one you can just connect a fan to lol
i wish i did lmaooo that would be great... zip tie like hang it from the top of the casE?
and 4400 18 18 18 38 failed 1.55
1 error 10min
so i put it to 4266
lol pretty much
try that & try 1.32 IO/ 1.36 SA
Which one?
And what’s more important? Hz or timings
Like 4200 16 16 16 36
Or 4400 18 19 19 39
mhz
to a certain point tho like 3800c14-15-15-35 is better than 4000c17-18-18-38 or something
usually every 1 tCL is about equal to 200mhz
in general but probably even more so for your CPU since it runs 1:1
I mean like gear 1 3600c13 v. gear 2 3800c14 isn't even close on 11th gen & up
4266 18 18 18 38 1.5drsm 1.3 io SA
Still solid 20min in
Didn’t try the higher io and SA yet tho
Wow I see mhz really matters
for sure
wait so you're looking good then?
Hahah no worries you’re good
do you still have 4/4/16 set?
oh ya that's right
tRRD_L 6
Yeah
you can do those high ass timings to get high frequency really high
Bad for gaming?
you're lucky in that your CPU handles high frequencies really well so you can go either of two routes
I've never done 1:1 comparison but the guide says to prefer frequency over tight timings
Ok ok I see
That explains why they say best ram kit
Is 4800
Lol
19 26
So currently only thing that worked is 4266 18 18 18 38 1.5drsm 1.3 io SA
Currently testing 4266 16 16 16 36.
15min no errors
So it’s prob gonna work…
Remember we got 4800 to boot on 18 19 19 38?
Should we go higher timings?
Or stick with 4266 16 16 16 36
1.30 io sa
1.5dram
4800c17-19-19-39 is the best kit for DDR4 I think
"godbin" bdie
they have Rev E kit that's like 5100c19-26-26-46 I think
I can't remember if those are the primaries or not but I know it's the best Micron kit
Oh hell no lol
you'd want tWR at like 10-12
tCWL 14 or 16
😂😂😂😂
Assuming start lower to higher?
ya that's the best bet
starting with lower is how you end up clearing CMOS a lot lol
tWTR_S/L 4/8
before tWTR_S/L tho you gotta tune tWRRD_sg /dg
the guide pretty much tells you & the order to do them in
although I disagree with their approach of not doing tRFC/tREFI last or very near the end
Which guide? I might’ve missed it
So what’s the order you suggest then
Just give the safest timings I guess
That should work
well the ones that auto does the worst are the tRRD_S/L, tFAW, & tWR
it usually does tWRRD_sg/dg pretty bad. Those two timings are directly linked to tWTR_S/L which end up bad as a result of that lol
Also if I have 1.35 v core 1.3 io and SA. 5.0. What should I try for 5.2?
Bad as in tight? Or high
that's really a huge difference based on silicon lottery
bad as in like really wasting performance bad
really high
Ohhhh ok ok
Ok so 5.1 more realistic attempt
that gives you an idea of what a golden sample 10900K does v. most 10900Ks
i.e., if your 10900K can do 5.2ghz at 1.21v under load, it's top 1%
mine can do 5.1ghz at 1.305 under load which is trash
Ohhh
so I need 1.415v vCore with like Turbo LLC lmao
That’s on a 10900k?
And also the ram mhz is for fps?
Or for latency
Like 4266 16 16 36 vs 4533 18 19 19 39
FPS difference? Or feeling different latency wise
yes
timings i.e 18-19-19-39 is for your latency
you probably wont feel it as latency is measured in nanoseconds (ns)
but idk as I've never tested gameplay with XMP timings
but in my experience dropping timings has improved my latency by 200ns based on the loose settings detailed in the memtesthelper guide
Ballistix 32gb Rev E
my bad been painting motherboard's fingernails
how's it going?
4266 running nicely
16 16 16 36
5.2 oc
Kind of want to try like a 4533
Or is it not worth
Like a 19 19 19 39
4533 flat 19s is good
I think you might do better with 4266c16s
there's a formula for latency
So what is 3600 cl14?
3600 is 7.7?
So, 4000c16 is good?
Even with loose timing?
I haven't had a chance to tighten them.
ya 8 v. 8.89
Ok cool
Don't they make bdie with flat 14s?
I'm not home at my pc so I can't look at the list I have of all the bdie ram, but I think some are flat 14s
Oh what the lol
14 15 15 35 to hard on 4000?
at 4266?
Keep 4000
it's really just a function of voltage
I believe the list has 3800 14 14 14
Expensive af though.
ya no point in paying so much for dying DDR4

Ddr4 is new to me lol
The last pc I had was ddr3 and that was about 10 years ago lol
Its gotta be good for a good bit longer yeah??
A little bit longer yeah, but not really for the high end kits
Ballistonks is disappearing now for real which makes big sadge 
Not sure what you mean by that.
Ah yeah, I'm glad I picked up a kit when I did. Was 4 months ago and now I'm they are scarce.
What about 4400 c17 17 17 ? Or 4400 c16 18 18
Or is that worse that 4266 c16 16 16 still
My game feels SOOO much better thank you lol.
Going to look into those timings on my gfs ram 4000 16 19 19
It feels slow etc I’ll send a couple of pics of timings and see if u know the issue
It passed stress testing etc so it’s confusing
Also 1hour of gaming my temp is 44 still 1.5dram
On her pc trying to copy what u sent but I’m not sure for this part
32GBTwo sticks of 16GB ram
4000 16 19 19 39
25
This is the oc i put on it awhile ago. It still just feels input delay or something. Similar to the default xmp setting on.
@modern walrus I guess there aren't any 3800 flat 14s.. there are 3600 flat 14s and 4000 14 15 15
Die ultimative HARDWARELUXX Samsung 8Gbit B-Die Liste
Inhaltsverzeichnis / Table of Contents
» 8GB Module & 2x8GB/4x8GB/8x8GB Kits (SS, SR)
» 16GB Module & 2x16GB/4x16GB/8x16GB Kits (DS, DR)
» 32GB Module & 2x32GB Kits (DC DIMM)
» SO-DIMM 8GB/16GB Varianten (SR/DR)
» Häufig gestellte Fragen...
3800 is a really weird bin and rare to find as a XMP
I just took a 4133 and clocked it to 3800 manually
3800c14 is one of the best b die bins, costs a metric crap tun tho
Although oloy have 4000c15 sticks at a decent price sometimes so i don't see the point tbh
Yeah it's a good kit but rare like I said. Most of the time they'll stick it to 4000+ if it's good b-die.
Ya Im tempted to get those 4000c14s but I think DDR5 is the next move
Is this also on an msi board?
enable rount trip latency again then send a screen shot of timings
ideally you want to get to the same 60/61 7/7
but it depends on the board
once you set round trip latency, boot & see what it says on asrock timings
if, for example, they say 74/75 12/12 you want to find the number that reduces the second group of numbers to 7.
12 - x = 7
x = 5
74-5 = 69
75-5 = 70
12-5 = 7
12-5= 7
so you'd put 69/70 7/7
the timing is based on tCL so higher tCL, higher numbers
oh okay even better
it seems like a pain but it helps latency
I've had wacky ones that ended up like 64/66 7/8
as long as the first group of numbers are within two of each other & the second group are within one, it's good
the higher the frequency, the harder it is to get 60/61 7/7 then you'd try 62/63 8/8 etc
here's a more detailed explanation if you have absolutely nothing else to do https://www.overclock.net/threads/official-intel-ddr4-24-7-memory-stability-thread.1569364/page-776
I found out that tFAW, tWRRD_sg and tWRRD_dg tightened makes my RAM heats way more then the other subtimings, any explanation to this?
slightly compared to 60/61 7/7
you may not see a huge difference cuz MSI boards train them well in general
others like Gigabyte will be nuts
oh wait on MSI you have to put other things in too like initial values, compensation, & some others
what are the four numbers?
oh my bad just saw this
I'm on my laptop in Domestic Arbitration Law class
okay for IO compensation put 21 for both
for RTL initials 69
IOL initials 4 for both
try 56/57 6/6
the lower the better lol
ya just leave 65 then
the initials don't matter as much anyway. I've only ever seen MSI even expose those timings
the RTL/IOLs matter but the initials not so much
IOL CHA & CHB should be 6/6
hers should be 58/60 7/7
which is kinda strange I guess but I've had that happen on my 10700KF with 2x16gb DJR
oh ya initials you can likely just put what auto has
one second
58/60 7/7
they can be two apart on RTL side
wait does she have 2x16 ram?
okay
it's basically already tuned if you look on the left 58/59 7/7 is fine. you could try 57/58 6/6 if you want tho
ya MSI is pretty good with that when Round Trip Latency is enabled
okay hers are pretty good already
but you could try 58/60 7/7
only worry about D1 btw with 2x16gb (dual rank)
coult try tWRRD_sg 34 or 32 which will drop tWTR_L to 10 or 8
you can change all _dd to 1 but that's just kinda aesthetic cuz without 4 DIMMs they don't do anything
tRTP you could try 11 or 10
otherwise this is pretty darn good
maybe tRDRD_sg/dg/dr/dd 7/4/7/1 and tWRWR_sg/dg/dr/dd to 7/4/7/1
oh whoa
missed oe
tCWL to 14
if that doesn't POST change tRDWR_sg/dg/dr/dd to 15/15/15/1
tCWL is tied to those tRDWR settings
one down the other up type thing
ok
you might notice that the "initials" are what rank 0 ends up with on dual rank so ya they're kinda unnecessary on 2x8gb anyway
yes. your RAM has chips (ICs) on one side of it, hers has them on both sides
nice
weird MSI will set tCWL higher than tCL but oh well
that tWRRD_sg can go down to drop tWTR_L
at 3600 you can probably drop tRFC to 276-296
tRFC is very susceptible to heat so maybe do 296
oh wait on MSI
there should be an option set to auto for those tertiary timings
change it to fixed
it's like directly above where you put those timings in
on auto it might change them despite what you write in
yes
or alternatively,
change tWTR_L back to auto
cuz when you change tWRRD_sg, it affects tWTR_L
so if tWTR_L is set, it won't let you lower tWRRD_sg
same with tWTR_S & tWRRD_dg
so auto for tWTR_L/S, set tWRRD_sg/dg then you can just fill in where they land
really you can leave them on auto if you wanted
those four timings can reallllly mess up stability
same with tWR & tRTP I've found
you can try tWR 10 or 11 but 12 might be the lowest you can go
tRTP same
looks like tRRRD_sg/dg can drop further too
try 28 for sg & 24 for dg
that should put tWTRL_S/L at 4/8
if you have trouble with that do 30/24 for 10/4
otherwise you're basically done
I meant for this setup
actually this one too lol
28/24 or 30/24 so 32/25 12/5 goes down to 8/4 or 10/4
You almost certainly won't be able to feel the difference with your mouse
Check with imlc or something
That's even worse lol
All in your head
BlessedBeams moment
okay nice nice
wanna try 4000?
3600c16 is
is it a huge difference?
8.89 latency
4000c19 is 9.5
but we haven't tried to bring down primaries yet
just seeing how high it'll POST
i wonder if this current OC is gonna work for games
but 3600c16 is a lot better if you run gear 1 (1:1) instead of gear 2 (1:2)
3600 is usually the max frequency 11th gen will run gear 1 in
are we gear 2 rn?
yes
anything >3600 is gonna run gear 2
that allows 11th gen to run really high frequencies
not sure where exactly where they even out
the formula is (tCL x 2000)/frequency
so 22*2000/3200
ya your old latency is 13.75
It's hard to put a percentage to it because memory overclocking for gaming is more for smoothing out the 1% lows, and that's not very easy to see
ya you definitely want to avoid 3200c22 lol
if it'll post i'll keep it like that
yeah i understand that
my fps dips in csgo/valorant were crazy. id be at 400fps, then it dips to 200, then 500, etc
this oc should help me get a more stable fps correct?
Yep
I'd guess you're bottlenecked at memory then
yep definitely
I think 4k should POST honestly like
maybe like 4400+ is where it might get more difficult
but you have timings pretty loose still
at 4k you can probably even lower IO2/SA
ya but at high frequencies
like it'll run 20-30-30-50 or something
at like 5333mhz lol
decent
Intel right?
you might be able to drop to 4000c18 or maaaybe 17 or 16
yeah 11700kf
Good
ya AMD we'd be having a 3800mhz party for sure
Yeah you should still be able to do 4000 without going into gear 2 and taking a penalty
really?
With the proper voltage tuning yeah
I couldn't get over 3600 in gear 1 with 11600K but I have a potato I think
It's the same IMC as 12th gen
alright
Bad board could totally cause that
gonna try posting 1.5/auto/1.45/1.45 4000cl19
I tried 3600c12-13-13-33 to max out gear 1 but it didn't really do great. 4800c16 on rev E did better
wish me luck even though i might not need it
why
I don't think you need to raise IO2/SA for that tbh
it might not POST cuz too much voltage there (it's a thing lol)
oh i just listed the numbers you gave me off the top of my head, probably remembered wrong
posting
you could probably do 1.35/1.35 on IO2 & SA for 4k honestly
I'm at 1.43/1.43 run to POST 4600c15-15-15-35 on dual rank bdie lol
single rank DJR is a lot easier on IMC
should i go back in and tune stuff or keep it as is?
I'd test that it's stable for a bit first
how can i do that?
OCCT can do a quick test that's reasonably good at finding glaring errors, but for a deeper test you want TM5
what about aida64?
Ehhhh
cuz that's like almost p95 processor/memory murder
Good to bench it but not so much for stressing
ya that's also me testing to see if I can run a bench with it not daily use
usually to see really quick if I like drop tRTP from 12 to 10 or something
but ya definitely not good for daily stability by any stretch
my current oc?
I meant using ycruncher 2.5b to test stability
ah
the top zip file has TM5 & has anta777 in it already
oh boy
you just gotta unzip then load it
what happens if i fail the test? bsod?
PCBdestroyer & absolut are really good too
depends how unstable it is
if it's really unstable then BSOD
if it's kinda unstable it'll have errors in the test window thing
you gotta click the load config & exit cuz the stock test isn't that intense
oh all that crap is in there? lol
my bad
click on TestMem5
this thing bussin
lol delete that crap
someone sent me that & I used it
but if you look at the config, it's the same as the stupid stock one
LOL
I was kinda mad ngl
ya you can just move that rar file somewhere & unpack it there
& delete the rest of the crap my bad
didn't realize I had all my junk in a compressed folder
For some reason anta777 fails to load on my system
Other profiles are hit or miss
tm5.exe?
click that then go to the folder with TM5
& look for a folder called bin
you want this one
hmm that's weird I wonder why... anta777 used to stop testing for me sometimes. It would run for hours then just stop moving the blue things around & keep counting time
Yeah it's super weird, it tries to load and just crashes every thread immediately even at jedec
even if you unzip the rar file again & let it overwrite the old one?
are you getting errors?
how do i know?
OH
Try running it as admin also
me when I edited a version of TM5 to use and attached it to a guide no one uses
Yes it starts on open
oh ya run as admin lol I forgot that
where? I want
That's expected
these need to be pinned somewhere or something... like just link to NeweggTechie github at least
It's the type of thing you should like run overnight to be sure
ah
extreme anta?
The test runs for several hours and stops automatically when done
idk whatever ggood linked me
yes
ah, ill try leaving it on overnight
should go thru 3 cycles & take like an hour & a half per cycle I think
before i go to bed
extreme anta running 2 cycles in 5 mins isnt a thing
takes like half an hour per cycle
yeah
oh might want this to see all your timings in windows btw. This is the newer one that shows you which gear you're in too
default config is crap
ya serj is stinky
load extreme anta
simple test is really that bad?
Or just delete that version of tm5 and launch the one from my guide thats linked
it defaults extreme anta
ya sorry I sent you ancient TM5 lol
I'm gonna download the one RAWR linked rn cuz I think mine is like a year old
thats the latest version
jesus christ
yeah i can tell this is more demanding
my typing is loadiing slow
so slow
ya it's beating your RAM into submission rn lol
LOL
why is my freq 3990?
is that just the
specific number
and they just say 4000 to round up?
you may not wanna do the whole mission of setting all the timings but at least drop tWR to like 12-14, tRRD_S/L to 6/6, & tFAW to 24
I dunon why it does that
does it to me all the time
like rn I'm not running 4600
I'm running 4596
while my pc is on and im doing the test?
sure
can you highlight where i set them?
looks confusing since theres a ton of TWR
in ur bios
ya just go to BIOS & do it there
the ones I said up top are like super safe ones
okay yeah we're fine for cycle 1
trrd and tfaw looks high
you wanna get 4 16 there so test is faster
dram timing control?
okay i'm here
okay send a picture real quick lol
okay the first ones under Secondary Timings
put 6 for them
so like the 5th & 6th from the top I mean
the first autos
uh
done
okay just leave REF cycle alone
alright
scroll down more to after REF cycle Time 4
my bad I put my only ASUS board in my dad's build & it was AMD anyway so I dunno these BIOS by heart lol
all good LOL
send another picture
not even if you like use the arrow keys?
I'm like damn I've never seen them let you change that many but not the rest lol
14, 12, 16
leave those auto

i ❤️ asus
for Write Latency put eh... 18 safest
isn't it already 18?
Their hardware can be good sure but I find their interface to be... different from everyone else calling things weird names
ya cuz it'll do the next even number below tCL
so at 19tCL, 18 is auto
so you can leave auto
asus has given me nothing but problems and dumb issues
just making sure cuz MSI will let tCWL be higher than tCL which makes no sense to me
featuring keychron keyboard
The IMC can silently override invalid settings
lmao how on earth do you have three tWTR settings? oh well staying auto lol
what's in "Skew Control?"

ah true. somehow it still reports like tCWL 16 on Asrock timings with tCL 15 dunno how
just asus being asus
oh heck no just go back lol
On Die Terminations heck that
Yeah the board will report the settings you put but the IMC will internally use different ones
ya just go ahead & exit. tertiaries matter but we're not going all out so just test this
ah okay
it's like ya right MSI no thanks
might take slightly longer cuz training so don't worry if that happens
nice
ty man
what do asrock timings look like now?
I mean you could try dropping primaries & all sorts of fun if you wanted to but
lets see
I doubt you wanna spend the time & you're already way better than before anyway
ya I mean as long as that gets thru anta777 which I think is highly likley
like I'd be extremely surprised if it didn't
you can chill on that
there's plenty more to tighten if you feel like squeezing a little more but that should be considerably better than 3200c22
I spent almost 4 hours yesterday with afterburner and timespy fine tuning my 1650s, I got a 10fps increase in the main game I play so I'd say time well spent :P
W
Out of 38202 benchmarks for my gpu.. I'm at least 81 so thats not to bad I suppose. Can't seem to get above that though, I think it's because the people above me have better cpus.
They're also probably using LN2
Oh, makes sense.
I'd be surprised if anyone has run a 1650 on ln2, maybe there are a few results on volt mod, but most likely it's a 12900k @ 5.3 with ddr5, that'll help time spy a ton.
Link the 3dmark result
Looks like bavor and traktor have all the
s, either on 10900k or 5950x
Higher is better but there's a point of diminishing returns, where it costs much more for not much more performance. 4000 is good and still a good price for a decent binned kit.
You'll probably have to run gear 2 but it's worth the small penalty if you have a good kit, 4000 with 18 or lower cas.
This would be my recommendation https://pcpartpicker.com/product/dqbTwP/gskill-ripjaws-v-32-gb-2-x-16-gb-ddr4-4000-cl18-memory-f4-4000c18d-32gvk
That's actually cheaper than the cheapest 3600 c16 kit right now
Learned 12th gen has this flaw where uncore underload usually drops from 4.7 and matches the e cores so I overclocked the uncore from 3.6 to 4.4
So now underload it stays at 4.4 and drops to 4.0 when idle instead of 3.6 underload and 4.7 when idle
Oh and @sudden torrent I couldn't get to 6000mhz again but 5800mhz cl36 cr 1T works and taking your advice setting it to 1.2 volts only draws 177 watts
Very nice, above average scores and less power/heat
That's my goal
It's pretty much squeezing the life out of the uncore and e cores as much as possible while keeping the p cores undervolted and cold
Oh and working on ddr5 latency
Yeah messing with a CPU OC will often destabilize a memory OC, that's why I usually recommend memory last
I'm a bit afraid of base clock to touch it
Most of the time it doesn't help much anyway
Understandable gonna leave it be unless I one day magically realize I've run out of things to mess with which I doubt will ever happen
I feel so out of the loop what I knew about 9th gen feels useless for a 12900k
It is a pretty huge jump, not just clock speed changes and cache increases any more. Now we have a whole new architecture with e-cores to deal with, and everyone has to relearn overclocking for DDR5
I guess that's what 6 generations of 14 nanometer knowledge does for you when superfin comes out
Sorry for long link, I don't know where to cut this one.
There seems to be only a few 12900k above me on the list. A lot of 10/11th gen though.
Yeah, figure, the only guys benching that card on a 12900k are doing it as a hobby as that card probably isn't the daily gpu
remember to set leaderboard mode
oh hey, jedi is #1, I know that guy
I think he's top 3 hwbot enthusiast team usa
4000 c16 like that is going to be either b-die or rev.e and either is great for overclocking
CR 1T and 2T would be the 10th gen equivalent
You can do 1t on 12th gen gear 2,, 10th gen and lower only has gear 1
To clarify, it's a completely separate parameter, 10th gen and lower just don't have options for gear 2 etc
Yea CR is command rate, which isn't the same as gear 1/gear 2.
Oh, well in Leaderboard mode I'm rank 31. That's even better
@modern walrus is it possible to tune VRAM timings?
Mmm fair enough
Increasing the frequency reduces your net latency anyway
Yeah I remember that from ddr4 oc at least

speaking of memory tho
does rendering use GPU core clock at all?
Depends on the program
I do 4200c18-24-24-46 on dual rank DJR on my 10700KF
I've never tried tbh
I'm using waifu2x to upscale some of my old animes with GPU but it only seems to use VRAM
the workload is still heavily CPU lol
Never used it so I wouldn't know about that one. Sounds interesting though.
if you do one of those rig jobs hanging it by zip ties lol
you could get one of those RAM coolers too
they're alright generally. I have the Corsair one with two fans
overpriced tho imo
I used an L bracket & one of the screws on the top fans/rad to hold a 92mm fan before but it ain't pretty
Pics or it isn't true.
eh let me go thru all 10 million of them hold on
Does good pc airflow not work good enough to cool ram so you gotta get a dedicated fan on it?
it depends on how much voltage you're running thru RAM in my experience
damn I really can't find the picture with the L bracket
I hated it so much that I just ended up doing this
ya on 5900x & 10900K
for me the issue with it is it stuck too far out & hit ipes
but it doesn't go much higher than the top of the RAM
oh damn
now that I look at it, it's gonna be tight on your rad if it even fits at all
There is no way
that's the Corsair one I have ya
I use it in 5000D now cuz I went with the loop
for whatever reason my DRAM temps were getting nuts
I mean probably the reason is I run 4133c15-15-15-35 at 1.64v
I'd rather run higher voltage tighter but my IMC needs too much IO voltage
that's actually kind of lucky cuz mine would immediately BSOD at 48C
honestly I've been slacking on benching for FPS
I usually measure improvements using like ycruncher or something liek that
in my experience you'd either get BSOD or crash
it also depends on ICs
like Rev E or DJR can handle the heat pretty well
bdie just quits
it could ya
best bet for those unexpected shut downs is to check event viewer
cuz GPU & CPU could also be the culprit
ah okay
ya stupid sweaty RAM overclocks are nice for benchmarks but for gaming it'll just be a pain cuz it crashes when it heats up
could also depend on secondary timings
all things being equal the 4000c16 should feel quicker
secondary & tertiary timings
if you run 4000c16 on XMP, it'll be likely be slower than 3600c19 tuned
ya XMP is lazy honestly
not on your part
but the motherboard only cares about working
& people who manufacture RAM can't tell the motherboard any of the timings to pick aside from primaries
I mean they should feel pretty similar because every 200mhz is about the same as 1 tCL according to the guide
some things work better with more frequency, some with tighter timings
From what I've experienced, temps shouldn't be bad unless you're like north of 1.55v
but--and this feels like a bit of a cop out--it depends on a lot of things & every kit is different
heat could the be the culprit, especially if your tRFC is really tight
wait hold on one second I might've missed something here
which CPU?
oh okay nvm that is always gear 1
er 1:1
was CPU overclocked?
ya bdie can start acting crazy at about 40C
see I've only ever had it crash
like it'll be completely fine then for me it's 48C where it immediately BSODs
there's like tRFC but also like tRFC2 & 3 I think
this was on stress tests
lol I waste so much time overclocking I rarely get to actual gaming
I should do a tm5 and check temps right?
Just to make sure my temps are good?
I never thought about it.
ya TM5 with anta777 extreme but I find PCBdestroyer gets DIMMs hot a lot faster
Idk, I never experience slowdown on my pc.
do you overclock your ring core?
oh okay good
what memory tests are you using?
oh okay ya TM5 is the best I've seen so that should be good
no chance your CPU or GPU is throttling?
is it possible that warzone just benefits from higher frequency?
I know that game is massive. What storage are you using?


