#graphics-cards
1 messages ยท Page 75 of 1
20 min in and cpu load is still goin well
40 min still good. Im sure cpu is fine
So this power settings and vsoc loadline?
In bios doin settings
VCORE soc?
Don't touch vsoc load line
Just normal loadline @frank dagger
Set to normal
These are all some pretty normal settings for a stronger imc
I gave up. Hired a technician online, technician didn't help either.
Pc is junk
Even after a clean install, it crashes
Psu is probably at fault, i don't know. It still crashes in power test, im done
That sucks
Well either way it's an issue with the IMC or ram, broken or not
Those bsods only occur pretty much from that bar for exceptional circumstances
Change of cpu would be my next suggestion
Maybe even go to new platform with AM5
A problematic EPS cable may be able to create a situation like this but usually psu has a few different errors
Cpu stress test was fine though, never had problems there.
But ima entertain the thought and switch cpu anyways
Kernelpower 141 being a particular one
Yea I know what you mean
We threw a bunch of tests and nothing
Yet a game immediately kills it
That's highly unusual
Power test as well
PSU was in 2021. It shouldn't go bad
It can still happen
I have a 2022 psu with a dead plug
Silverstone DA1650G if you don't believe me
It's not supposed to happen but can
Then fudge corsair
I'm not totally convinced of the psu honestly lol
Maybe SSD before it
If it's HMB
If you're able to run the 3D test just fine then I have doubts is PSU for the most part
Especially if no audible click
The ssd health should be fine though, its samsung
Samsung had a massive scandal in early 2023
I know, but this drive was the tier before it in 1tb
I'm only speculating here, not flat out saying it is broken lol
There's only so much that can be done from this side of a screen
Well there's a new problem in motherboard if ya wanna see it
Alr
Is there any way to stretch old gpus into doing stuff they shouldnโt
For fun
Clarify " stuff they shouldn't do"
Well everything you can do with them they're kinda designed to do?
But push even further beyond
Like make a 2gb gpu into a 2.5
Like crack the frick out of it
I mean you can with a soldering iron and some bios modding
But then you're going to have to buy some vram which isn't sold commercially
a friend said 2080ti has 33gb vram but it was deactivated or not supported
Your friend is a moron
๐
lemme check so I dont misintrepret that
44gb of vram
idk how one can make 44gb vram on 2080ti work
when it has 11 GB actual vram.
Modding
Very, very difficult and complicated modding
I can't imagine.
I read an article on it over a year ago atp, that's the only reason I know about it
So is it possible or no?
Basically you take off the vram chips, and solder on denser memory chips?
So I'd imagine 2080 ti is 11x1gb vram chips or whatever they're called, so they take them off and put on 11x4
Interesting
It sounds possible but complicated
And maybe risk burning up capacitors?
Anyways genuine question
330watt pulling power, normal for 79xt to reach high usage?
I changed the cpu from a 3900x to a 5600... it doesn't crash... maybe the cpu socket or cpu cord is bad then. Considering how the 5600 is 80w while the 3900x is 150w
OR the CPU socket is damaged, but if it was it would've shut down even with the 5600...
Hope PSU box has 2 CPU cables
Yep! 2 cords, ima revert the back and change the psu cord!
OMFG!
It is just a bad CPU PSU CORD!
I did all that sacrafice for nothing!
It was just the CPU PSU CORD!!!!
It hasn't crashed yet!
Okay, then, so my CPU overclock must've killed the CPU cord
3 days over one cord... 3 DAYS LOST!
Hey tech
You got it working?

Though tbf the fact it was so consistent at spitting wheas really threw me off the psu cables tbh
But glad the problem was finally found
Cos ye it'll usually say other stuff like KERNEL_POWER_STATE_FAILURE or CLOCK_WATCHDOG_TIMEOUT or other stuff like that when it's a dead EPS
I think DRIVER_IRQL_NOT_LESS_OR_EQUAL is another
There's a few
Got a lot of experience with that from dead 12900K's, lol
Probably had the makings of dying over time from sample quality tbh
How can i check for next time?
Mhm
I probably have to buy a external tool don't i?
Ye it's not easy to figure out most of the time
Psu failures can be really painful
The fact you made it to that in 3 days is in one sense really lucky
Usually takes a lot longer
Bruh... i should've done the "stupid" stuff first before goin into software and drivers
Dead cable is not a common one tho, especially when the errors were so consistent
I've had 2 psus fail on me with one rn on a dead plug. The two that failed had wild issues, and the dead plug one would just really refuse to start
Bsods would be completely random, and switchoffs are really common
100+ frames at 3440x1440p
So a cable fail like this and found fast is a good outcome imo
Sounds good lad
Just make sure not to change the settings too much in amd drivers and it should stay good
Too many changes at once can overwhelm it a bit lol
For context btw one took a casual week, the other took 4 months
The dead plug one took like a month on it's own
Tons of testing
Lots of rma's on the 4 month one, I lost 3 cpus and multiple ram kits
Perfomance on 79XT is wowZERS
144 fps in Witcher 3
80 fps with RT ultra
FSR 2 on
Well if we're lucky Nvidia will grace us with that gift sometime this decade
By that time amd will probably have stable drivers too ๐
Maybe by then Nvidia will fix some of the bugs in their drivers that they've had for 8+ years
And finally move away from the settings UI that they made in 2001
unlucky*
1200w 4000$ msrp
maybe will be x2 faster than 6090
Double the power for double the performance, and only at triple the price! Look at that great deal they're offering!
Hello everyone! Recently My graphic card started giving me frequent black screen, random lines and artifacts once in a while. After few days I could even run any games or benchmarks and I noticed that there was some issue in graphic card driver. I tried DDU and clean installed drivers. For some reason my pc is not recognising the installed driver. I tried reseating the gpu, update bios, still nothing. Finally I used my gpu on my friend's pc and it had the same problem but his gpu was working fine on mine. I checked Gpu z and I noticed that its not detecing the memory size or speed. So I assume its faulty vram?
I believe so. If it doesn't work in either PC and the other GPU works in both PCs then that GPU is dead
Try undervolting the gpu
And underclocking the memory
That wouldn't permanently fix it, but can potentially give it a few more months, or even past a year
Depends on how bad it is/what the issue is
What is strange is the Its still able to give output but its limited to 60fps and works only for basic tasks. Anything GPU intensive, it crashes. Is it possible that its using my RAM for memory since the vram is faulty?
If the VRAM is faulty then whenever that part of the memory is accessed the returned data would be corrupted
GDDR6 is quite good at error correcting enough to keep it from crashing
You can use tools to correctly verify this but if gpu-z is detecting less vram than the card is supposed to do, that's a pretty telltale sign that the card has in fact lost a chip or more due to a bga break or dead nand chip
I don't know what tools you'd use for this as I've never encountered this myself irl, but gpu repair channels might have a demonstration
What would apply at this point would be gpu repair/scrap tbh
Vram is really hard to fix if it's the module itself
https://ibb.co/9c0MqM9
https://ibb.co/pLv7FmL
Im not able to share image here. This is what I see in GPU Z. Memory size is 0
Ah then not good honestly
Hoo man yeah
Full vbios but no detected vram? Could be something
I was planning to stretch it till 5000 series release sadly Its gone before that : /
Ok
Go to device manager
Look at video adapters
Check your 2080ti for code 43
If that pops up...welp.
Yeah i get the code 43 error
Tried it. Still same
Well, not every fix... do you happen to have an old toaster oven you don't mind sacrificing?
Will a microwave oven work?
No
If you're willing to give fal a chance there's odds this would potentially work
u are suggesting to rebake?
Yep, reflow that solder a little bit
Hence sacrificial oven
It's also going to make whatever you cook it in toxic
Don't do it in your normal oven
Nah my mom will kill me. I will just get a new card at this point. There might be some guys who repairs gpu on board level. Maybe I will pay them a visit
It's a 50/50 shot of it working but you don't have much to lose at this point
You could see if someone like buildzoid (actually hardcore overclocking) will buy it to fix on camera
Does he have a reflow station?
Yes
A nice hot plate too for preheating
$240 3070 would slot right back in there as if nothing was really lost tbh
Used ofc
A little better RT performance but otherwise quite similar to what you have
Sell the 2080ti as for parts
120-150-ish
Cos it works but not correctly
Would cut your losses really short that way
Higher odds of parent approval too
Yeah $100 is a drop in the water
Just explain the pc is broken
It's a fair bit of money but to fix something that's not a gross amount
Yeah looks like I got no other option. Rn its giving some output so I will just use it till it dies completely.
Actually I saved up enough to get a 3060ti. Just a little more I can get a 4070 super
4070 super would slot right in too
Or 7900 GRE
What psu you got there mate
Psu and cpu is what I need
its a 5 years old pc with 850w psu. It should be fine i guess
Need the psu name my friend
850 is just the capacity, tells me nothing to it's performance
Full name not brand
corsiar rm850x
Ok beautiful
Same as mine, great unit
Either of those gpus will go right in
Rmx is still top tier even today
7900 GRE can be had for as low as about $520 new
6900xt/6950XT would also go in there for some $450
Or 3080Ti
If you upgrade to 40 series nvidia you can get this to go with it
https://www.corsair.com/us/en/p/pc-components-accessories/cp-8920284/600w-pcie-5-0-12vhpwr-type-4-psu-power-cable-cp-8920284
yeah it still got 5 years of warrenty left. Looks like it will be the last component to last in my build lol
Adapter also works in the box tho
But ye
Depending how much you saved, anywhere from about 350 to 550 will net around the same-ish performance if you don't mind used
Then you leap to 7900xt at $700
I always prefer the cable. I got measurably better results going from an adapter to a dedicated 12 pin cable on my 3080 ti.
I am looking at a some used 3080ti with 1 or 2 years warenty. I could buy it right away with the money i got and no need for a new cable.
I can play just about any game at 1440p high or ultra on mine, you'll like the 3080 ti
Oh, when you do @exotic viper , get a gpu support brace
$7 for a magnetic stick on amazon
Saves lost vram...lol
And broken slots
30 series and beyond are very susceptible to it due to the large coolers
In an ironic twist there
Lol
Totally unexpected and no one saw that coming ||/s||
So you mean sag might have caused this issue to my 2080ti?
Heavy coolers can bend the PCB over time, causing it to basically snap off solder
Very common problem on today's larger cards*
Yep
It wasn't as much an issue for 20 series
Just unlucky really
If there's at least 3 points of contact with the case and motherboard then you nearly eliminate the problem
PCIe slot on mobo, rear slots where the video plugs go, plus a support from the bottom
Well, 3 good points of triangulation ye
Its a MSI 2080ti gaming trio x. A large card hanging there for 5 years, so it could be possible.
It's possible ye
When you talk to your parents tell them you spoke to a pc support group
It's not really lying, not really truthtelling lmao
We do a lot of the pc store's support here
Heck you could probably tell them you talked to newegg's technical support since their CS team uses us like that lol
Honestly yeah
Its not actually a big deal. Im a working grown ass adult ๐
Hard to tell sometimes, we get all kinds here
Should have mentioned earlier lol
Even though discord is 13+
Yep still happens tho lol
my fans are at 100 rn and i can hear the damn clicking
Thanks for the chat guys cya
Np
Did you swap the aio
??? That wasn't the aio then
Stop the fans with your finger one at a time and see if it goes away
sure sounded like it
at 1500rpm?
Yes
Finger in the middle
I don't mean jam your finger in there, touch the hub
Bro it's an imposter
An unbalanced fan might do that
let me turn them back down then up again
im mad because my aio tube run isnt as clean
idk if i put it in wrong but the tubes are weird for this one
I thought you figured it out it was the pump ๐๐
Lmao
๐
also did i put the new aio in wrong or smth
this was the first time
this is now
the tubes are all wacko it seemed like
ok let me go render like 90 chunks in minecraft thats the only time i can manually recreate the noise it seems like
Oh you have a different twist in the tubes
If you rotated the radiator 180 or 360 degrees in one direction that would probably fix it
Project for another day
First thing is finding this fan i guess
i cant get it to make the noise idk
ill play around with it tomorrow its almost 2 AM
once I find it and swap it out ill mess around with the rad to fix the tubes, i was just nervous playing around with it i didnt want to bend and break anything
You'll be fine
Takes a lot to break an aio tbh
Unless you willfully go for the radiator channels
first time (i guess 2nd since its a replacement) handling one the first time i was super careful lol
i bumped the block on the case once and i thought thats what might of done the noise or smth
Nope
I've been pretty nasty with my aios lmao
They're still fine
The fact they're in water means they can handle a fair bit lol
fair
Radiator are very simple as well, yoy have water channels, and copper fins
Quite similar to the heatpipes you see on air coolers
The can handle a fair bit of abuse honestly
Would it look good tho? No 
Yep
Ye
Y
Metal with water running through the small tubes
Small reservoir at either end
Very simple design honestly
also do either of you have tips for managing rgb fan cables
i have all 7 of mine into one hub
but its just stuffed into the back i like
Zip ties
could not keep on it
I often use zip ties yeah
Or twisties
But you have to be ultra careful
I zip tie to the case normally but not tight
Too tight and it's cancer
i want one of you to come do it for me cables are my worst nightmare i just cant handle managing them for some reason
i should of done it as i went
Sure just pay for my plane ticket and I'll be omw
yes i want felix to come roast me irl and not in discord!
Let's see short notice plane trip to dallas
This friday, from a major Au hub
$2200 usd pls

wrong state
Well I gotta fly to one of the international hubs
Ok well then we have the connecting flight from dallas
i havent flown before idk how that works lol
Don't you want to look at my back panel before you make that offer 
I promise it doesn't look that good either lol. I've made a lot of upgrades since the initial build and I kinda gave up.
Hope I don't get flight emails now
i cant remember if ive seen your build before or not
My inbox is pretty sanitary don't want to trigger a mailing list omg
tourist destination ohio
It's independence day this week so that's about right
Oh you've seen the front... not the back
Ironically it's barely any different than the price I paid to go a few years ago
is it white with like 4 monitors
It's in fact cheaper
No, it's the Torrent that I butchered into fitting a 420mm AIO that doesn't actually fit
doesnt ring a bell ngl
You pc reminds me of a semi
Lol
brb
idk I guess Iโm used to seeing glass boxes like mine with rgb
That board is kinda sick
fractal torrent
The Torrent still is at the top of almost every performance chart
am I tripping or does your 30xx have green rgb on the top or is that a reflection
Mostly because it brute forces its way to the top with 2x 180mm and 3x 140mm fans
Yes on the high tier FE cards that LED is ARGB
Huh I never knew that was a thing
3070 and lower it's just white
woah
Thatโs neat
You just need to use the EVGA PX1 software to adjust it, or openrgb
Those are the only 2 I know of that can adjust it
I used openrgb to change the color dynamically based on GPU hotspot temps
fancy
And the bottom fans on my case run based on the GPU hotspot temp too, thanks to fan control
I want to actually spend time to learn fan curves and proper software but I donโt think Iโll ever be able to or really need to
Fan control has gotten easier, and you can hook into hwinfo for extra sensors like the GPU ones
My image of the board has been ruined ๐๐
how to fix this display thing? im pretty sure its the gpu being slow, but i updated drivers for it and its still like this
asus rog strix rx580 o8gb
I think your gpu is artifacting, might be broken
hi :)
i recently built a new pc, around march 19th
ever since then ive noticed weird screen flickers.
it happens usually when i mjust clicking around the desktop, watching youtube, doing work in tabs
sometimes both monitors, mostly just the primary one though.
its usually a split second, but sometimes stays on screen for up to 1 full second.
its usually some pixelated checkerboard blocky black squares. sometimes some triangles in there too.
pictured is a very horrid representation similar to what happens. but every time its somewhat different.
i did scrape i think 2 pins on my gpu upon insertion. but its never really been an issue apart from this? games run fine, my fps is great
one of my friends i think had a similar issue, but he said ram fixed it so im not sure what to make of it.
i have a zotac 4070ti btw. :)
any help?
oh, second note:
sometimes it will capture on a recording if i use nvidia. i can see if i can catch it rn
My guesses are either ram instability, windows corruption, graphics driver corruption, or GPU instability (more likely vram than core)
I would recommend running dism and then sfc, then try ddu and testmem5, if ddu fixes it, that's it, if testmem5 comes up with errors, you'll need to mess with the ram and that'll likely be the issue.
does that weird pattern cover the whole screen or just a small section(in a corner maybe?)
okay, ill try all this in a second. :)
thank you
most of the screen yes
what cards do you guys recommend for a budget of 150-250 usd? I'd like to see some recommendations between those ranges. New or Used
Used, 5700xt, 1080 ti, 2060 super+ and 6600/6600xt
New the only real option is 6600 for gaming
thanks very much
Ah so Id recommend looking at the other components, test against a different monitor and with a different cable. If there's another video out port, use that too.
If it were a smaller area I'd worry about the card like bad memory or something but that sounds like something maybe in the monitor is acting funky
how would you rank them from first to last?
5700xt/1080 ti/2070 super/6600xt are all about equal
1080, 2060 super, and 6600 are all close enough
whats the best sub $200 gpu? ideally as cheap as possible
6600, but if you'd buy used there's other options
5700XT at $120 is untouchable
where?
Used
which brand of 6600 is best? if it doesnt matter much, which ones should i avoid?
XFX and Sapphire are generally the go-to for Radeon cards
i had an asrock 6700XT last year that treated me well after I sorted out some driver issues
maybe avoid MSI if you could? ive heard mixed things about their radeon cards iirc, ask someone more knowledgeable if you like
generally whatever is cheapest should probably be fine, run it by someone here if you want to double check
For a 6600 I'd say the cheapest one
Also at ~$30 more a 6650xt might be the better buy
Is GIGABYTE GeForce RTX 4080 SUPER GAMING OC 16G good? I read a lot bad reviews about it
Just bought mine for 750&
For 750 yes
Use an anti-sag stand or something makeshift to prop it up
Gigabyte has problems with sag on their cards
It is very big. Any suggest for vertical mount kit for that big boy?
$7 amazon magnetic stick
Oh vertical mount
Ezdiy-fab is nice
Cooler master vertical mount v3 kinda nice too
its only like 15% performance though right?
I wanna say more but not quite sure
if you can fit it into your budget it may be worth spending the extra for it
im on a pretty tight budget unfortunatly
if you cant strech 6600 is a solid card for its price
there is always the used market too
i really wanna avoid the used market tbh
i dont wanna risk the part just not working
I cant find hags on windows 10 with my 6800xt
is gigabyte 6600 eagle good compared to other 6600s?
Iโve heard mixed things about gigabyte cards, again, likely whatever is cheapest or looks the best to you should be fine
I paid the premium for a XFX 6950 XT over the โFoundersโ version from AMD since I liked the look and design of the card better
I bet if I got the cheaper one it would of been just fine as well
Narrow it down to like 2 and look at reviews for those specific cards if you want to learn more about them if you are unsure
I remember since I loved the look of the XFX 6950 so much I was looking at videos and articles about it for months beforehand
The reference is actually kinda mid
Specifically the stock pads
how is the titan xp?
a 4060 beats it, and the 4060 isnt great
also its on an old architecure, better off with getting smth from 30 series or 6000 series or higher
what is your budget and currency for just your gpu, and are you able to buy it like rn if you could/wanted to
and are you sticking to new
Just get the cheapest 6600
If you spend any more than the cheapest model you'll see a much bigger gain by going to the cheapest 6650xt/6600xt
Can be excellent, but not worth more than $10 over an alternative
Just don't get a Mech MSI one
I think even those are alright for a low powered card iirc
All the 2 fan ones are more than sufficient
Kenzie haunts my nightmares with the hardware that was abused in their hands
Mech is still kinda bleh even on 6600 ๐
Well at the cheapest price it's the best imo
The only 3 fan
Cooler at a similar size (Not cooler density) to the 6700XT variant
yeah I'd take it over the challenger D. not a bad 2 fan card by any means but they're the same price
also the challenger backplate is so ugly edit: maybe not all models?
Challenger is like the real cheap option beyond mech
yup, here's the 6600 one
Good card but asrock has trouble silk screening the core
yeah better flow through
The right side has flow through, also the edge around the front allows for better ventilation
It's functionally smart for being cheap, they just can't silk screen paste properly over at asrock 
Ask @wanton edge he has one of those coolers
Iirc
eagle is a great looking card imo. Surprised it's the cheapest rn
It runs like completely MPT unlocked at some 60-70C
It's a very good cooler
Don't expect an amazing bin from it though
It's just very quiet
Ironically way overbuilt when they had 2 fan Eagle 3060's at the time
that sounds a bit like my previous card, which was a gigabyte 3 fan 2060 ("windforce" I think). Never was able to OC very well but super quiet
what is bin and oc?
Oc: overclock
Bin: some chips of the same sku/model are slightly better than others and/or will overclock better than others
Manufacturers use the best parts of the silicon to make the high tier products, like i9/r9 cpus and xx90/x900 gpus. The silicon that isn't quite good enough for that will become a step lower, like an i7/r7 or xx80/x800
Some of those lower binned chips might just need a bit more voltage or something than is within spec for the manufacturer, so by fiddling with it yourself you can get a decent performance increase
Other binned chips will just run as intended for the product they end up in and won't have as much OC headroom
6600 is a little faster I think, probably pretty close though
Oh this says 5700 is faster
pinned and hasnt been updated in awhile
imo just get a 6600, its on a newer architecture and will be supported longer
a good deal on a 5700 XT may be good, but if i had to pick between 6600 and 5700 non xt, 6600 easy
either of those would be perfectly fine
cheaper+better cooler
https://pcpartpicker.com/product/G9ytt6/gigabyte-radeon-rx-6600-8-gb-eagle-video-card-gv-r66eagle-8gd
im buying used now
A friends' GPU stopped working suddenly, no signal at all, and he already tried different monitors/cables/ports.
He brought it to me to see if I could work on it at all, so I put it into my second pcie slot and plugged it into my second monitor. I'm also not getting any signal and it isn't showing up in device manager nor the nvidia app. Not sure where to go next so please ping if you have any ideas, thanks!
Also the fan is not spinning
Sounds like a fuse went POP. This is the kind of thing that happens when a short happens. If he's lucky nothing but the fuse blew but it would take an expert to diagnose and repair it.
Oh and if there's a short then plugging it in can cause irreversible damage
ok, thanks. I'll tell him that.
is new architecture a big deal?
In this case, not really
But in theory you'll get driver support longer on a newer GPU
This might be the wrong channel to ask this, but i have a 3090ti that is water cooled, i recently upgraded the thermal pads in the water block, when doing so i didn't realize the CPU on the graphics card was not touching the thermal paste (due to the thicker thermal pads). Now when plugging in the Display cord i get no response from the GPU, what are my best options for trouble shooting?
because you can edit a message
i know lol, it just looked silly
btw the die on the graphics card is still called the gpu; it's not a cpu
and there is a chance that the gpu might just be dead, im not sure though on how you would troubleshoot that exactly
@tough plinth when i mentioned newer architected, falcie said what i meant, it will technically be supported longer
ohhh ok ty
So, going back to my previous issues with my gpu, ATX 3.0 PSU solved heat problems, stability problems, and performance issues across the board.
Really??
This was the intel cpu yes
I want da 4090 so bad T.T
As a 4090 owner, I prefer the XTX in my other rig....lol
I paid half the price for the XTX and I get near 4090 perf
Yo Dr feet
just want to say my 7900xt is stable after 3 hours of gaming on cyberpunk 2077
Nice man. So all going well?
I'm a dentist
Damn
Yes. No hard crashes or anything at all
Rt ultra 60 FPS
100 FPS ultra settings
I'm okay with it. Story was enjoyable
Also I noticed the Hotspot temps
It's much better compared to merc 310
Hwinfo64 reported 63C hotspot while i was playing Cyberpunk 2077
However something unexpected to the 79xt and it went away with reinstalling driver with touch of DDU
I have three monitors and they all work, so I unplugged 3rd monitor to test PS5's input. played for 30 minutes. took it off. Screens flicker (i think thats normal when you plug it in)
But the #2 and #3 didnt come on while being plugged in, only main monitor came on. I couldnt use AMD adrenlin software. Opening up Device manager it revealed display adapter is Mircosoft Basic Adapter. So did the usual, DDU then Driver only AMD graphics driver and it was fine again.
Yeah for sure it's odd but it's still working as expected and I'm relieved that I don't have to RMA it
Heโs a dentist for all but an orthopedist for me
@odd siren best foot doctor there is
Yes, this is my rig with the i9 13900k that was having gpu issues
My cpu is getting clean power and dropped from 80c in game to like 65-70
And I actually am getting better performance out of it
I went with the Corsair 1200W shift because I wanted the headroom
If you havenโt looked into it already, windows has a tendency to overwrite display drivers, I believe there are some work arounds on windows 10/11 home but pro allows you to edit update policies through group policy. Which is what I did
That's perfect I'm on windows 11 pro with altasos optimizations installed
I'll go do that group policy so it don't pull a amd failure jump scare ever again.
Thanks
Np
Gpedit only works if you have Pro Windows or higher feature tiers, just fyi. If you have Home Edition it won't work.
If you do have Pro then the key you want is: Local Computer Policy > Administrative Templates > Windows Components > Windows Update > Managed Updates Offered from Windows Update > Do not include drivers with Windows Updates > set Enabled
That is actually really interesting. What was the old psu?
Imagine thermaltake smart
asus Thor 1000w
Im gonna blame the transient power spikes degrading the psu and causing it to have unstable voltage delivery
Oh do you use one or two eps
two
i can send the old unit if you were interested in testing it
it worked fine with the 3080Ti though which is weird, the temps for the cpu were going up though over time
they dropped back to where they were when the psu was new with the atx 3.0 unit
Appreciate the offer but I'm in Australia mate
Haha
I'd enjoy messing with a thor, if only
Is it just a pain to ship something due to customs?
Very very interesting behaviour though, maybe the Thor EPS were degrading/one was
I've seen what dirty power does, so I kinda believe the assumption you're making
Yeah, I could do some testing on it and identify if/which cables are causing the issues. If its the EPS then I could use it in another system. My girlfriends old pc has an 8th gen i7 which I could try to turn into a local server or HT pc
I mean it definitely has me interested tbh, I genuinely wonder if it's a warranty-worthy behaviour
Am I allowed to ask for help here or do I have to do it in the post maker channel?
You can ask
really trying to decied right now for my $1500 bulid what would be more optimal to get the 4070 ti super or the 7900xt or 7900 gre
7900 gre is best value
7900xt would be better value than the ti super
And still good value
$50 over budget
https://pcpartpicker.com/list/s8dNt7
Part List - AMD Ryzen 7 7800X3D, Radeon RX 7900 XT, Phanteks XT PRO ULTRA ATX Mid Tower
Cheaper, either drop gpu to 7900 gre, or cpu to 7600 and either use stock cooler or whatever $15-25 tower cooler you like (evo dark from thermalright is great)
alright thank you
ill probably go with this but I already have a mobo
its tell me this - The MSI MAG B650 TOMAHAWK WIFI ATX AM5 Motherboard supports the AMD Ryzen 7 7800X3D 4.2 GHz 8-Core Processor with BIOS version 7D75v13. If the motherboard is using an older BIOS version, updating the BIOS will be necessary to support the CPU.
The mobo is pretty costly and doesn't offer a performance benefit over the b650 eagle ax or b650-p wifi, but it shouldn't be an issue
oh dope
Just make sure they work before you purchase
how would I confirm that
Ask for a video with it working
Though I can't really be one to talk, I bought five parts used w/o proof and the mobo is dead
Hey @odd siren
When we were talking about my gpu the other day I forgot one detail
When my pc freezes, the power button doesn't work either
The only way to shut it off is holding it
Which makes me question whether or not it's the gpu
It's various times, just now I turned my pc on, opened discord and valorant, the icons showed up at the bottom and then couldn't do anything else, couldn't even see the apps, but my cursor could move
Before it's happened multiple hours into a gaming session
It honestly sounds like ram to me, but there's no reason it would be having issues
I might get ridiculed for it but
I think the 4090 is overkill if you don't do content creation and/or streaming
Only for gaming it's just crazy
Even then it's a stupid purchase imo, unless you get paid very, very well for said content creation
Unless workstation based or that, I would never spend $650 for a 30% difference
I mean, can GPU driver crashes cause OS halts? I feel like it can with edge cases but idk.
valid opinion
i think each gpu maker should have a halo product going forward for those who will buy, but instead focus on making the entry-mid range actually good
4090 the content gpu
Arguably true, I'm just saying that the 4090 can make sense, but not for gaming
i thought you were saying what i said
yes 4090 has valid reasons
Iโve had that issue before on unstable ram overclock
upgraded from a 3080ti to 4090 surpime liquid x, made sure to uninstall previous drivers using DDU and clean install the new ones. I notice in some games, specifically UE5 I believe, that around 5 or so minutes in my pc starts freezing completely, or getting like 5 fps. Not just in game, but after that it appears to run normally. There are also some games that randomly start freezing for a second or two and then are back to normal, any ideas what the issue could be from that?
Gpu overheating perhaps
U should send the specs of ur pc
def not overheating, temps stay <50
4090, 7900x3d, x670e asus crosshair MB
@mellow glen if the software was running fine with the 3080 ti the gpu may be defective (just a guess)
Iโd reinstall drivers to see if it fixes it
anything that could have hypothetically went wrong during install as an explanation ? was gonna go back through and make sure all connections are ok. I also notice when this happens the GPU loses wattage, from my UPS when I am in game I would pull ~5-600ish from my entire setup, when this happens it drops to 280-300.
@mellow glen first question: likely not. What wattage is your psu and id try undervolting it for now that might help
That could also explain the frame drops and freezes bc of the sudden wattage drops
1000w
My guess is that the gpu clocks down when it reaches 600w for a certain amount of time bc itโs only a 240mm rad
Not a psu issue then
also saw some conflicting info... the psu supplies 2x pcie cables, which has daisy chain connectors so unsure if that has anything to do with it
hmmm the weird thing is that it only happens on startup in the case of UE5
let me rephrase, it only occurs after 5ish minutes so not startup. doesnt matter what I am doing in game though, could be idling in main menu could be in game
Ignore this, I misread 600w as the gpu power draw
gotcha
Sorry but im not sure whatโs wrong tbh
One of the techies with the yellow names might be able to help
No worries, was wondering if anyone had a similar experience or if these were common symptoms of something obvious.
Has long u used two separate pcie cables itโs fine
2 separate cables, but 2x 8 pin on each cable since 4090 requires all 4
I just googled ue5 4090 lags and it says itโs normal sorta
Ue5 pushes the 4090
A lot
lag is one thing but this is SEVERE stuttering and system halting
Ram?
64gb ddr5 (2x32)
how much?
Just lower it down to the lowest to see if it fixes it first
Then raise it back up in increments of 10 then 20
did not fix, was -500 clock. while this happened GPU load% went from ~80 to 0.
temps dont spike, fans dont stop spinning, power drops
@mellow glen r u sure u have enough ram then? Thatโs the last thing I could think of
100%, nothing other than the gpu changed since the issue started
which was immediate
could run memtest to check if there are stability issues
U can download furnark to stress test your gpu around 30m to see if itโs fine and working
That as well
really hoping there is human error somewhere rather than the gpu being defective lol
Btw make sure hot spot temps r good just in case
how can i check that
nvidia app + nzxt cam
Idk if those check it or log hottest temps recorded
I think nzxt can should have the hotspot temp
And nvidia app
Iโd download heinfo64 if do stress test tho just bc it records the highest temps reached
Windows 32bit/64 bit is the one to download
alright, will see what hot spot is rn while it happens after launching
not the stress test, but the application where the issue occurs
hot spot temp max was 44.5
Thatโs fine when idle
thats what it was when replicating issue
Do u have steam?
yeee
Thereโs a free version of 3dmark on there and u could run a benchmark then compare it with other 4090s
Or u could use furmark to stress test the gpu for around 30m
I've done that actually, nothing out of ordinary. within margins of others according to 3dmark
also the issue doesnt happen with that
Furmark can kill a 4090
Oh
LOL
It has a tendency to make the GPU ignore power limits so fuses go pop
anything glaring or suggestions that you can think of ?
I just got here let me catch up
no worries
let me know if you want a video demonstration lmfao
Have you installed chipset drivers
I believe so, did that after the issue started happening
Idk how heat sensitive ddr5 is but if the gpu rad is an intake maybe the internal temp of the pc case got raised causing instability - I doubt it tho
What model PSU
the 240rad replaced the 2x rear exhaust fans I had
it is used as exhaust
Corsair RM1000X
using the supplied adapter from MSI
and takes 4x 8 pin connections
Which ports on the PSU are you using, and did you use 4 separate cables or daisy chains
2x daisy chains
Do not daisy chain a 4090
Corsair cables won't melt but the voltage will be unstable
Keep in mind each cable is specced to run at 150W
Corsair overbuilt them so they can do 250-300W but you might lose stability
I had similar problems with my FE 3080 ti on a RM850X and getting a dedicated cable helped. I gained score in benchmarks changing nothing but a cable.
hmm okay, was thinking that initially which is why I mentioned conflicting info. some people say absolutely not, some people say its fine as long as the cables are supplied from PSU directly
The conflicting answers are what happens when something is perfectly acceptable in theory... but in practice there's more variables that can influence results
Another thing to double check is that your UPS is rated to output enough wattage constantly
I just saw you had one
1000w ups
That'll be peak, what's the constant load rating
If you don't know I'll check by model number
CP1500PFCLCD
Based on CyberPower's own charts, you'd be getting about 5-6 minutes of runtime at your load level, which suspiciously lines up with your estimated gaming time before it slows down.
You can always test if it's the UPS by plugging straight into the wall, at least for a little bit.
interesting... it doesnt have problems holding the 500-600w during long periods of use though, and this specific issue doesnt seem to care about the wattage, I can be mid game which is when its pulling the most power or just in the main menu when im <450w total. only thing that is constant is that its around 5ish minutes after launching the game.
will still test that though
and grab that cable as well
if that doesnt fix it you thinking it could be defect?
Possibly. I'd like to narrow it down a lot more first before making that call.
fair enough
Oh are you just watching core temps or are you looking at VRAM too
I know some of those liquid cooled GPUs don't cool the memory well
was watching these:
Second one, Memory Junction Temperature
was lower or very very similar to the hot spot temp
nothing alarming or significantly differnet
Ok the memory doesn't throttle until 100C so as long as it's below that
Alright it's 3AM so I need to get to bed
no worries - will report back after trying those two suggestions
should get cable tmr,
?ddu
DDU (Display Driver Uninstaller) Guide by Pato
https://github.com/NeweggTechie/Graphics-cards/wiki/How-to-use-DDU
@odd siren I'm trying to ddu and can't find safe mode in bios
Everything is telling me I'm on the right page
And it's just not here
Been at it for 30min and I'm losing my mind
I want to break something
I've turned my pc off and on over 20 times going back and forth between windows and bios I want to cry
Easiest way is the restart method within windows ime
https://support.microsoft.com/en-us/windows/start-your-pc-in-safe-mode-in-windows-92c27cff-db89-8644-1ce4-b3e5e56fe234
Find out how to boot into safe mode in Windows from Settings, the sign-in screen, and a black or blank screen.
Yeah I found that part in your ddu guide and it helped me keep my sanity
Ty
"Found" being used loosely
Iโve got this laptop with a 3060 that is capped at 210mhz and is misreading its wattage. Nvidia, msi afterburner, and gpu-z believe itโs consuming 752 watts. Laptop is plugged in and every โperformanceโ setting is set to max and no the Gpu is not thermal throttling at all
Itโs a cool 39-40 degrees
And itโs not like the Gpu just isnโt being stressed because games like r6 at the absolute lowest settings run at 20-40 fps
What does hwinfo64 say? It kinda sounds like it doesn't have drivers rn though
Give me a minute to run that I havnt checked it
I'd check if you have drivers installed first
Low usage + no reason for throttling + low performance is usually either no drivers or bad drivers
Drivers couldnโt be the issue. Iโve uninstalled them and reinstalled them twice
Iโve also reset all bios options and reset windows
Try rolling back to an older driver version
Where can I find older nvidia drivers?
I forgot where Nvidia puts it on their site, but TPU has them if you scroll down to version history and choose a stable version a couple of months old
https://www.techpowerup.com/download/nvidia-geforce-graphics-drivers/#changes-556-12-whql
Power draw since you brought it up before, 750w sounds crazy, and the frequency, 210mhz sounds too low unless you're idling doing nothing
Also, if it's on a laptop, what Optimus setting do you have set?
Optimus?
The fancy name they gave to whether or not to use the igpu and switching between the igpu and the dgpu
Ah
Iโve tried both โhybridโ and discrete only
Neither changes the clock speed
Prob use discrete only for now
Watch the clocks and open up a game and see if they increase to a normal freq
They donโt. Iโve checked before
The only thing that boosts clocks is manually โoverclockingโ with afterburner
But
It doesnโt even โoverclockโ to its base clock rate
And it wonโt turbo above what itโs set to
Does it let you access the voltage/frequency curve on afterburner?
Oh great
Would've liked to lock a v/f on the curve in afterburner to try to force it
Btw have you used DDU for uninstalling drivers previously?
I did it with control panel the first time then did it with nvidias cleanup tool
?ddu
DDU (Display Driver Uninstaller) Guide by Pato
https://github.com/NeweggTechie/Graphics-cards/wiki/How-to-use-DDU
Try using ddu
Boot in safe mode?
It is in safe mode
Nvm I figured it out
To extract ddu you have to launch it
You canโt just use 7z
Is this the right one for Mobile Gpus?
installed the drivers and it still doesnt work
Unfortunate
Maybe something vbios related or something, I'd try to warranty it if possible at this point
Idk if I can ๐Iโm pretty sure this laptop was bought like hardly more than two years ago
Thatโs honestly really sad
Maybe it's still in warranty? I dunno what else to try rn, maybe other people have more ideas
Sounds almost like an optimus system that's got the optimus bug (Stuck in iGPU no matter what)
no
it is using the gpu
and also using msi afterburner to "overclock" it works
but it doesnt properly "boost" or "turbo" itself cuz it believes its at the 100% peak of its capability
So if you check hwinfo, the igpu doesn't come up at all?
idk
i havnt tried but im telling you the igpu is not the issue lmao
and the issue you described the igpu could cause is not what im experiencing
Sounds really familiar to a problem with an optimus laptop I just tried to help with like a week ago
Hence me asking
the gpu was stuck at 210mhz and 752 power?
It was stuck at 210mhz yes
Laptop would only use the igpu
Nothing would make it switch over to dgpu preferred
It ended up in a similar scenario, after regedits, ddu, optimus settings, forced cmd overrides, display adapter resets, etc.
We opted for warranty
Genuinely ran out of ideas
If you can see it's trying to utilise one or the other then I could be a bit more conclusive about this, but it still sounds very similar
210MHz is the idle clock for ampere
752W will just be a readout bug
Mux?
Mux means a discrete complete disable of the igpu to boost dgpu performance
You can know by googling the name of the laptop
It will say in the product advertising
Costs extra to have one
Only other thing I can think of is a complete wipe and reinstall of windows or using nvflash to redo the vbios off the top of my head
Done a reinstall of windows
Using device manager to disable/reenable may also work
Reset is not a wipe
Reset carries over registry errors
Vbios flash COULD void warranty even though thatโs technically not legal
They would if I messed it up or if it just unluckily bricks itself
I would be surprised if that happened
Usually the nvflash will pop up if there's something really amiss
It will write but not complete if it detects a mismatch or error
I have wiped the m.2 drive from everything and did a fresh install of windows
Yes
Although I couldโve missed something in the bios cuz this msi laptop bios highkey sucks
The categorization is kinda bad
Ironically enough, the optimus issue on the other laptop was an msi laptop...
Ge76 raider
Quick google of similar specs gives this answer or a broken mosfet
Also try nvidia driver 462.21 specifically
Ok
It's a power delivery chip for the gpu
Yikes
Ah
Unless there's, very few
Like 3
Lol
Which it won't be
On a gpu like this at 145w tdp it'll be minimum 6 or min 250A
dang
But your exact laptop
Exact problem
Someone tried 462.21 and got it to work
Ddu first
Where would I be able to find 462.21? I only see 465.89 then 461.92 on tpu
Shouldn't
ok
Win 11 is just win 10 with a mask, as is win 10 being win 7 with a mask
Someone also suggested that geforce experience could be the culprit forcing minimum power
Don't install gfe
the one you sent me is 462.31 and the one you said is 462.21
and i never had GE installed
i do have the nvidia beta app which might be the same thing?
so i wont install it this time
So do I need .31 or .21
.31 will be basically the same as .21 but with a hotfix
Mk
This one was also targeted at msi laptops specifically from nvidia
https://www.guru3d.com/download/geforce-466-55-hotfix-driver-download/
Getting a lot of results now that I have the laptop name
Common result is not to install GFE
GFE seems to be very commonly related to this boost lock issue
3060 specifically
Iโm guessing the nvidia app is just a reskin of GFE?
Pretty much
Control panel is fine
You need that one
Rollback of bios is suggested
Fixed someone with the same issue
Bios or vbios?
Bios
I donโt have a usb on me
Fast startup may also do something
So Iโd need to do that tommorow
Download straight to C drive
You'd need a fresh vbios off of tpu
Exact clone
Iโll try that out when I get home
Yeah so 462.31 doesnโt work with either my version of windows or my Gpu
which makes sense
its for 20 series cards
I looked up 3060 on the nvidia website for that one
It's compatible
What's happened is it's not detecting the gpu at all
i.e it didn't come up, it's now using the igpu
Try plugging it into the charger, restart and try again
3060ti supported but not 3060

