#general-chat
1 messages · Page 257 of 1
is 3700x decent compared to 5700x
@mortal grail nice 5700x guy got back to me
200cad i can prob get lik 175
i think worse than 5600x
oof
@mortal grail can you do a small favour
Can you make me a build for $900 and a build for fiancé at $800
how did my c drive go from 182-178 gigs... ik storage fluctuates but damn
TV was fixed.

🤔
Can anyone provide some help
You might wanna provide details and I recommend we pop over to #building-and-recc-chat
Bet
the 0x129 microcode update is insane, i managed to pull a greater undervolt @grand kernel @idle nexus @mortal grail
gonna see if i can go lower
red or blue ?
Blu
good
violet
good
ok it went down to 131w
free AMD Zen 3, 4, & 5 performance uplift apparently https://community.amd.com/t5/gaming/ryzen-9000-series-community-update-gaming-performance/ba-p/704054
The new “Zen 5” architecture first introduced in high end laptops earlier this summer is now available in a full stack of desktop processors with the AMD Ryzen 9000 Series. And while “Zen 5” is delivering leadership performance across content creation, productivity and AI applications, we’ve heard f...
cant feel it
you feelin it now, mr krabs?
lmao we need to talk https://en.wikipedia.org/wiki/Rachael_Gunn
Rachael Louise Gunn (born 2 September 1987), known competitively as Raygun, is an Australian academic and competitive breakdancer. She is a lecturer in the Department of Media, Communications, Creative Arts, Language and Literature at Macquarie University Faculty of Arts.
Gunn gained media attention after competing in breaking at the 2024 Summe...
memory integrity disabled...?
thats kinda insane that it has so many reviews in the past 3 days
Where
lol uh did you read it?
the Windows Insider Preview ISO
supposedly has like optimizations for branch predictions
can you tell I'm parroting what I heard? I don't actually know what that means; how it would help; or whether it even could help lol
whoever remembers my audio issues ive been having
i plugged in different headphones
and i have audio in the games that werent working
like my old turtle beach ones i used with my xbox
I didn’t read
I’m dumb
TLDR just read here
Thought that was just for 9000
How tf do u beat the valkyries in god of war
I have to go to easy mode and it still takes me 10 tries
Yolo it
And i have to revive each time
Hello
Thanos
screen rant wants diversity in this game lol xD
why is that even a xD requirement in a game
there are a bunch of animals and mythical entities in the game
what are they smoking is beyond me
So an update for the LG TV. Techician said something inside the TV came loose
But alas it's fixed
Thanks windows I hate it
Am I blind? What's wrong with it?
i think task manager tricked him
it seems that he's running 32gb ram instead of 64
supposed to be something like this
heh
hol up
64 gb installed. Bios shows 64 gb. Windows says there's 64 but 32 is ""hardware reserved""
Its ether windows not allowing me to use the other 32gb of ram, or something is damaged on my motherboard because I can confirm all 4 sticks work, and are the same make model and speed
Via is just done through the browser
Chrome specifically
ah nice thnx
connectx4
goes from my desktop to server
then my server connects to my switch which connects to the rotuer which connects to the ONU which provides all the interwebs
at 2.2GBPS
Anyone here use a Acer Nitro V?
some will say overkill
i see...
i guess check system info
no but whats the question
thats alot of tests heh, very quick speeds
ye
they should've included other animals ig
does this make your GPU run at x8?
Palworld FTW
except I have stupid GamePass version
so none of the .pak mods work right lately
cuz some IO Store thing they changed to on GP version but reverted on Steam version
.lua mods work fine but .pak is .b.o.r.k.e.d
it's fine that's just it chillin cuz it's not doing anything
if you click that little questionmark there
you can run a like tiny render thing so it'll show you what it runs at when it's actually doing something
still showing 4
i read it backwards
mine just started 2nd grade
i also run z790 so maybe thats why it doesnt go down?
don't worry it sucks but it doesn't last thaaat long
cause that network card is running at x8 too
damn that's pretty nice
what if you like, put all 4 LAN connections from your router? would that make it any faster? lmao
its 2x sfp
so it wouldn't do anything
at most i could do full 50gbps from server
I've been meaning to set up a server lol
mines on a 5900x and i still want to switch to something with IGPU
but havent done it
server has 72tb storage too
128gb ram and 4tb nvme for game servers
kicking my ass lmao
what games do you host?
Worth it
I had a 5900x MC server for like 3 days lol
I want to make another MC server but I’m waiting for my friend group to get back together
@idle nexus https://images.morrenus.xyz/reku3/sUWeJIZi85.png what do you think
Minecraft Satisfactory and 7DTD
we're terrible about starting a MC world and playing for like a couple weeks and then never touching it again
put all that time and effort into it for nufin
425w TDP?
I could host a server for Newegg but I would probably break this server’s rules in the MC server lol. It would have to be a sub server but I’d imagine that would also be against Newegg’s rules lmao
no real worth
i had one up afte rforks took down
and it sat empty for 4-5 months
Yeah there was one up a long time ago as I recall
this 3080ti is 400w & I always have it at max which is 450w
I remember that one lol
my 4090 hit 746w
I mean I could host one but it would be in a sub server
i could too atm since i have ways to mask my ip better now
I would just host mine via Tailscale lol (more secure this way imo)
right now its sitting at 475
also this is funny tho...
50 more watts 4 degrees more
damb what kind of cooling are you using? 💀
3090 has that 360 rad
Gotcha that makes sense
the case is a neso p1 from gamdias
still need to clean the case and take pics
man So I bought this Razr+ 2023 phone a while back and the first one screen broke (not because of me, like it glitched out) in a month, so I returned it but had to pay like $325 to get a replacement. The new one charger port broke after a week. I left it and just used wireless charger cause I wasn't gonna pay another 325 for this dumb phone. I finally decided to just buy a port replacement and replace it myself. I've done charger replacements before as well as many other phone repairs and never had an issue. It came apart and went back together super smooth, no issues. Charger works now. But neither of the screens work at all now. The screen stuff is all in the top half which I did not take apart. I've done everything I can think of to fix this stupid thing and now its a $1000 paperweight. 😩 or $1325 if you count the replacement cost of the original. I'm never buying Motorola again.
I need to take my GPU out & redo memory pads
you gotta go to headquarters to find light switches in order to turn em off
now imagine someone so mad at intel they just sneak in and just keep turning off lights around the office building
AMD 9000 series didn't get a good look after launch
very little improvement it seems
well better than a slowly degrading 13700kf
best part i just need to swap my motherboard and cpu
no ram needed to swap
maybe cooler
tho tbh kinda wish there was a x3d version taht would be worth getting
is there?
i dont really follow stuff anymore
7000 series?
7800X3D remains king afik for gaming
ok i may look into getting that as a replacement
now whether that changes when 9000X3D launches remains to be seen
is there an eta on that?
given that the 9xxx non-3d haven't been all that impressive
what motherboard would you go with
ah crap i realized something
i have 4xnvme in raid 0
ASUS TUF GAMING X670E-PLUS
seems good
we may get a 9000 x3d chip maybe late this year
honestly amd should've released the x3d lineup first
._.
there are 4 NVMe slots in my X670E Taichi that's reasonably priced
yeah the x670e plus has 4
X870E should be out soon but the only advantage to it is newer USB thing
@idle nexus Please help me... If you can. Remember the GPU mem bluescreen issue?
well, its back, and its not going away...
have you checked for gpu driver updates
what is the actual error code
MEMORY_MANAGEMENT
this is the third time its happened. 1 time a week ago, and 2 times back to back now
Ok, just happened a third time in a row. So 4 times now
try MEMTest86 and make sure you get 8 full passes
As much as I have avoided doing that, looks like it's my only option. Time to find out if I need to buy RAM I can't afford, an Mobo I can't afford, or a new GPU I can't afford
you could also check event viewer, click find, and search for memory diagnostic and see if there are any errors.
windows logs>system, then click find
Oh god, my event viewer is an absolute nightmare of errors and warnings
oh my godddd, thousands and thousands of them
Oh dear
16,345 events
Event viewer jumpscare
I don't even know what to do right now
I already know either my RAM or my Mobo is faulty
check to see if you have any windows updates
ok did you get any ram errors in your thing
that's when you search for memory?
I have no idea how to even search in here 😓
oh, this
sorry
The computer has rebooted from a bugcheck. The bugcheck was: 0x0000001a (0x0000000000042001, 0xffffad8eda486650, 0x0000000000000000, 0x0000000000000000). A dump was saved in: C:\Windows\MEMORY.DMP. Report Id: abe893cb-8ce4-413a-be52-87ffea326a32.
yeh cause I have some errors in mine too all throughout but my pc work fine
yeah, but do you have like... 100 in one minute?
no 😅
here is a list of things you can try: https://www.drivereasy.com/knowledge/memory-management-blue-screen-of-death-on-windows-10-solved/#f1
if those don't work I would do memtest, then if that doesn't help or fails, take a stick of ram out, check, then swap, then check, then if you have alternate gpu or integrated, check without gpu, etc
I would not be surprised if it is GPU driver related. I have been apprehensive to upgrading my GPU driver, as this is the ONLY ONE I have found that doesn't have dual GPU IO drop out issues
yeah
why
Machine learning and AI tasks
oh neat
I am trying to get a second 3090, but I am worried this 3090 is dying
4090 is meh for AI
lol it was a joke
4090 is a horrible value
its a ever so slightly faster 3090 that costs like 3-4x as much
I have a 3090 and a 3060ti at the moment
I could be wrong but I believe windows doesn't play well with dual gpus so that might be your issue
I have had it like this for over a year, and I have had some IO issues for a while, but that was NVIDIA driver related (its no better on Linux)
I just happened to update to this driver, which hasn't had any of those issues, but I am worried there might be something wrong with it
aside from that, there is for sure something wrong with my Mobo/RAM, that I know for a fact
I will update my GPU drivers. Even if it causes IO issues, I would rather that than not being able to use my PC at all
I had a long stretch of no issues. Like 2 years, and then since I have had a dual GPU system, I have occasional pop ups of issues
nvidia's site sucks for gpu drivers now
jesus christ
I had the IO issues, which are well known and just a down side of having dual GPU
And now I am having my mem and (potentially) GPU issues at the same time
my bad I was going to war with a Roku real quick
I think it's time for update lol
what's the issue? what does event viewer say?
nvcleanstall is best
@tacit geodeI don't even remember you by the way. Did you used to go by something else?
Dallas
oh, right
what do Service Control Manager ones say?
In this video, we dive into how Unreal Engine 5's Nanite technology is dragging down your game's performance and debunk misleading claims made to convince you it’s superior to traditional rendering.
This is a crucial topic for both gamers and developers and we offer a unique insight that aims to change how you understand mesh rendering and pote...
I am so overwhelmed with what to do. All of these issues just kinda sprung out of nowhere. I assume they are all related to my mobo/RAM being faulty, whichever one is the culprit
in Windows update, what's installed recently?
these people are complaining it's the result of an update
note the date in the picture lol
this general error happened a week or two ago for the first time, and then stopped
and now its doing it non stop back to back
and my RAM started disappearing like... 2-3 months ago?
well
i have tons of errors too
related to other things
mostly the fact that windows wants to control rgb for a device thats disconnected
XD
DistributedCOM? I hate those cuz they're nonsense lol
sorry, remind me again which CPU?
12600k
yeah I got a ton of distributedcom errors lol
those are just permissions nonsense
Metadata staging failed, result=0x80070490 for container '{00000000-0000-0000-FFFF-FFFFFFFFFFFF}'
wat
i think its fine
i have a bunch of those now too
tho i just updated drivers
hmm you might wanna bump system agent voltage in BIOS a tad bit... have you cleared CMOS & reenabled XMP or anything like that to see if it helps?
heck I just randomly lost internet
computers are heckers
I can;t change voltage at all on this mobo, and I haven't done XMP again cause every time my other half of my RAM dies, it resets
what kind of motherboard do you have?
dell?
tried swapping the DIMMs into different slots to see if it's the DIMMs themselves?
Its intel B760. I can't over volt, under volt, power limit change, boost change, anything. I can basically look at my CPU name and thats it
well some of those start at 150?
do they still make H boards? B760 shouldn't be that limited
only $300+ motherboard i own is a x570master
I have the mobo I have
I haven't checked particularly recently
it was like... 8 months ago?
but I don't think a behind BIOS will suddenly start not working, if the issues wasn't there from the start. Might be wrong on that tho
I am gonna be doing memtest now
oh, I need a USB flashdrive. I don't have one of those at the moment 😭
BSOD
2 separate (I think) issues
MEMORY_MANAGEMENT error
ok i mis understood
it's not an H610 board so it shouldn't be really bad
i thought it was related to the eventviewer
so i was so confused you were going this all out for those errors
the event viewer was just to hopefully see what was going wrong. Event viewer said "yes"
what the heck
I keep losing internet & reconnecting... dunno why I'm trying to send this since oh wait it's back
1: Half of my system memory just vanishes sometimes when I start my PC. It shows only 1 stick installed, it disables XMP, and it doesn't even throw an error or anything
2: My GPU keeps causing BSOD's for mem management whenever its VRAM capacity gets above ~11GB. I am not sure if that is because the GPU is hitting a specific limit, or the system RAM has issues, which is being tripped up on it
it sounds to me like it's time to just burn it and build a whole new one
how much mem does the gpus have
24GB
yes
YES
It shows as still having 24GB, but in programs that use CUDA allocation, it will say insufficient space
For example, an error I got was along the lines of:
CUDA DEVICE OUT OF MOMORY. Tried to allocate 20MiB, available capacity 12.7GiB, insufficient space
Anyone know where to get a recommendation for a good gaming PC that wont be out of date in a year, etc? Want to run Raft, Gmod, AW2, RDR2, Res Evil
so it will say I have 12GB free, but 20MB can't fit in it. And programs that don't use precalced allocation will instant BSOD
honestly exciting to see
same
im always down to see more competition
i would love to try these gpus out
I seriously have no idea what to do right now. I am so defeated. I bought this new PC to get rid of issues like this that plagued my old PC cause of its fault mobo issues. Here I am, having arguably even worse issues now

Might be a program bug
its way worse than that. But I do wish
If it can report you have free vram
But the bugcheck is "no vram left"
Bug is my immediate suspicion
Would be a pretty big bug to be BSODing my PC, and be affecting all prograns
Well...we talked about remapping issues
If not the program itself, or some sort of software conflict
Potentially that, maybe the clock it runs at is too fast
Trips a chip as it pools to vram
Can't recover
Aggressive vram downclock might reveal something if it's stability related
Nvidia drivers don't always recover using a black screen error, if the lockup is big enough the pc will reboot
Takes a very big degree of instability to do that on a 3090 but I've done it before
Multi monitor in fact makes it hilariously easy with surround
I'm alaready so insanely overwhelmed and defeated, and all of this additional information is just overwhelming me even more. Seems like no matter what I try to do in my career, I just keep getting a kick in the teeth every second of every day
I mean, can't diagnose a problem without trying more ideas
fix me
Your problem sounds so unique I doubt any normal tech would know what to do
They'd probably write the gpu off
I am just not sure what all to try. I need my computer to be running, and doing all of these superficial tests will be a great way to waste an insane amount of time and energy that I should be dedicating to making myself a viable career
It just sucks so damn bad
Have you tried swapping to the other ram slots
no, I haven't. Mainly cause I will have to completely unplug my PC, remove it from where its being stored, and then disassemble the whole thing to get to it (both GPU's, and the CPU cooler), which will be a ton of down time
Well, we've been over this how many times now? Gotta start trying these things man
aside from that, if you can get away with it for a lil bit, I would try running a single gpu and see if the problem just magically goes away
I can't offer ideas if we don't try changing boundary conditions
its definitely not dual GPU, that much I know, cause I usually have my second GPU completely powered off
Unplugged?
Then it should effectively be single gpu
Still suggest swapping ram slots, if it trips again, both conditions at once, suspect a vram chip
its likely something to do with that has damaged my drivers from inconsistencies (I was having realyl bad crashes yesterday)
That really does sound like a lot of work to be told its the same things already suspected 😓
Thats what I was meaning about down time
I'm not saying I won't, of course. I am just really defeated right now
I don't have better ideas at this point
If this was an a100 instead, we could consider some sort of ecc error
Even the motherboard is a bit suspect
But it's not the first on the list
Gotta start with the possible small problems
I think this all stems for sure from the memory. There is a severe memory issue somehwere. Whether in software, in hardware, firmware, IDK. But I am hoping its that, and not my GPU dying. Cause if its my GPU dying, then I quit. I will resign from my career fully, and just burn the last 2 years of my work for all I care
Well downclocking the vram is the easiest idea
Download afterburner and stick -500 or -1000 on the vram clock
I am doing some really important work right now, and constantly blue screens are gonna be a nightmare. Let me grab my laptop and see if I can do this stuff on it instead
Tongues are so frickin weird man
https://pcpartpicker.com/list/vJqfN6 @mortal grail how’s this
Part List - AMD Ryzen 5 7600X, GeForce RTX 3070 Ti, Phanteks Eclipse G360A ATX Mid Tower
I think even if the card had a problem, if you found a shop that could repair same day you could get it fixed for maybe a hundred, hundred fifty
Dead VRAM?
hmm... No shot for same day, but I could at least ship it off or something
Nvidia cards run to specific row mapping when it pools, technically if the chip died you're better off as it won't read the vram as there
Ok, I got my laptop. I can do this stupid stuff on my laptop, and go back to crashing my desktop over and over
Change case and it looks fine
Argb-q5
I know for sure that a VRAM down clock isn't gonna fix this issue, but I will do it for the sake of exhausting your recommendations @mortal grail
-502mhz
actually, wait
I can test in a way that won't BSOD me
If it's degraded instead of a poor connection bga then a downclock will reveal this
my training with CUDA allocation will OOM
If you run nand, too hot, too long, it can cause a degree of electromigration
Omg
Um
I need to go
I just felt an earthquake
Where are you located?
Yup
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 2.00 MiB. GPU 0 has a total capacty of 24.00 GiB of which 15.99 GiB is free. Of the allocated memory 6.45 GiB is allocated by PyTorch, and 300.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
@mortal grail Please stay safe. Here is the error message when you get back. This time it says I have 15.99GB free... great
Ty
what were you at before?
buy a mac
Hmm... Maybe. Then I would have an even worse supported piece of hardware. Could be fun
Hockey
my bad I had to reinstall LAN drivers
this Dark KP has always been a little hecky
with networking I mean
so clear CMOS, DDU, or even heckin sfc dism does nothing
how the heck have you been bro
it's been a little while
lol
Since when were you a mod?
he's OG mod
Oh I haven’t seen him
like was already mod when I first showed up here asking dumb questions lol
That’s cool so a long long time
about my (verbatim) Intel one zero nine zero zero K
first time in voice chat someone told me, "dude, just say 10 900 K"
I'm like oh
cuz I was talking about it & it got pretty annoying lol
"so that works on a one zero nine zero zero K?"
Me with my ryzen 3 1300x
I still remember that time I thought silicon & silicone were the same thing
12100F arrives 8/27
Can someone do this question plss the bluebook explains bad
i believe it was around 140w/1.110v
the 131w undervolt is unstable thanks to OCCT linpack
my new stable voltage is now 1.100v
XD
Is there any good way to automate restarting explorer when it exceeds a certain amount of RAM used?
Changing the config file doesn't change the outcome?
Also I'm fine

config file?
The error spits to a config file for pytorch
Can you edit it to increase the limit?
Need a hex editor and rehash?
oh, no, I can't
its not a fixed value or something I set
when I don't have this GPU mem error, it just... works
been working on projects it's been going well
The requirement is determined by the params of what I am running. There isn't a way to cap it or change it or anything
can't say the same for my 13700K
neat
i'll have to find some time to try it out
forgot to reply directly @mortal grail
wassup
Maybe it's core stability
Downclock gpu core too
glad to hear. heck you're what? junior now?
You can get memory errors like that benchmarking things like ycruncher for ram
@grand kernel hopefully last UV post of my 13600k, this is how much the cpu cranked out
I can really assure you its not that. Why would core satbility throw a memory error, and specifically error out at a specific capacity?
you're getting new CPU?
no no
dont get me started with ram
take Jay with em plz
glad mines is stable
IDK why I can never just have good RAM. This is the third time I have had critical RAM issues on 2 PC's
Just getting more info rn
Nope, all entirely different hardware
Oh yeah
well no crap
Check the ek reddit
they've been dead since like april
Orders aren't being met
lol
oh wow
Yeah I know that
This is a level worse
Insolvency signs
I don't have a confirmed source yet
well heck, no more RMAs for me lol
Fanatec and recaro went under the same business pattern a few months ago
damn bclk 102 didn't POST
I should probably drop memory down one bin lol
@mortal grailI'm doing the GPU driver re-install right now
it's not gonna see anything coming it doesn't have eyes
why? I'm just going to rip its IHS off & try to force it to run memory at 8k+ & see how high core clocks can go before it hits 100C
nah i got 1 quarter left before graduating
dude time flies wow
however, its gonna feel a huge voltage bump
what's the next move?
getting a job at nvidia whilst messing with drivers maybe
I'm so annoyed that X3D doesn't keep boosting if there's thermal headroom
||Probably find a job that doesn't utilize the degree he got at all
||
"Unlimited power" aah scene
57C? boost you heckin bad word
i keep hearing horror stories of people with good PHDs only to never land a suitable job
huh?
no no no NO YOU WILL DIE
haha voltage GO UP
my G7400 is still fine
i mean
it hit 100C lol
it's screaming
Ohp I figured it out
Slovenian authorities have been investigating into EKWB as of two weeks ago
They sent more notice to employees that payments for wages can't be made until october
Kenobi and Anakin on Mustafar aah scene
The accounts have been frozen
5.6ghz G7400
In other words
._.
oh wow EK is hecked
EK sounds like it's been repo'd
oh god what has EK been up to
Frozen for insolvency entirely
time to file bankruptcy & pay its creditors with whatever can be sold off
They lied.
maybe der8aur will buy it or something
@mortal grail
It seems to be working after the driver update
well you plant uh crap seeds you get crap weeds
I can only assume that all the severe crashes my RAM was causing last night with the HDD page must have destroyed my driver
They should have watercooled their brains
If it happens whenever you update drivers then always ddu
maybe then they would have understood how bad this would be
3090 gave me a huge headache with driver problems
my GPU driver crashed like 30 times last night. That was cause of the HDD paging issues I had. I had basically everything paged to my HDD. I fixed that now. That was hell
4090 does too but I believe the 40 card may have been caused by a faulty chipset board
I had over 11GB paged to my HDD. It was so bad even snipping tool took like 20+ seconds to start
Now I disabled HDD page and have it on my fastest and most free SSD
Funny how they issued an apology, did nothing about it and no one said anything
tf is going on here?
binbows michaelsoft moment
It was a nightmare, cause cause I only had 32GB RAM, it was constantly paging cause I kept running out of RAM. Man it sucked lol
i suppose to be crashing more than both of yall thanks to my intel 13th gen + amd combo
💀
btw Felix do you disable iGPU? that helped with FCLK heckin on 5600g but might be comparing apples & oranges ig
turns out the pagefile was just slow as hell
Yes
@mortal grail nevermind. Blue screened again
bro has it rough
5600g before AMD Adrenaline drivers installed was memory monster
I wonder how the floppa is
Haven't talked to him in a long time
4933MHZ!?!?!?!?
slow
this was gonna be my dad's memory OC until I realized Adrenaline drivers super borked it
bro is crankin dat ddr4 stick
that pc will last you a decade
my guy im running DDR4
real
ignore my terrible timings. That was June 13, 2022 lol
my pc will last longer because my cpu is starving for voltage

hmm longevity... never really thought about it tbh
mine is cause intel is a crap compnay
I was gonna say you're being elitest, and then I realized that you're right. That CPU is lucky if it keeps working at all 
i dont think ill buy any cpus from them for a while
Really glad I went 12600k over 13400
im gonna get probably an x3d chip
only excited for their battlemage lineup IF they don't mess it up
Alderlake is like super stable
lol if AMD can mess up 9xxx launch while Intel is selling broken CPUs, anyone can
Would consider swapping ram stick slots
amd when given the opportunity will mess themselves up
how idk?
its freaking impressive
does your computer bug out with two sticks?
what
single threaded
eh not in gaming
oh
no AMD CPU hits 6ghz
yet 7800x3d beats 14900KS in gaming
I need 6ghz dammit
nah
i swear the 14900ks probably wrecks itself in less than 5 months
is 10k+ DDR5 <3
actually no, 1
i've accidently cut the internet cable wired up to the anntena ( WHY IS IT CONNECTED TO THE TOP OF THE SCREEN ) ajd somehow my internet still works but only when i sit next to the router
why is that
how do you accidentally cut an antenna?
just strip it back & splice it back together
coax?
heat shrink would be ideal but electric tape will get you by
Yeah... I guess I have no other choice anymore. I'll have to tear my whole PC apart in order to test that
I had windows totally crap out on me
do you have enough space to put a partition & put windows dual boot there
I had that happen many years ago too. Windows ate the entire file system
to see if windows is the problem
I could put it on my hard drive. I know its not ideal, but I could do it just to see
that way its on its own drive too
pain
instead of worrying about backing stuff up I just put it all on stinky OneDrive
so reinstalling windows becomes more like upgrading a phone
I'm curious. Lets see if using my second GPU crashes me too, or if its just the 3090
Thats about all I can do without tearing my whole PC apart right now. And its way too late to do that
I dunno if win10 or win11 will run right on an HDD anymore tho
Driving hard or hardly driving?
no need for hdd
for os
Also, I try to say it frequently, but thank you to all of you guys who have been trying to help. I appreciate it a lot. You guys are flat out my heros, even if I am not always in the best head space
I can see if I can shovel off the stuff from my 512GB SSD onto my HDD and use the SSD for dual boot. Been curious about Linux cause its mostly better for AI
I always feel bad accepting a "thank you" when I haven't solved the problem but that's a me problem. What I mean to say is, you're welcome & I'm happy to [try to] help
The thank you is specifically for the effort, cause its not something you guys owe me in any way, so it is massively appreciated above all else
hmm...
so the 3060ti didn't crash me
Its at 7.8GB/8GB
I knew someone who swore the only way she could make her 3090 work was to run it with CSM enabled lol
CSM?
Compatibility Support Module I think
hmm
it's basically what you use if you wanna run Windows 7 still lol
if you would be so kind as to remind me what the command is, how did we check my GPU mem in CMD before?
or old hardware
ohhh, right
is it nvidia-smi.exe in elevated cmd? I just found this now ngl
it was a command that showed the remapped sections of my VRAM or something
kinda cool actually
wait, how did you do this?
opened cmd as admin then just typed in
Nvidia-smi.exe
huh, interesting
ah, my power limit is on the 3090. Its real TDP is 420 watts
in the non OC BIOS
I'm gonna try something real quick. I'm curious
I'm gonna load 2 different models into VRAM on my 3090 to see if its an issue with 1 big file, or what
so I flipped the BIOS switch (the one that's like performance or silent) & it seemed to run better lol
make sure system is off tho
I seriously am worried it's my 3090
heck if you press Win+R then type uh
I tried to load, and it blue screened at just 1.9GB VRAM
dxdiag
there's a DirectX Diagnostic Tool
doesn't really give you much info ig tho
but it does detect problems apparently
where would one find this?
I wonder... Let me try something else
thanks by the way
screenshot capture right before EDF6 crash
fun lol
Those are supposed to be grenades
the game froze before the crash which I took as an opportunity to grab that screencap
apparently nvidia-smi has a bunch of commands to make it log various things https://nvidia.custhelp.com/app/answers/detail/a_id/3751/~/useful-nvidia-smi-queries
@idle nexusI just loaded a different model using a different backend, and look...
but it crashes using 3 others
and 2 trainers
let me try llama cpp... Wonder what that will do
I am so confused...
damn that uses a heck ton of memory
24GB is pretty pitiful for AI TBH
I wonder if the workstation driver would be better than gameready lol
but its generating images with much more VRAM that what crashed it before
Tried that, its actually slower
a lot of the CUDA stuff used for AI relies on acceleration things for games
I am totally over where it was crashing before
but if I try to load an LLM, it dies...
hmmm...
how big is an LLM?
in this case, 4-32GB
but it almost always crashes on all of them at exactly 8.7GB. Though I did get that one above where it crashed at 1.9GB
maybe the 39.9gb total isn't enough
oh no, its fine
I've used these models daily for months
it gets to 8.7GB and blue screens
ah okay. I thought that might be why so much was being dumped into HDD paging file or something
oh, that WAS why
It fills system RAM then VRAM
and it would squeeze my RAM, then my VRAM
& since it's so massive it dumped the rest on HDD paging file?
yeah. It would push out the other files to a page. Thats why I have 64GB, cause that stops it from happening, but I only have 32 right now
maybe disabling that HDD paging means it just doesn't have enough memory
so it freaks out & BSODs?
Maybe? It had this issue before tho. Once before
And also, these other models I am loading into VRAM are really no smaller, so I am not sure...
heck it's like you need a 500gb NVMe just to be used as a paging file or something lol wow
not laughing at your situation just amazed at how intense this all is
makes me feel like I barely do anything intense on my setup
Oh yeah, for sure
I have a friend with 4x4090's, 4x P40's, and 2x A100 80G's, and hes still always strapped for VRAM lol. AI stuff is no joke. my 24GB 3090 is chump change
wow
You look at an open source model like Llama 3.1 405b, which is as good as GPT 4O, and you need 1 TB of VRAM just to load it
Hell, even mistral large needs minimum 2 A100's
LLM's are crazy. Probably the most GPU intense thing you can do on your system (in terms of VRAM. Actually running them is a walk in the park)
does like CPU-Z or anything else say there's 32gb for size?
like it's crazy that it would disappear I dunno
I believe it I've just never had it happen like that
Yeah, my whole system says only 32GB. I can keep restarting my PC and it will randomly boot with 64, and I can use all 64 no problem too
when it loses 32gb, is it one channel is out or like one stick from each channel?
I have 2 sticks, and 1 goes offline
cuz even with 4 DIMMs it's still just channel A & channel B no?
ohh I thought it was 4 sticks
nope, dual 32GB sticks
ah okay
one disappears, and my RAM speed gets reset down to 4800mhz. No crash or error or anything
ig they're in A2/B2 slots?
it just happens. And sometimes it unhappens
I wanna say yes. I read the manual when I installed it, and it worked fine. Its been over a year since I have seen the RAM itself
I believe its TeamGroup
Yeah, thats the RAM
sometimes boards revert to JEDEC default when they fail to train but I didn't think they dropped a whole channel
I have some 2x16 & 2x24 of those lol
Its been fine RAM. And then one day I had a critical crash. It was scary, and when my PC turned back on, it was only 32GB
and since then, the other 32GB has decided to put the random in random access memory
I haven't had the chance yet, cause its a huge ordeal/undertaking. I might try it tomorrow
I will need to fully disassemble my PC, and also get it out of where its installed, unhook everything, deal with both PSU's
oh ya I know that headache
I need to remove 2 GPU's and my CPU heatsink to get to my RAM, so its no small task
well the one where it's unreasonably difficult to get to components lol
damn & I thought the balancing screws on the end of the screwdriver act I have to do to swap NVMes was a pain
however, I will say... I see on the amazon page that this DDR5 has lifetime warranty... Hmm
ya RMA sucks cuz of downtime but the constant crashing is pretty bad too
lifetime can mean a number of undetermined years
but some manufacturers actually honor that
like crucial
Yeah, I mean I already have the downtime of half of my RAM. It's not like both sticks are failing
If I could just send one in and then get one later, that's fine. For all I know this issue could potentially be caused because the faulty one is in the slot. Maybe it's trying to address it or something. I have no idea
they usually make you send back both
But at the same time, this error and the blue screen was also happening when I had all 64 GB
I use these cheap ass 32gb DIMMs https://www.amazon.com/gp/product/B0CGWFYBXB/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&th=1
Overwhelming Performance Improved DDR5 Memory: KLEVV DDR5 memory module delivers starting speed at 5600MT/s, 75% faster than DDR4. Lower Voltage, Highly Improved Performance Built-in PMIC: While achieving 20% lower power consumption compared to DDR4 memory, DDR5 memory can reduce dynamic voltage ...
but $85 each isn't really that much cheaper
That's way slower than the RAM I have, but honestly I haven't been using it with XMP for a while now lol
I forgot that DDR5 got inexplicably expensive again
lol well I force it to run 6200c28
which sounds great until something gets borked or BIOS update means I gotta redo it & let it run ycruncher tests for hours & hours
My 64GB DDR5 6000 mHz CL32 RAM was $130 about 14 months ago
damn I wish I'd gotten that lol
Yeah honestly, from what I remember, I bought my CPU, motherboard, and RAM all in some weird price dip. It wasn't planned or anything, but it seems like I got everything right when they were all at their lowest cost
the only good thing about these for me is it saved me the headache of removing the heatspreaders
I run them watercooled to help me feel like running high voltage thru them isn't so bad
The fact that my GPU isn't 100% reliably crashing at the exact same spot every time is giving me some hope, however I still highly highly doubt that whatever problem I'm having is going to be solved without me having to drop considerable amount of money
It continues to be weird that my 3090 is crashing consistently every single time, yet my 3060 TI is not
well if you can find a way to be sure it's the RAM, that would be good. Screwed up RAM screws up everything
However, at the same time, my 390 usually crashes around 8.7 GB, and my 3060 TI only has 8 GB
I doubt it's the CPU but 13th gen Intel crashes usually say "out of video memory"
so much so that nvidia basically announced that people should stop bothering them about that cuz it's their CPU's fault lol
the point being, I wouldn't necessarily blame the 3090 yet at least
I doubt RAM dropping out is the GPU's fault
Interesting... The instantaneous moment I tried to load a bigger one, it blue screened
I'd be much more suspicious of the motherboard, CPU, or DIMMs before I'd think it's the GPU
heck
which motherboard is it btw
do i wanna try smokeless uamf on my laptop
on one hand i wanna optimize my laptop's cpu
Intel LGA 1700 Socket: Supports 13th and 12th Gen Intel Core Series Processors DDR5 Compatible: Dual Channel DDR5, 4 DIMMs Commanding Power Design: 12+1+1 Twin Hybrid Phases Digital Power Design with 60A DrMOS Cutting-Edge Thermal Design: M.2 Thermal Guard, Fully Covered MOSFET Heatsinks, 6-Layer...
another i don't wanna risk bricking the bios
thats my mobo
heck I wonder if this smokeless thing would run on my SurfaceBook lol
can it get around all that spectre microcode crap that makes my laptop run like a volcano?
it's for amd only
smokeless allows you to fiddle with oc settings and whatever the oem decided to hide
on amd platforms
for intel you might as well use throttlestop
which BIOS version are you on??
Ok, I just loaded a different model and it OOM'd at 11.7GB
it won't work cuz of plundervolt microcode thing. I tried. Many times lol
oh yeah..
ya it's pretty crappy ngl lol
best you can do is tdp i guess
ya which borks the 10th gen i7 to even sadder clock speeds
it's Microsoft's fault tho tbh
the thermals on that thing are a heckin joke
& it's one of those glued together piece of crap laptops that are virtually impossible to disassemble for 99% of the world (myself included)
@idle nexus just booted I to BIOS
That's interesting
to repaste the CPU, I'd have to remove the screen which is like gauranteed it's gonna crack
64GB showing up
lol you're on F3 & latest is F10f
i.e., you're 7 BIOS updates behind
What the hell? Do they release on every damn month?
7 came out in 2023 so close to every month lol
I would flash it, but I don't have a USB drive that I know of
cuz of Intel 13th gen microcode update to stop borking CPUs with crazy voltage
ehh I'd suggest using a USB but
This Mobo doesn't have the vastly superior direct bios update. Asus had that one right
Gigabyte's software will usually do it in Windows if you're feeling crazy &/or desperate to update lol
The thought of doing that on a system that is already chronically blue screening makes me want to gag
I've updated the BIOS on this motherboard before, I know I have a USB flash drive somewhere
The big problem is just that somewhere could be anywhere lol
Yeah, but I need a flash drive for that. That's the problem
Damn it, I have no idea where that damn flash drive is
Oh! I found one!
Let me make sure there's nothing on this
Ok, so bios is 100% reporting 64GB. Let's see if it boots with 64GB
does it matter if you jump from F3 straight to F10? I wonder if it's better to do like F3>F5>F7>F9>F10 or something
If not, maybe that's a huge windows error
Sounds like a pain in my tired ass, but maybe
I currently running 64gb at 6400mhz
@idle nexusWindows booted with 64GB this time
you might just wanna download Gigabyte Control Center & let it tell you what to install
I turned XMP back on
yeah, fair enough
plus it's gonna want to update Intel Management Engine
there's some thing on Intel CPUs called like GNA Accelerator that's for AI iirc
not to mention chipset drivers, various LAN, BT, etc. drivers
its so nice to see my HDD at dead 0%, instead of constantly screaming at me all the time lol
apparently there's a firmware update for RGB related things
Thanks a mill for the help with that
no problem. Even if you don't update BIOS, I'd at least download like Intel Driver & Support Assistant & see if it has any driver updates
Uninstall it after cuz it's annoying as heck lol
not sure if only F9 is for my board, or if it needs to go F3 to F9, then F10
or at least remove it from startup programs imo
is that the gigabyte software saying that?
yeah
my guess is either (1) you need to go F9 before F10 like you said or (2) F10 only matters for 13/14th gen CPUs
12th gen is pretty solid from what I've heard
My guess is F10 would be only for 13/14, cause 12th doesn't need to have its balls chopped off to not off itself lol
lol ya 12th didn't have that whole dump crazy voltage thing going on in the microcode
on the charts I saw (forgot where) Alderlake was even more stable than Ryzen. It was fantastic, especially the 12900k. Thing is built like a tank
just be careful. I dunno if Gigabyte does it but MSI tries to sneak Norton installs even with Afterburner now lol


