Tech Lab forum - General Discussions!

[mossy gull]

They are, I hope they improve from here on out

[jade scaffold]

Though really I want a super mini lunar lake pc

But no one's making them :(

[jade scaffold]

But really my next personal socket I want is

[mossy gull]

Why not get an Epic ITX board 🙃

[jade scaffold]

There are none

[mossy gull]

I know

[jade scaffold]

Won't be an itx board for my server this time round anyway

[mossy gull]

Would an Epic even fit on an ITX?
Even 1st gen Threadripper barely fit.

[jade scaffold]

Might just walk mount the thing tbh

[jade scaffold]

Sp6 would

Which is the same size as 1st gen tr which was SP3 then there was sp4 also the same size and SP5 which is the big square one

There was never a tr itx board though

Threadripper requires a chipset epyc doesn't

There was a mini "deep itx" epyc board

Bit that was only for epyc Rome

Anyway want me my 288 core Xeon

[mossy gull]

I am no fan of Xeons

[jade scaffold]

Why not

The new ones are just as good efficiency wise

And xeons do hold advantages in certain tasks

For like fluid simulation epyc is better but for saying AI Xeons are better

Intel's chips have hugely lower latency internally and chip to chip and especially system to system which in AI and other tasks is a big advantage if latency doesn't matter and edges out raw performance only barely though and Io wise they are both currently pretty equal

[mossy gull]

Because last time I looked at 1 they were 20K😅
I do not have 20K for a CPU

[jade scaffold]

you dont have 15k for an epyc either

[mossy gull]

Nope, but I had 3K for a Threadripper 😅

[jade scaffold]

which is in no real way comparable to either epycs or xeons

TR is just what AMD does with crap silicon too bad to use anywhere else

[mossy gull]

I know, I just needed a CPU that could have 2 full fat PCIe 4.0 x16 slots😐

[jade scaffold]

its why TR Pros were basically paper launched being unable to even buy half of the stack of them despite the usefullness of having somany PCIE lanes on an afforable platform

[mossy gull]

I was able to buy a TR Pro, but no motherboard for it🥲

[jade scaffold]

theres only the 24 core one

which is dog shit value

its pretty much only worth the pcie lanes because the CPU cores in that thing are dying and struggling

in the end im glad i couldnt get a TR pro because it would have been so much worse an experience than just buying a very slightly damaged epyc

[mossy gull]

I just didn't want to deal with the fact most Epyc boards I could get didn't have enough ports

[jade scaffold]

my super micro one had 3 16x gen 5 slots on it

the gigabyte ones had like 6

asrocks also had like 5

[mossy gull]

Asus has 1 with 7

[jade scaffold]

wouldnt have had full memory support though

and you loose alot of performance dropping from 12-8 channels of RAM on Epycs

[mossy gull]

But the problem is 2 GPUs, because they ain't 1 slot, but 2 slots each, means I already lose more than half the slots.

[jade scaffold]

a normal ATX mobo has at most 7 slots

[mossy gull]

I know

[jade scaffold]

the super micro one would have still left you an aditional full 16x slot

[mossy gull]

But a GPU fills 2 slots as the cooler is over another slot

[jade scaffold]

which is more than any of the TR boards will give you asside from Asus 1 TR pro board which costs like 2k us

for a mobo with a shagged chipset

thanks asus

[mossy gull]

The Asus board cost me 1200.-

[jade scaffold]

Threadripper had the potential to be amds best product

it is by far their worst

you dont have the TR pro version of it then

1200 is the price for the none pro one

weirdly the none pro one has better vrms despite the only 96 core TR chip being a TR Pro

[mossy gull]

Nope, I have the sTR5 variant, it was back then the only 1 with more than 4 slots😅

I have the the 7960X

[jade scaffold]

yea which uses the uber crap TRs

[mossy gull]

Hey, it works so far😅

[jade scaffold]

shining example of why Modern TR is such a failure

Its quad channel which would be fine if the chips didnt have 3-12 Chiplets

but alas

they do have 3-12 chiplets so you just loose performance no matter what unless you are on a 3 chiplet chip

[mossy gull]

Just wait till you hear the motherboard is a triple channel board with only 6 DDR5 slots😅

[jade scaffold]

even then the memory controler is so unbelieveably fucked up that you cant get high speeds on them

[jade scaffold]

its not its a quad channel mobo

its got 4 slots 1 dim per channel

[mossy gull]

😐

[mossy gull]

That looks like the newer 1

[jade scaffold]

yes

its the only asus board that supports the 7960x

the literal ONLY one

[mossy gull]

Mine looks slightly different, it's a full sized I think ATX XL sized board.

[jade scaffold]

then its not an asus board

Threadripper 7000 supports quad channel DDR5
Threadripper 7000 pro supports Octo channel DDR5

no boad for either serise across all companies has 6 dimslots with only 3 channels

[mossy gull]

Ah, I don't know, not willing to pull out the system.

Could also be an AsRock, lately going back and forth between Asus and AsRock.

[jade scaffold]

this is the ONLY other Threadripper board ASUS make

and this ones ONLY TR PRO normal TRs dont work in it

ASROCKS boards for TR are normal ATX

[mossy gull]

I don't have that many PCIe slots😅

[jade scaffold]

and heres asrocks TR Pro mobo

Gigabyte make a TR mobo but it onlyhas 2 full size X16 slots

[mossy gull]

Those fans look cute

[jade scaffold]

and dedicates the rest of the PCIE lanes to M.2 and other connectivity

stupid 40mm screamers

no idea why the TR pro mobos need that level of active cooling

they just infact dont need it

[mossy gull]

Rule of cool

[jade scaffold]

and this is gigabytes

These are the ONLY Threadripper 7000 mobos

there are no others

Threadripper 7000 is trash

all the mobos are kinda shit

[mossy gull]

I got mine close to launch, so they may have refreshed their boards in the meantime

[jade scaffold]

and any rev1 mobo is like borderline none functional

No they havent

the only thing thats changed is the Chipsets

because the rev 1 chipsets were all FUCKED

[mossy gull]

I am planning on replacing my 7960X with the 285K or the 1 that comes after.

[jade scaffold]

theres also this TR Pro mobo from gigabyte

very nice looking acually

[mossy gull]

AMD keeps doing 180s everytime, like I thought the 9950X3D would have X3D on all chips, and now they don't

[mossy gull]

Classy

[mossy gull]

I hate Gigabyte's WiFi antenna slot, like I can't get different antennae if I so desire

Also, I got my 1st Gigabyte uATX board, and it's damn slow on boot 😅

[jade scaffold]

getting intel 7529 socket mobos is hard

like i can get a mobo no issue cant get one of a fucking normal form factor

[mossy gull]

I guess they're all SSI-EBB boards?

[jade scaffold]

like this s the most nromal thing i can get i think

[mossy gull]

Last time I seen CPUs in the front was in the BTX days😅

[jade scaffold]

gigabyte do their server mobos like that

[mossy gull]

Also, is that a dodeca channel board?

[jade scaffold]

12 channel yes

Xeon 6900 serise support 12 channels of DDR5 CRDIMMs

so 12 channels of 8000mhz

on upto 2 sockets

which is like 2 TB/s of memory bandwidth which is insane

[mossy gull]

still waits for CAMM2 RAM🥲

[jade scaffold]

never getting cam in the server space

In theory the new Xeons suppor SDRAM

so you could get 12 channels of CR-DIMM DDR5 SDRAM

which would be absolutly insane memory performance

[mossy gull]

I know, I know, but I am also not gonna get another server board or CPU.
I am planning on making a workstation for in my future campervan

[jade scaffold]

such a silly socket

the 288 core chip isnt out yet i dont believe

so im gonna be waiting a while for someone to drop one so i can get one with damaged grounding pins in a corner or something

whole gaggle of Xeons

on the left the 128 core all P core chip
the middle is a 144 core E core chip and as you can see its on a smaller socket
then on the right is the 288 core all E core chip on the same socket as the P core chip with literally just 2 of the little E core CPUs dies stuck to eachother

[mossy gull]

Left 1 is the size of my Pixel 8 Pro😐

[jade scaffold]

probably

yea the 2 6900 serise CPUs there would literally be wider than my phone and only just shy of the height of it

they are about half an Epyc Genoa taller than Genoa is which is silly

[mossy gull]

When will we see Nvidia CPUs on the consumer market

[jade scaffold]

like Epycs big

[jade scaffold]

im not sure we'll see desktop chips for a while

if ever

but we might see a laptop SOC in the next year

[mossy gull]

That sounds interesting

[jade scaffold]

well there is a mini super computer desktop chip they are releasing soon

[mossy gull]

I wonder how "super" that will be

[jade scaffold]

GB10

210 Billion parameter AI model

AND AAAAND this is the big part you can keep buying the little mini desktops and stacking them ontop of eachother and they just auto link together and form a cluster its glorious

[mossy gull]

I still mean it.
Like you can't just launch a mini pc with a super computer chip that is louder than a rattling laptop fan at full speed.

[jade scaffold]

i mean if all its doing is AI it wont need super silly ammounts of power per one i dont think

[mossy gull]

Oh god, AI............
It was the most common word during CES

[jade scaffold]

yea its basically an AI dev kit for smaller companies to get started with

128gb of unified memory and more performance than pretty much anything else for 3k

the price for what it is honestly really isnt bad

[mossy gull]

How long till ARM CPUs support external GPUs😏

[jade scaffold]

they already do

[mossy gull]

I mean for us consumers

[jade scaffold]

that would require a consumer ARM chip exsisting

which they uh

dont

not outside of SBCs which arent designed for GPUs

[mossy gull]

Both Apple M and Qualcomm don't support a dedicated GPU

[jade scaffold]

though you can get a GPU working with a Raspbery pi

[jade scaffold]

no shit

why would they

[mossy gull]

I can't run my progams on a Pi😅

[surreal moss]

That the Nvidia Digits thing?

[mossy gull]

Why not?

[jade scaffold]

yea

Because its completely at odds with the 2 chips purposes

[surreal moss]

Did they release more info on it?

[mossy gull]

But it would increase the amount of people buying the Qualcomm chips though.

[jade scaffold]

the snapdragon chip is a thinn and light laptop chip designed for pretty much ONLY that and they are still working on their own GPUs functionality never mind other peoples

and Apple, well 1st their GPU is insanely powerful far more than any occulink GPU could do

[jade scaffold]

it wouldnt

the issue with Qualcoms chips isnt the GPU

its the software

which is why NVidia comming to the Consumer Laptop chip space is exciting because NVidia have a really good software suite

[jade scaffold]

sadly not yet

[mossy gull]

Apple M4 Max is still slower than any laptop with a 4080 or faster😐

Although their battery life at full performance is pretty good

[jade scaffold]

thats a lie

[surreal moss]

Yep, I’ve got standing orders to get a couple when they release

[jade scaffold]

the M4 max actually even keeps up with the 4090 laptop chip

which is a actual 4080

[jade scaffold]

i would like to grab one for personal use

but i wouldnt be doing anything propper with it i dont think

[surreal moss]

Yeah, I’m curious to see how they perform before I decide what I’m going to do at home

[jade scaffold]

honestly the GPU doesnt seem too huge

it seems heavilly optimised for AI

so i dont have super high hopes for the GPU performance

[mossy gull]

It's almost on par with a 4080 in Blender, I can't find any Maya3D benchmark because which sucker uses that.
So, still slower😅

[jade scaffold]

Allthough it might be using Blackwell 1.0 and not blackwell 2.0 like the RTX 5000 serise uses

Blackwell 1.0 is actually built on a superior node to 2.0 so despite its smaller size it might actually be alot more powerfull than its size may suggest

[mossy gull]

I can't wait till we get GPUs the size of the cooler

[jade scaffold]

theres a max size to GPUs tbf

the 5090 its literally as big as its possible to make a single die on the Machines

weird GPU tbh

dont know why they didnt use their newer fancier node

they could have turbo leaped over AMD and just chose to keep their mind you already huge lead

I really like the case of it

i love the gold DGX stations

[mossy gull]

Too expensive for us peasants

[jade scaffold]

and i love the mesh they use on the front

[surreal moss]

I just hope it can be rack mounted

[jade scaffold]

so smol

[surreal moss]

Wait, what

[jade scaffold]

WAIT did you not know they were micro PCs?

[surreal moss]

Nope

https://tenor.com/view/the-universe-tim-and-eric-mind-blown-mind-blown-meme-mind-explosion-mind-explosion-meme-gif-18002878

[mossy gull]

Smol boy

[jade scaffold]

200 billion param AI mini desktop

[surreal moss]

Itsy fucking bitsy

[jade scaffold]

that can have multiple combined to make it run even bigger models

[surreal moss]

I think one will be “dev” and the other will be “prod”
I have no idea what they’re planning. I was just told to buy two

[jade scaffold]

yea they are super cute

kinda why i want one

releases in May

[surreal moss]

That’s over another whole month

I’ll diiieeeeee

[jade scaffold]

also NVidia are going to be doing ARM for windows at some point

though i doubt itll make it to the didgits computer

and will likely be focused on their laptop chip i keep hearing rumors about

i wonder if someone will get games on linux working on it

that would be pretty fun to see

im kinda curious on the exact specs of the GB10 super chip

anyway im not gonna try get one till much later

[surreal moss]

I’ll definitely post about my experiences when I get it

[jade scaffold]

Im actually curious as to how they make that mesh they use on them it actually looks very good imo

[runic thorn]

Any idea why these three are like 300-600 € cheaper than most of the other 5080s? Did someone forget to add the taxes to the price and I should check again for corrected prices tomorrow?

[crude dirge]

As someone who lacks alot of the in depth technical know how of GPUs what are thoughts on the 50 series vs the 40 right now? Internet is full of arguments and im struggling to pin down reputable sources. Been eyeballing options for a full PC rebuild in the next 6 months or so

[sacred seal]

Its not so much a question of 40vs50 as it is what you can find any given part selling for

In terms of pc hardware l don't recommend looking at parts until about two weeks before you plan to buy. Prices just fluctuate too much

Good chance 40 series parts won't even be on the market 6 months from now

[crude dirge]

Thats fair. Right now im kinda mentally in the position of buy things slowly. I have the funds to just buy it all now but not in a hurry if i can score better deals without losing quality/longevity

[sacred seal]

For the most part you will get better value the longer you wait, but it's not going to be crazy . Personality I would wait till march when amd drops thier cards

[crude dirge]

Trying to do a big upgrade thatll last me another 4-8 years without concern. You think the new AMDs might be worth? Or are you thinking from the angle that the nvidia stuff will drop in price?

[sacred seal]

Both.

Depending on the price point. If you are looking at a 5090, then nothing is going to change

Also a pc lasting 8 years is a long time. You tend to get the best value if you upgrade peacemeal. If a 5070 is good enough for you, spending 500 now and another 500 four years from now on a hypothetical 7070 will get you a better experience in 8 years then if you got a 5080 now

[crude dirge]

5090 is probably excessive for my use case, just need something to keep up with the high end gaming needs that can handle software if i need it to. Im aware its a long time, just in an awkward position where alot is coming up due at once yknow?

Mobo is due, processor will be due soon/depending on mobo upgrade, ram is due if i want ddr5, graphics will be due within the next year or so probably

[sacred seal]

Most parts don't need to be updated that often. Pretty well any cpu on the made in the last 4 years is good enough for modern games. Built my personal rig nearly 7 years ago now, same mobo

Got a x370 board and a 1700x, and I only upgraded to a 2700x because I got it for a steal

Going to put a 5000 series part in it next time I see a good deal and get another 3 years out of it

[crude dirge]

Nice

[mossy gull]

Except with how powerful the new GPUs have become, anything higher than the RTX x070 and RX x700 we pretty much need a faster CPU😅

[jade scaffold]

50 series is slightly better has some extra features honestly unless you really want DLSS4 multi frame gen or are a professional and want what is the cheapest and most powerful 32gb GPU or an enthusiast 40 series will be fine for you

[crude dirge]

Ive heard mixed things about the frame gen, for example in situations where accuracy is important the generated frames can interfere with that since they arent native and were basically guessed by the AI behind it. Not sure how true it is but if thats the case im sure i would notice it in use

[jade scaffold]

Frame gen isn't perfect yet

But it'll get better and better over the next 2 years

It will eventually be next to impossible to tell generated from native frames

Frame gens main issue is that it's not actually extra performance it can make 30fps 280fps sure but you won't get 280fps level responsiveness just the visual smoothness

Nvidia do have something that could change that over the next 2 years though called reflex 2 which would make you able to turn and look during a generated frame which would make 30fps generated upto 280 actually feel like 280fps which would be insane and make 50 series truly superior to 40 series in a big way

[mossy gull]

Frame gen is more about "smoothing" the game, but actually doesn't make it run much better, 30fps still feels like 30fps

[jade scaffold]

Yea for the moment atleast

[mossy gull]

Yeah, it does get better though

[jade scaffold]

Like I said reflex 2 really could make Frame gen FEEL native which I think is what Nvidia are going for

[mossy gull]

France gen.......

I wonder how frame gen will change the scene

[jade scaffold]

auto correct lmao

the 5080 reviews are out and uh

wierd ass performance

in some rare instnaces its not far off the 4090 other times its not much better than the 4080 super and 7900xtx

super fucking unusuall

[mossy gull]

I know, but it's funny 😅

[jade scaffold]

yea lets not generate france

infact lets generate literally anywhere else

[mossy gull]

I saw, I don't know what to make of it, I hope it's driver instability or something, but it's weird how it sometimes loses against the aging 7900XTX.

[jade scaffold]

yea

its really unusual

its performance while raytracing performance is atleast rather substantially better

[mossy gull]

Even the 5090 isn't that impressive, still impressive but not that impressive, as DerBauer calls it, 4090 Ti

[jade scaffold]

with some games REQUIRERING raytracing now thats a good to have tbf

the 5090 is atleast mostly consistant though

a bit over 30% faster than the 4090 at high resolutions

which isnt amazing generational uplift

but 1.3x the 4090 is still faster than the 4080 by almost 2x

[mossy gull]

I'll have to see what will happen after a few months

[jade scaffold]

yea ill be interested to see if driver improvements can maybe bring the 5080 up more

and if not

what about Wukong and Cyberpunk Make it so much faster than the 4080

[crude dirge]

Here's hoping. Ive ran AMD cards all my life and am only considering a switch now because i am much more financially stable then i used to be, and i love pretty graphics and raytracing haha. After w4sted hyped up how much better founders edition stuff is to literally everything else i was hoping to snag one of those whether it be 40 or 50 series but i guess i should wait a bit and see if prices drop in the near future

[mossy gull]

As someone who had both, both have their advantages and disadvantages.
If you like running on Linux, then AMD is still the way to go, if you run Windows then both are good.

[mossy gull]

I also want raster performance because many games I play still only use raster graphics

[crude dirge]

Always windows for me. Cant stand macs and never messed with linux. AMD has always been more problem prone for me. When games dont explicitly support AMD on launch they always seem to have issues whereas games dont as commonly have issues with NVidia. Heck even EDF6 had frequent crashing issues with AMD until just a couple weeks ago. I guess part of it is im a little fed up with the problems and want the better raytracing nvidia has, but am not sure when/what to go for and have to upgrade my mobo/related components around the same time

[cinder lagoon]

Hey, anybody got any idea why Blender won't start on My XPS 17 9750?
It says that apparently the OpenGL Version is not supported by the "graphics configuration", however that thing has a 12700H iGPU and a 3060 Mobile, both of which support new versions of OpenGL as far as I could google.

Blender started fine in ubuntu....
Setting the exe's to use the nVidea GPU through windows also didn't help

[mossy gull]

Intel is known to downgrade their OpenGL driver on older CPUs in the past

[cinder lagoon]

Just recently installed both graphics drivers

[mossy gull]

Try the older Intel GPU drivers

[cinder lagoon]

💀

Alright, will try tomorrow

Still appreciate other possible solutions tho

[mossy gull]

1 reason I dont like Intel is because especially on laptops they downgrade their CPUs.
It happened to 3 of my laptops.
1 Dell, 1 Alienware and 1 MS Surface Book 2.

[sonic meadow]

https://cdn.discordapp.com/attachments/125959358336073728/1334357073026220154/FB_IMG_1738179286062.jpg?ex=679c3c5d&is=679aeadd&hm=ee218bf12238dd46d2c727101f6ab841aed3cfaa5f45df5b09876651000c2031&

[sacred seal]

Love that with 96 cores at 52.8ghz, it still has a Intel uhd 630

[tender plank]

9080ti in SLI? How fucking big is the case?

[sonic meadow]

its an extra case

per 9080ti

honestly though, you buy an NUC style pc, which as an expansion port on top. you connect another NUC-sized box to that, and it contains an expansion graphics card

[surreal moss]

that's not a real card

[sonic meadow]

noooooooo

I never would have guessed

[tender plank]

I know. I was extrapolating at the rate GPUs have been growing over the last few generations.

[surreal moss]

[mossy gull]

I guess the 1 free Ubisoft game makes it worth the price

[mossy gull]

https://youtu.be/O46XXFu_578

[sonic meadow]

https://cdn.discordapp.com/attachments/546133905946705920/1334479602659360859/ae99Gxv_460swp.webp?ex=679cae7a&is=679b5cfa&hm=9f4808b020bfa939fec1ff2aa2a64f58ef81a81f936f3239d8c985372243a708&

[jade scaffold]

thats uh entierly as expected

the new strix halo APU is like 7800xt equivilent

like desktop 7800xt specs wise lower power but still specs wise

so its gonna beat the 4070 laptop because the 4070 laptop is 4060/4060ti equivilent

[jade scaffold]

1 Hour to go till new GPUs

[jade scaffold]

Nvidia basically paper launched the 5090 there was a bunch of 5080s but like no 5090s

so

so

so

ive decided im gonna temporarilly get a 5080 then sell it back on later once 5090 supply has increase due to the lack of any decrease in 4080 price and the fact some aib models of 5080 are going new from manufaturer for 1.5k+ i dont really think ill loose any money on them

[cerulean helm]

I've had Home Assistant installed for a couple weeks now, and man this is cool software. The Home Assistant Voice is really promising with local hosted LLMs. Its not as good as a Amazon Echo (yet), but I have my own assistant controlling my Iot devices.

Finished setting up a wall mounted kiosk using a tablet thats always plugged in. And added my TrueNas NAS to Home Assistant to monitor it.

[cerulean helm]

So, my HA voice assistant has been okay. Sometimes slow. I figured that was normal. That was until I realized the DDR4 ram that I had lying around was only 16gb, and one of the memory slots on my used motherboard was apparently dead. Ive been running a local LLM on a pc with 8gb of ram... Whoops. Quick run to the store and it now has 64gb of ram.

[jade scaffold]

lmao

[mossy gull]

Poor RAM slot

[cerulean helm]

Ebay motherboard, an old Z170 chipset.

[jade scaffold]

This thing overclocks surpisingly well

3ghz already with like minimum effort

[cerulean helm]

The 80? Yeah, it looks like its really been held back to yet again segment the product stack. With the price difference between the 80 and 90 its all but a guarantee there will be a super/ti model/s.

[jade scaffold]

nah that just means its not power limited

but it shows how insanely limited the 5090 is power wise

the 5090 can bearly crack 2.45ghz

the 5080 is breaking 3ghz quite easilly

if the 5090 could hit 3ghz it would be running at over 25% higher clock speed all it needs is more power

[jade scaffold]

you know what after playing darktide with frame gen on
i think i might leave frame gen x2 on for darktide its surprisngly not very artifacty the only place i see the artifacts is in the hub world at the giant hollographic table then a saw a bit of weird ghosting on a really distant glowing enemy through fog

and that was like it

[cinder lagoon]

Guys help I accidentally overwrote my partition table of my laptop :((

It was done via diskpart clean (I assume thats what windows uses if you specify a partition layout in an unattend.xml)

I've check with linux#s gdisk and both the backup partition table and protective MBR seem to have been overwritten with the default

There's not really any valuable data on the laptop but I would hate to have to set everything up again

I've found a tutorial online that tells you to use testdisk, which should apparently not only scan for protective / backup partition tables but also find partition boundaries by walking the disk

Can't seem to figure out how to tritter the advanced search tho

Well nvm on the cant seem to figure out - i needed to confirm some changes which would not be applied anyways and in the next menu was the "deep search"

Still taking recommendations tho

I was able to stop the automated Windows install at like 5% so it probably literally was just done with the part table

[mossy gull]

That's...........smart.

Well, getting data off it you need another computer, but you can't fix it

[cinder lagoon]

Ahhh craptastic

I wanted to use my laptop to install W11 on an unsupported machine that had a bios password.
Took out the drive, connected it via USB to my laptop and hit the ventoy stick.
Forgot I configured it to automatically clear the largest available drive and install the system to it 💀💀

Not to mention that the Windows installer doesn't allow installing to USB lol

[jade scaffold]

Lmao

Nice going chaos xD

[mossy gull]

I once had that when trying to install dual boot😅 Linux + Windows, only for Linux to completely overwrite Windows 😅

[cinder lagoon]

Yeah, only a hand full of linux installers support dualboot recognition and automatic setup 😬

Going to just call it a day if photorec and DMDE turn up with nothin (maybe make a copy of about the area of my windows partition) and do a fresh install...

Was overdue anyways 🥴

[mossy gull]

Ubuntu supports it, but I did a dumdum😅

[cinder lagoon]

Ahh yes, it does xD

[mossy gull]

I selected the wrong partition, I forgot at the time I gave Windows the smaller partition, so I installed Linux over Windows.
So Dual Boot menu found Windows boot loader, but Windows didn't exist anymore 😅

[cinder lagoon]

Ooof haha

Btw yeah if you delete the partition table and create a new one, the old one is completely gone.
No File systems, no nothin.
Every tool only finds the new stuff

At least I'll have longer windows support now (:

[jade scaffold]

Nvidia should have shipped these things at higher clockspeeds

im sorry but im getting like 3.2ghz

people are getting like 15% performance increases from Overclocking 5080s

making them like 25% faster than the 4080S why didnt nvidia just ship these things at higher clockspeeds

[cinder lagoon]

Cuz they neede product NOW, the yields don't allow it on all the cards and now they still don't have enough

[jade scaffold]

3.2 isnt seeming to be a yeild issue

[cinder lagoon]

Given the fact that AIBs only got an FE card like 4 weeks prior to launch

[jade scaffold]

almost everyone whos tried has been able to get that clock speed

NVidia should have just shipped the card at a higher clock

[cinder lagoon]

Keyword being ALMOST

If they would have enough product, the lower ones would just be a 5070

[jade scaffold]

yea just ship them at 3.0

rather than 2.6

[cinder lagoon]

They don't stho so they accumulated it all into the same skew

[jade scaffold]

the 70 cards arent the same die as the 80 ones

the 70ti is an 80 core the normal 70 one isnt though

and the 80 cores are pretty much perfect specimen cores they have their full hardware enabled so pretty much no defects

[cinder lagoon]

They dies are designed in a way so that you can cut off a bit or disable it and get a lower tier version

[jade scaffold]

theres no reason for nvidia not to run these things 400mhz over their current rated boost clock

[cinder lagoon]

At least they were last time

[cinder lagoon]

Except not all of them CAN

[jade scaffold]

im running 600mhz over

im pretty sure every single 5080 that isnt a total dud will manage 400 easy

[cinder lagoon]

Also they need to set apart the top tier 5090 some bit

[jade scaffold]

ITS ALREADY 1.6x FASTER

its spec for spec just 2 5080s

but with a lower clock because nvidia didnt want to give it 2 power connectors

theres no good reason for nvidia to ship 5080s with as low as a clock speed as they have done

espcecially when its so far off the performance most everyone has been able to get

So what if a few dont make the cut then? just make them 5070tis

[sonic meadow]

More clock means more power and heat, which also means shorter card life.
These things are already drawing silly amounts of power. Amounts they have been criticized for increasingly over the last few generations.
Plus people in general have been complaining that stuff is less overclockable

[mossy gull]

Only very recent files are there to stay, anything old might be overwritten.

[mossy gull]

I can't wait for the day where Nvidia requires a 1300+w PSU, which last time I heard were illegal in California 😅

[sonic meadow]

too high for outlet current draw I am guessing

[mossy gull]

A buddy of mine said so yes, that anything higher than 1200w you need a 2nd power group or 1 that can push it😅

[cerulean helm]

No higher than 1800w in the US with standard 120v wiring on a 15 amp circuit. At max you have 20 amp circuits for the kitchen which is just under 2000w.

Gets more complicated when you have 240v wired in as well. Like I do in my garage. Literally can also just combine 2x 120v lines and make 240. when its on a 20 amp circuit.

[jade scaffold]

No it doesn't

This things doing the extra clock speed with little to no power increase

And power and heat only reduces a cards lifespan if you are running at over 85c

Anything below that and you aren't seeing any kind of accelerated degradation

[mossy gull]

1800w, basically my PC with 1 tiny monitor😅

And nothing else

[balmy flicker]

any recommendations for tax software? I am tired of paying turbo tax 60+ to file a 1040easy

[surreal moss]

IRS.gov has a direct file option for 25 states this year as well as a free option for all states for income under $84k

[balmy flicker]

thanks Ill have to look into that then

[cinder lagoon]

Not how the GPT or MBR works lol. Thats how file systems work (usually)

[mossy gull]

With files I mean their remains, with good recovery software you can fetch the data as long it hasnt been overwritten

[cinder lagoon]

Yes

Does not seem to work if there is no file system header (the location that says where files end and start, path, etc.)

[jade scaffold]

GB202 vs a Zen5 CCD

the GB202 chip has a bigger area occupied by L2 cache than 2 zen5 CCDs combined

[cerulean helm]

Does that still exist? A certain Musk decided that whole department needed the axe.

[surreal moss]

Without getting directly political I’m pretty sure that it was established by congress and would require an act by congress to abolish it.

Either way the site is still up and after Intuit’s bullshit The IRS rescinded subsidizing Turbotax and is piloting the direct file program I mentioned above

If things continue as they planned last year (yes, I know it’s all up in the air) then Direct File will be the go to option for most Americans with the rest of us that have higher income still using private preparers like Turbotax or TaxAct

[balmy flicker]

getting rid of the IRS would be a real treat but nearly impossible the currently floated Ideas are general tariffs and or a national sales tax but so far all of that will have to start in the house and they have bigger fish to fry atm

on a side not after trying the wilds bench mark I want to know how to tell what is bottle necking my performace
based on Task manager my GPU was barely working while my CPU and ram were nearly taped (I will run a gain later after steam stops updating stuff)

[balmy flicker]

I turned on ray tracing and got better performance??

[jade scaffold]

What game

Well implemented raytracing can actually run really well

[balmy flicker]

the Monser hunter Wilds Benchmark

[jade scaffold]

Is monster hunter wilds the new one?

[balmy flicker]

yep only the bench mark is out atm

[jade scaffold]

Ah

Interesting

What card you running and what kinda performance you getting

[balmy flicker]

I found some settings that give me better performance

[jade scaffold]

Planning on getting it and want to know how it runs

[balmy flicker]

3060 12gb at an average of 50 FPS but the 1% frames is the issue

there is a shader setting that nearly compleetly fixies it

[jade scaffold]

Interesting

[balmy flicker]

the only issue is the textures seme to load in as it renders instead of loading the texture first

[jade scaffold]

You on a fast SSD?

[balmy flicker]

it is on an old intel SSD

more to the point my ram is taped out at like 80% the entier time with nothing elce running

[jade scaffold]

That may be an issue

[balmy flicker]

yep I boosted the graphics to high but changed the texture loading and render distance to over all improve performance over the defalut

[jade scaffold]

What Res you at

[balmy flicker]

1080

[jade scaffold]

Okey cool

I'll be sure to mess with the benchmark

[balmy flicker]

I think I will run a test on the steam deck for the lols

[jade scaffold]

Yea that'll be interesting

[mossy gull]

https://youtu.be/gIK7iUwRgaE

[hasty rose]

the RTX 2060 has no vram...

[balmy flicker]

Even wit fsr you only get 30 and you can see how little fsr has to work with

[mossy gull]

It has 6GB

[hasty rose]

the 1070 has 8gb, is 8gb not the bare minimum?

[mossy gull]

In 1080p I would say currently bare minimum yes.

1920x1080 = ±8GB
2560x1080 = ±10GB
2560x1440 = ±12GB
3440x1440 = ±16GB
3840x2160 = ±20GB

This will vary on games but it is the average usage I seen so far.

[hasty rose]

Well, the 2060 can't handle the bare minimum

[sonic meadow]

It's also 6 years old, and had a sufficient amount when it released

[hasty rose]

1070 has more vram, sure it's not good but...

[mossy gull]

You mean 8 years old?

[sonic meadow]

The GeForce RTX 2060 is a performance-segment graphics card by NVIDIA, launched on January 7th, 2019.

[hasty rose]

Gtx 1070 was released in June 2016

[mossy gull]

Oops, accidentally typed in 1060😂

[sonic meadow]

The 2060 and 2070 both have the same amount of vram as their 10 series ancestors

[hasty rose]

So did the 1060 have 6gb vram a well?

[sonic meadow]

Though the 1060 also came in 3 and 5GB variants. The 3GB was shit though, lower clocks

[sonic meadow]

Yes

No wait, it had the same clocks but less shaders. Thats right

[hasty rose]

Well I went from a 1070 to a 2060 and the lower vram is giving me issues...

[sonic meadow]

Really? Maybe should have gone to the 2060super then :p

[mossy gull]

I went from the 1060 to the Vega FE Crossfire to the 7900XTX.
I miss the extra VRAM

[jade scaffold]

Not just lower clocks literally less cores

Why would you upgrade from a 1070 to a 2060?

That's more of a side grade than an upgrade

[mossy gull]

Shitty RT support 😀

[jade scaffold]

What fucking rt support

That pile of shit would dettonate on contact with a raytraced render

[sonic meadow]

if you scroll down a bit you will see me correct myself

[jade scaffold]

Ah yea my b mobile discord not loading messages cos of shit WiFi

[sonic meadow]

fair fair

[jade scaffold]

No seriously though why upgrade from a 1070 to a 2060 it's a sub 20% uplift on average

[mossy gull]

I remember it being advertised along side Half Life Alyx😅

[jade scaffold]

That games not raytraced though

And it'd be fine for running hl alyx on an index

[balmy flicker]

how do I tell why my ram is running at only 2/3 of its rated clock speed

[balmy flicker]

never mind it turns out I had to disable and then reenable xmp now I have all 3200mhz

[grizzled oracle]

I just noticed this as well for me while tinkering, bios was saying 3200mhz xmp but aid64 was giving a weird 1800mhz something read

Will try

[cinder lagoon]

You've heard of Tetris in a PDF,
You've heard of Doom inside a PDF

Now get ready for:

Linux: Minimal system in a PDF-Document

https://www.heise.de/news/Linux-laeuft-im-PDF-10272022.html

[jade scaffold]

power util is like nothing

in some areas i was at like 50% gpu Util

i need a new cpu man

absolutly maxed settings

[jade scaffold]

I am rather cpu limited it would seem

[balmy flicker]

Still ram limited but then it is cpu

[jade scaffold]

Getting 100fps at less than 230w is insane

[deep zenith]

Ahoy, Techsperts, I have an interesting question and, as is becomign increasingly common, GOogle is not givign an answer.

Suddenly: My Print Screen key is not functioning correctly.

Until like, yesterday or so, pressing the Print Screen key on my keyboard opened the snip tool, and pressign it with Alt/ctrl would give the normal effects of Active Screen or Active MOnitor.

Now no matter what, it brigns up a loadign window and just captures all my monitors as a single image.

ANyone know what might have caused this?

[sacred seal]

First thing to do is figure out where the problem is. use a keypress test website to make sure that the computer is recieving the correct input

Also make sure you have tried turning your computer off and on again

[deep zenith]

HMmm, yup, not registering print screen there

Oky, any thoughts on the next step?

[sacred seal]

I would run the standard windows repair commands
sfc /scannow
DISM /Online /Cleanup-Image /RestoreHealth
In an admin command prompt, but I am guessing it may be an issue with the keyboard itself. You have access to a second one you can test?

[cinder lagoon]

There's many screenshot programs that just override the printscreen key and by default capture the whole screen.
Alternatively microsoft changed the print-screen key to capture the whole screen via snipping tool.

first step would be to check if Win+Shift+S also saves the whole screen.

[jade scaffold]

Getting some wild rumors about Intel's nova lake chips

52 core kinda wild rumors

2026 chips these will be so 2 gens out

52 cores on a consumer desktop platform is nuts even if they are mostly E Cores

[balmy flicker]

is fire fox les of a ram hog than the better chromium browsers

[mossy gull]

Lowest RAM usage I actually seen on Edge with all power and resource managements turned on

[balmy flicker]

shuders

[mossy gull]

32 tabs, less than 6GB usage

[cinder lagoon]

laughs in 78 Tabs

[cerulean helm]

I have spent most of the afternoon adding Zigbee devices to my Home Assistant. I may have found a new addiction... And as always its expensive.

[balmy flicker]

Well orders a second kit of my ram any advice on instalation

Do I Wan to just fill the empty slots or keep the pairs together

[sacred seal]

You want to alternate

And typically I believe it is best to have your weaker kits in slots 2 and 4

[cerulean helm]

Well, holy crap. Der8auer made a prototype waterblock for the 5090 FE.
https://www.youtube.com/watch?v=8JDqy0jmLVY

[cerulean helm]

Just 2 days ago, I had zero Zigbee devices setup in Home Assistant. Now...
And I still haven't installed the new light switches (10x of them) nor wired in the garage overhead door device.

[balmy flicker]

well in theory they should be the same so Ill just slot them into the open slots thanks

[mossy gull]

https://www.reddit.com/r/pcmasterrace/s/CFgqxvZjyD

[jade scaffold]

What's going on with intel their new laptop chips are faster and more efficient than AMDs best laptop chip why can't we get this kinda effort on desktop intel what are you doing

[mossy gull]

AMD uses older chips, that's how.

But oddly enough it changes per few years lately.

[jade scaffold]

They don't though

The new intel laptop chip somehow beats out the strix point zen 5 chips

[mossy gull]

But Intel's chips ain't APUs like the ones from AMD are.

They might be faster, and more efficient, but when it comes down to graphics, they still are worse than AMD.

[jade scaffold]

They actually aren't though

Intel's new igpus kinda insane the one in lunar lake is just a better GPU than AMDs the one in the new arrow lane chip is weaker yes but it's still pulling insane performance for an I GPU

[mossy gull]

I think AMD been very focused onto their Strix Halo, something Intel has no answer for currently.

Also Intel's GPU drivers are still shite.

[jade scaffold]

I don't think they need an answer to it tbh

What's strix halos advantage over a Nvidia GPU and intel CPU combo

The only thing would be addressable VRAM which isn't going to matter to 99% of customers

[mossy gull]

Strix Halo probably has an advantage of it being able to be cooled easier as you don't need cooling for 2 powerful components.

So instead of 1 fan per component like most laptops, you have 2 fans for a centralised component.

Causing the device to be slightly more quiet and easier to cool.

Not to mention it does save space on the motherboard, an U7 + Nvidia dGPU takes a lot more space than lets say the Strix Halo with tons of RAM.

[jade scaffold]

You'd still need the same amount of cooling regardless lmao

If anything you'd need less for the intel Nvidia combo because of the cassam that is the difference in efficiency between them

[mossy gull]

True, but you won't have the problem of a CPU fan or GPU fan, like on laptops if the GPU gets hot it won't instantly speed up the CPU fan, this means the GPU fan will run much faster making more noise.

If you have 2 fans for a centralised component, the 2 fans can work together to cool the entire chip, keeping the device more quiet.

[jade scaffold]

And like nvidias GPUs are tiny with the exception of the 90 class

[jade scaffold]

You can still do that with 2 discrete chips

In fact the strix halo isn't monolithic

It's a separate GPU and 2 CPU dies

You still have ramp fans if only the CPU gets hot or only the GPU gets hot

[mossy gull]

But it doesn't happen.
A laptop will always have a CPU fan and GPU fan unless they're low powered enough to have 1 heatsink.

[mossy gull]

It's still centralised

[jade scaffold]

It does though all the time

Most all laptops the CPU and GPU are cooled by the same cooler

[mossy gull]

No it doesn't, both my Alienware and Asus wouldn't spin up their CPU fan in a GPU heavy load, and vice versa.

[jade scaffold]

It's only super thick laptops that tend to separate the coolers and even then they don't always do so

[jade scaffold]

And that's fine and that's because they were thick ass bricks

Thinner laptops share the cooler so if the cpus not doing much but the GPU is doing alot you can shunt more power to the GPU and vice versa

[mossy gull]

Even my Titan doesn't spin up both fans in a GPU heavy load, and that laptop is thermal constraint.

[jade scaffold]

Why would it

Spinning both fans would increase noise

If both coolers are separate you not spinning up 1 fan makes sense

And again you can run a 203 die at lower power than strix halo and you'll still get better performance

Strix halos ONLY advantage is the large addressable ram amount which is only useful for 3d modeling cad or ai and nothing else

And let's not get started with AMDs severe idle power draw issues

I think strix halo in a mini pc would be great but as a mobile laptop chip it'll have the same failings strix point had but even worse

[mossy gull]

Spinning both fans also can mean more surface area for a centralised chip.
2 fans are usually more quiet than a single fan.

Look at handhelds, the Ally I think it was, it has 2 fans for a centralised component, it's pretty quiet for how small it is because the fans don't have to run that hard.

Now lets take a laptop of similarly power but with separated GPU and CPU, in CPU or GPU heavy loads it spins up 1 fan, not both, making the laptop more noisy.

[jade scaffold]

And again not all laptops split the cooling for that very reason

Infact most all modern n laptops do share the cooling and again a similar performance laptop would pull less power than strix halo

[mossy gull]

But enough do.

That's why think centralised components like the Strix Halo are actually a good thing.

[jade scaffold]

And not a little less like 30w less that's ALOT

Then don't buy shittly built laptops

[mossy gull]

So far Strix Halo hasn't been released yet, so we don't know how it actually will perform and how efficient it will be.

[jade scaffold]

Any good modern laptop shares the cooling so it can put all of its power into a single component

[jade scaffold]

Except we do

[mossy gull]

I wish I could see before I buy it, but it was either Asus SCAR or MSI Titan at the time.

And I need 128GB RAM.

[mossy gull]

I haven't seen anything about it and the devices aren't released yet.

Before you say to get something like an XPG, yeah I couldn't wait a month to get a laptop.
Next in line was a MacBook

[jade scaffold]

It's zen 5 CCDs we know exactly how they perform

It's chiplet so requires IF to run all the time

It has the same GPU architecture as strix point and it has a quad channel memory controller which if it's anything like the epycs memory controller will chug jug power

We know exactly how this things going to run

Basically the same as all amd products on mobile go it'll have great full load efficiency but it's efficiency at less than 100% utilisation will be complete ass making it a terrible mobile chip

[mossy gull]

I am still curious how it'll perform, nothing lately is as cut and dry as it sounds.

[jade scaffold]

Everything pretty much has been

The strix halo will be lucky to hit much more than 7700xt performsnce

If even that being so memory limited

[mossy gull]

1 report says 4070 levels of performance, the other 4060.
So I have no idea who to trust.

Till they properly release it I can say that is what I expect or don't expect.

[jade scaffold]

Also amd have to keep this thing "cheap" if a device with this In goes over 3k it becomes pretty much instantly worthless for ai

[jade scaffold]

4070 seems about right on the high end

Might drop as low as the 4060ti in some unfavourable circumstances

Physically the GPU is a 7800xt like design with a shit memory controller and an iod bolted to it

[mossy gull]

I still wonder how it'll perform really.
Not gonna make a pre-launch judgement.
Left or right, the Z13 is most likely the only alternative for my Book 2 15"

[jade scaffold]

A 9950x locked at 45w and a 7800xt locked at 75w will get you the performance lmao

That's literally what the thing is

[mossy gull]

I am not lugging around a desktop sized system 😅

I need something with enough RAM and VRAM, so far almost no portable laptop has that.

[jade scaffold]

Well except for the ones with quadros in that have nearly 50gb of vram

The strix halo has to come in at under 2.5k in devices

If it doesn't it's not appealing to ai people and probably not to gamers

[mossy gull]

I wonder where to get those.
Because I certainly can't find laptops with digitiser that have them.

[jade scaffold]

That's usually because you find them in environments that laptops usually dont get along with

The Military, construction sites the likes

Usually built like tanks

[mossy gull]

Yeah I don't have 20K on a shelve

[jade scaffold]

This ain't bad for a lunar lake laptop

Aaaaand god is tempting me

[mossy gull]

https://youtu.be/Ndmoi1s0ZaY

[jade scaffold]

Here we go again

At this point just make them out of ceramic that way they'll never melt and if they do melt someone's likely feeding an entire nuclear reactors worth of power into your pc

[mossy gull]

The cable also melted, almost like it wants the whole 600W through a singular or double cable

[jade scaffold]

Yea it's because Nvidia didn't put any load balancing on the FE cards

[mossy gull]

Which is really really bad

[jade scaffold]

Yea

Might mod myself a totally ceramic connector

[mossy gull]

I am curious, would they still have the same problems if they just extended the 8-Pin connector and make it wider?
The whole connector is bigger and cables were also slightly thicker.

Instead of making the whole connector smaller

[jade scaffold]

nah

the pins on the 8 pin connecter are huge comaritivly

they can handel way higher current

if you made a 12v high power with the same size pins as the 8 pin pcie and EPS you'd never run into issues at 600w

[mossy gull]

So basically we need the make the 12VHPWR bigger😅

[jade scaffold]

yea basically they should have just made it fatter

or made it out of something that just cant melt lmao

[mossy gull]

Aerogel

[jade scaffold]

or just ceramic

if you can melt ceramic

you have other issues

[mossy gull]

It was meant as a joke because it's very good at stopping heat transfer 😅

Still, I saw the problems coming 1st time 12VHPWR came onto the market, while there barely were any issues with the 3090 and alike, as soon the 40 series came it's just been 1 big mess with the 12VHPWR.
And the fact Nvidia more or less forces the connector simply isn't good either, I am happy Intel and AMD give the freedom to AIBs to implement it.

[jade scaffold]

the thing is on AMD and intels gpus they could use it without issue

because the most power an AMD GPU pulls is what 350w

same as the 3090 and thats fine

its when you get to the silly powers 450 575 600w that issues start to arrise

[cinder lagoon]

...wondering when 42/48V GPUs start coming to market to prevent this...

Probably next gen lmao

[mossy gull]

24v maybe, but not 48v.

[jade scaffold]

youd need a psu that can deliver that

and the PSU standards rarely change

[cinder lagoon]

Well duh but you also need a cable that can not light on fire and the standard already changed for one that does

[cinder lagoon]

Bruh, typo'ed 24*

[jade scaffold]

the melting cable issue is now weirdly an NVidia FE issue

basically the FE card doesnt load ballnece across all the pins

which begs the question why even go for a pinned connector rather then just a big fuck off bus bar

[mossy gull]

Just go for a barrel jack at that point

[jade scaffold]

barrel jacks are bad for low voltage high amperage situations

[mossy gull]

You can adapt them that they do support it though

[jade scaffold]

you want maximum contact area ideally

[jade scaffold]

no

you just cant

not without an abnoxiously large barrel jack

[mossy gull]

Like Dell's barrel jack?

[jade scaffold]

theres a reason the barrel jacks on most laptops when they had them are like 19V not 12V

[mossy gull]

Make the GPUs 24v and use a barrel jack

[jade scaffold]

thats stupid

[mossy gull]

And so is the 12-pin that keep melting

[jade scaffold]

the GPU already has to do a tonne of work to step down the voltage into the 0.6-1.4 range

so unless you want to make GPUs EVEN bigger and give them even higher power output 12V is still the way to go

[mossy gull]

What about Lenovo's rectangle jack?

[jade scaffold]

can it do 600w

i dont think so

[mossy gull]

Just imagine, 4 USB-C on the side of the 6090 instead of the 12-Pin

[jade scaffold]

no we need an actual built for purpose connector that isnt a fire hazard if inserted even slightly incorrectly or over used even sligtly

[jade scaffold]

yea im imagining the 120w of excess heat that only exsists because the voltage has to be brought down from 48V down to 0.6

one thing USB does really well is supply-Device communication

this is really what you want in a super high power connector, but we cant be using more than 12v Ideally

[mossy gull]

Neither can the singular pin the 12-pin currently has, yet the 5090 FE asks it from a single pin

[jade scaffold]

no it doesnt

and thats the issue

it doesnt ask for shit

it just pulls power through the lowest resistance route

which is a problem

you want it intelligently pulling power through all the pins equally

but because GPU makers are all cheap bastards no ones card does that because of the added complexity at the beginning of the power delivery

[mossy gull]

Which I know currently doesnt happen on the FE, but the way it behaves you better can use like a 3-pin solid wire cable

[mossy gull]

I am curious why PSU's dont have safety in place for stuff like this

[jade scaffold]

they do

but OCP only gets you so far

especially when you allow absolutly absurd currents

[mossy gull]

I mean that the PSU levels it over the wires if the GPU doesnt do it

[jade scaffold]

which is the same thing that blew up a bunch of 7800x3Ds on launch

The Mobo VRMs were so overbuilt the over current protection on the mobo couldnt tell when the CPU was internally shorting out and blowing up

[jade scaffold]

because the PSU isnt pulling power

thats the gpus job

the PSUs job is only to supply what its being asked for and as long as what its being asked for wont cause it to detonate it'll give the GPU that its really the GPUs job to look after that kinda stuff

[mossy gull]

Eh, so far the 12-pin doesn't feel like they thought it through when they designed it.
And that it is more an design over function connector.

[jade scaffold]

the 12 pin itself is ehhh

if installed correctly it works fine

if GPUs were smarter it would work fine

but but but

the fact its not turbo idiot proof is an issue

[mossy gull]

Costs too much money it seems

[jade scaffold]

yea aparently so

nvidia can put god knows how much into the awesome cooler design the awesome PCB the awesome GPU but makeing the 12 pin smart is asking too much aparently

[mossy gull]

I am gonna be honest, the way power plugs are, are kinda dumb.
We can make them smaller yet more fool proof.

But that costs too much money and most likely people wouldn't want to adapt to it.

[jade scaffold]

you cant make them that much smaller tbh

you dont want to be running 50A through something thats tooo small

the fact that your pins would be smaller would harm your reliability and life span alone

However

[mossy gull]

Solid wire can carry a pretty big load though.
But are tough to work with

[jade scaffold]

its not the wire thats the issue

its the connection on the end of it

surface area matters

solid wires also break really easilly which would 100% lead to more fires than we already have

[mossy gull]

Then make it so the cable also has temp sensors which it can use to communicate with the GPU and/or PSU with so the cables won't overheat.

[jade scaffold]

which then doesnt solve the issue

that just stops it from burning up

and would have to do so like all OTP by cutting power

crashing the computer or at best turning what you are doing into a slide show

[mossy gull]

Exactly, which is better than trying to solve the issue of pushing more power through the cables till they burn.

[jade scaffold]

no the actual solution is to build a connector thats idiot proof and way overbuillt for its spec

[mossy gull]

I rather have a few minutes of lower fps than the fear of my house being gone

[mossy gull]

Which probably will never happen

[jade scaffold]

its happened before see the 8 pin EPS see the 8 Pin PCIE see 24Pin ATX

all of them overbuilt for what they do

[jade scaffold]

this connector here used everywhere could do what the 12v High power could do and they dont tend to melt the only thing its missing is the sense pins which imo the 12v High power doesnt make good use of anyway

[mossy gull]

But with the 12-pin we are are version 3?, 4?.
And still have the same issue.

In Dutch we have a saying and if I translate it correctly it would say "3rd round is the winning round".

[jade scaffold]

yea its v3 now

NVidia have mostly dealt with the user error issue

but now the issue is 12V High power is only rated for like 30 insertion cycles

so like

anyone who messes with their PC alot is now at risk of melting their wire

additional issues comming from cheap and badly made wires

its a mess of various issues some of which nvidia can only "fix" by swapping connector

[mossy gull]

I rather for them to pull the card back and fix the power issue to be honest 😅
600w through a single wire is a big no-no

[surreal moss]

What’s a quick and free tool for benchmarking a cpu?
-# never mind

[rancid plank]

So you know how you can airdrop stuff directly from the camera roll to a mac or apple device with airdrop? Is there any way you can do that between an iPhone and a Windows machine?

[mossy gull]

I usually use Passmark Benchmark software, but there's always better ones out there.
But Passmark will do different tests unlike just stressing the CPU.

[surreal moss]

I normally use an intermediary like a cloud or messaging platform

[mossy gull]

Pairdrop.net I believe

@surreal moss you just reminded me to test my U5-245K😅

[sonic meadow]

Currently working on a friend's laptop to spruce it up (9ish year old celery).
Been going through the installed applications and there isn't much there.. except AVG. Classic antivirus is a bit of a performance hog, and a device like this needs all it can get. Plus of course windefender being a thing now
Soooo, yeet or keep?

[sacred seal]

Yeet

[jade scaffold]

Pulling 600 through a single wire is fine

The issue is the connector isn't built to handle that much power through 1 pin it's meant to do 5A per pin not 30-50A

[mossy gull]

As seen from DerBauer it wasn't, the cables was also melting

[jade scaffold]

You just make the wire thicker guage

[mossy gull]

You mean 8-pin sized cables?
I can get behind that

[jade scaffold]

The actual wires on 12v hpwr aren't even smaller than the 8 pins half the time

And if you ran 600w though an 8 pin it would also liquify

If they arent going to load balance they may as well use a tx60 connector

It can do 60amps easy

[sonic meadow]

except for the shit knockoffs Creality (3d printer company) used for a few batches

[jade scaffold]

Or a tx90 if you wanna be real safe

[mossy gull]

Only thing Creality is good for is making a treadmill printer

[mossy gull]

So ceramic connectors on both the wire and GPU, and TX-90 wires?

[jade scaffold]

Just a tx 90 connector with a suitably thick wire

[mossy gull]

That still won't combat the issue of the melting connector because the cable getting hot

[sonic meadow]

suitably thick wire

[jade scaffold]

It's not the cable melting the connector it's the connectors pins on both the male and female ends melting it

It's not the wire in the cable

The problem is the connector not the wire where have you got the wire being the issue from

[mossy gull]

This is from DerBauer's video, you see the cable sleeving is also melting

[sonic meadow]

the connector is dumping heat in to the wire, causing the wire to heat up and thus damage the insulation

the insulation has a melting point hundreds of degrees lower than the wire

most wire insulation is rated to sub-100c

[jade scaffold]

That's because of the connector though

[mossy gull]

The cable also heats up, and a cable is pretty good at conducting heat.
Causing the sleeving to melt.

It was the cable that reached well over 150°C, not the connector.
The connector just won't survive the the heat over the cable

[jade scaffold]

That cable is fine infact the cable is able to pretty safely do 8A

The issue IS NOT the cable

The issue is the connector causing power draw though the individual cables to rise above what it's rated for

[mossy gull]

The cable can carry 100 amps for all I care, but if the sleeving ain't good enough the sleeving will melt.

[sonic meadow]

this is like complaining about a car wheel being damaged after the tire exploded, and blaming the wheel

[jade scaffold]

The cable won't melt if the cable is having the right power pushed through it

What do you not understand here

[mossy gull]

If the sleeving was good enough, but the connector would be bad, then the connector would be the only melting part.

[jade scaffold]

If the 12v high power balanced power properly and want so prone to bad contact with a bit of force applied to it the wires would have 0 issue

[jade scaffold]

THE SLEEVING IS GOOD ENOUGH

Oh my god

It's melting because the wire is having 4-5x it's rated power pushed through it

Which is caused by the bad connector design

[sonic meadow]

the current carrying capability of a wire is rated in several ways. typically all of them relating to a maximum temperature and a rate of change of temperature.

[mossy gull]

So it's good for its rated heat resistant, but not for the excess.

[sonic meadow]

the fuze rating of a wire is typically waaaaaay beyong any of those ratings

[jade scaffold]

If the connector was good it would never push past 5A through the wires which is well below tolerance so it won't even get hot

[sonic meadow]

its rated for its rating

[jade scaffold]

Yea the wire would you believe is not rated for doing 30-60Amps

[mossy gull]

But it could be, but they won't, because it costs more.

[sonic meadow]

18ga is rated for 10A at 60c, and 16A at 90c. but has a fuze rating of 83A (10s) to 1.4kA (32ms)

[sonic meadow]

because there is no need to

[jade scaffold]

A 60amp wire would be like a half cm thick

[sonic meadow]

higher temperature insulation on the wires doesnt fix the connector fucking up, it just covers it up... maybe

[mossy gull]

Well, if everyone keep up to spec instead of skimping on money I would agree.
But as we see now, even Nvidia rather saves money than getting proper load balancing.

[sonic meadow]

those are two separate things

[mossy gull]

With how shite goes these days with the melting shinanigans, I think we better should build over spec more.

[sonic meadow]

stuff is being run to spec, the entire connector is not taking more than its rated load. however for whatever reason more current is being pulled through parts of the connector and thus going over their rated load

[jade scaffold]

Everyone is keeping to spec the fucking connector end is just got shit life span my god

[mossy gull]

I can understand kinda the short lifespan of the connector, like who in their right mind is swapping GPUs every 3 months.

But then again, those people do exist.

[sonic meadow]

every 3 months would be fine, thats still only what... 5 connects a year at most

a bit excessive, but its still within the 20-30 rated connections of the connector

[jade scaffold]

And the wires on the connector are overspeced they can do like 8A safely in theory the crap part is the connector and pins them selfs

[jade scaffold]

Yea but if it's only rated for 30 the connector is going to degrade pretty rapidly

Which could make the issues of improper connection more prevelent

[mossy gull]

Sad that not every case has so much space the connector is in the GPU nearly stress free.

[jade scaffold]

Sad that the connector can't just take the abuse

[mossy gull]

Something I seen on some GPUs is that the power connector os more recessed to reduce the stress on the cable bending down

I think Zotac does this

Or was Zotac the one with the LED to show the cable is properly inserted?

I already can imagine the green LED, constantly burning, when you try to sleep, telling you "Yes, the cable is properly inserted".

[sonic meadow]

I had a motherboard once that had an LED per pcie slot that lit up when said slot was in use

[mossy gull]

Was it an EVGA board?
Because I want that😅

[sonic meadow]

not EVGA, havent had an EVGA motherboard before. I think it might have been an asus, but really dont remember anymore

[mossy gull]

Maybe Gigabyte, they also have weird stuff

[jade scaffold]

I have a gigabyte x299 board like that

[sonic meadow]

well I never did 😆

[jade scaffold]

Light up pcie slots is a weird thing

You wouldn't see it like 90% of the time

[mossy gull]

I almost thought why the CPU is backwards, but it's just the mounting

[jade scaffold]

It's my 7980xe

Is a funky lil dude

[mossy gull]

And people keep saying DerBauer and the other guy didn't correctly put in the 12-pin and that's why the cable melted.
Even though DerBauer gives a whole explanation on why it happened.

Then they say Jay didn't have it even though I mentioned it's solely the FE card, and Jay has the Astral card.

On other news
https://videocardz.com/newz/amd-reportedly-working-on-gaming-radeon-rx-9000-gpu-with-32gb-memory

A midranger with 32GB VRAM, that's for people who use AI that's VRAM heavy a blessing.

[timber barn]

Sounds like a cult

[mossy gull]

The anti-truth cult?

[timber barn]

Team Green cult I guess

[mossy gull]

They don't believe load balancing is a thing and that electricity flows over all wires because a wire is made for 3.5 amp max, so it won't go past 3.5 amps and jump to another wire automatically.

[jade scaffold]

Each wire is not 3.5A max

It's rated for 5a each one can do more than that but the connector isn't rated to do more. The connector is fragile af leading to poor connection

Without load balancing this bad connection causes the electricity to take the path of least resistance running 30A through a wire where even 8A would be pushing it

[mossy gull]

They don't believe load balancing is a thing

[cerulean helm]

If anybody needs a cheap nas setup, just came across this. X99 cpu/mb combo with 2 2.5gb lan ports and 10 sata ports and supports ecc memory.
https://www.aliexpress.us/item/3256807468197329.html?tt=CPS_NORMAL&afSmartRedirect=y&gatewayAdapt=deu2usa4itemAdapt

[mossy gull]

https://youtu.be/oB75fEt7tH0

[cinder lagoon]

lads and gentlenerds, lemme present to youuuu....

drumroll

Windows

[sacred seal]

Now that is one i have not seen before

[cinder lagoon]

...like... how?

[carmine depot]

what the?

[silk jasper]

Anyone have any suggestions for some aesthetically pleasing ITX cases?

[surreal moss]

Fractal North

[silk jasper]

Bit smaller if possible xD

Tho on the note of Fractal the Terra has been one I’ve looked into

[surreal moss]

I see one called Jonsbo? I don’t know anything about them though

[silk jasper]

Never heard of them tbh admittedly I am vibing with this SSUPD Meshroom S never heard of it before but looks nice too

I am lowkey pissed though. If I had worried about this a month ago I could have gotten the case I wanted before it got discontinued

[surreal moss]

Oof

[tender plank]

https://www.fractal-design.com/products/cases/node/node-202/black/

Oh, I guess the Ridge is their alternative to the 202 that better accommodates modern GPUs.
https://www.fractal-design.com/products/cases/ridge/ridge/black/

[silk jasper]

Ty Ty

[tidal birch]

https://youtu.be/6FJ_KSizDwM?si=hz7Ou-3WIQymcbOh

[sonic meadow]

https://media.discordapp.net/attachments/275512286557241344/1340448052590411828/Screenshot_20250117-100203.png?ex=67b26508&is=67b11388&hm=7a122cc5654198a852d5397ecabcdd345b299fd5587f1f9ce448bfd46595b78e&

[sacred seal]

I hope that is a troll post...

[sonic meadow]

that poor 2GB ddr3

[cerulean helm]

Very much a troll post. Jack is a very well known ethical hacker in the cyber security world.

[jade scaffold]

Bought a broken apple watch ultra gonna try and fix it

[tulip valve]

Microcenter is so busy

[tidal birch]

The statistics are rising
https://www.reddit.com/r/pcmasterrace/s/kv34BEOqpi

I might need to use my high school statistics and 3 weeks of College Statistical Analysis and do some math and make a graph.

[mossy gull]

@tidal birch the connector has been a failure since the beginning, almost 5 years of failure

[tidal birch]

Of course. However, some think its not that big of a risk.

[jade scaffold]

Well 2 years really

It was fine for rtx 3000 because nothing was chug jugging 450w+

[tidal birch]

It came out in the 30 series and started there with a safety factor of 1.1.

[jade scaffold]

But with a GPU that pulls 575 it's a problem

[jade scaffold]

With 30 series the cable wasn't put under insane loads though

[tidal birch]

Yeah its always been shit. We are only seeing this, go this bad now.

[jade scaffold]

Nothing doing 4090/5090 power

[mossy gull]

Then came the 4090

[jade scaffold]

Yea and the 4090 was having issues with slight user error then comes the 5090 where the slightest most minute misalignment will cause it to melt

They should have just used beefier connector one that was massive overkill

[mossy gull]

What about, just hear me out, what about not 1, but 2 12-pin cables?

Or we go back to 3 to 4 8-pin 😂

[jade scaffold]

Still causes an issue

If load isn't balanced correctly

Also you'd need 4 8 pins for a 5090

[mossy gull]

Not if you have a Corsair PSU, which apparently can do 300w per 8-pin

[jade scaffold]

What we need is an connector that's just unbelievably beefy

[jade scaffold]

Not how that works

You are operating massively above the rating for pcie 8 pin

Sure it CAN do it

[mossy gull]

Solid, non bendable block, every case to be adapted to, that attaches to the front of the GPU.

[tidal birch]

The issue with making a super beefy connector is that the pcb connector has to be super beefy. It would be better to split the power between multiple connectors so the pcb doesn't need beefy traces.

[jade scaffold]

But 150w through them causes them to get actually warm to the touch something I experienced on both the 3090 and OCed 1080 ti

[jade scaffold]

They combine it at the PCB anyway

That's what's caused all.the FEs to melt

The pins just feed into a big bus bar

[mossy gull]

Then Corsair has huge balls of selling a 2x8-pin to 12-pin cable🤔

[jade scaffold]

If ya gonna feed everything into a giant bus bar you may as well just use a fat bastard of a connector

[jade scaffold]

That's a shit idea

Can't even begin explain to you how that's stupid

[tidal birch]

Btw could I get some assistance? I am trying to figure out how many 5080s and 5090s have been sold.

[mossy gull]

I know, really shitty idea, 1 of many.

[jade scaffold]

5090s less than 2k globally

5080s probably over the 10k range

[mossy gull]

They're short on 5090 from what I know

[jade scaffold]

We know that microcentre sold nearly 3k 5080s and just under 300 5090s

The 5080 was easy to get after all it is half of a 5090 it's way way way more reliable to produce and way cheaper material wise to produce per unit

[tidal birch]

Well the 5080 is the same as the 5090 the only difference is the amount of disabled section of the silicon due to manufacturing defects.

[jade scaffold]

Lmao no

5090 is GB202
5080 is GB203

completely different dies

Though spec wise and size wise the 5080 core is I believe exactly half a 5090 all be it with a higher clock speed as it's not as power limited as the 5090

[mossy gull]

The 5080 is also smaller, if you look at the GPU itself it's like half the size.

[tidal birch]

I see

[jade scaffold]

The core of the 5080 is actually UNDER half the size of the 5090s

The full GB 203 core is 10752 cores

The full gb202 is 24576

5070 GPU Vs 5090

The 5090 core is enormous

Also good news for 5090 winters Nvidia due to low ai GPU sales might dedicate some of that space to the 5090 so the market might end up flooded with thousands of them

[mossy gull]

I want to get the 5090, but I guess I am gonna get the Inno3D with Alphacool block.
Because the FE is not to be trusted.

[tidal birch]

Fe not to be trusted? Is that because of the melting issue?

[jade scaffold]

Yea

[mossy gull]

The poor power balancing issue

[tidal birch]

Well its fundamental to the cards themsevles.

[jade scaffold]

Asus for example have sensors on each pin and the GPU will tell you off you are doing too much power through 1 pin the Nvidia FE one won't

I need an FE card I want to do overclocking and the FE cards have a better bin

So I'm just gonna replace the connector with a custom one

[mossy gull]

I want to go full Star Citizen style with my PC, which means the PSU might reside in a different "case" and because of this I will use extension cables.

And with the 5090 FE's poor power balancing I don't want to take the risk.

[tidal birch]

I am looking to also get myself an FE. I will likely find some 3rd party adapter that will warn of unbalanced power load or something like that.

[jade scaffold]

Yea thermal grizly will probs make one

I'm just gonna replace it with a custom ceramic connector with beefier connections

[tidal birch]

If I really need to I will props just check my cables using a clamp meter every once in a while.

[mossy gull]

As long they don't fix the power balancing issue I don't want to risk with the FE, I hope Asus makes 1 with a waterblock, or else I go Inno3D because they have 1 from Alphacool

[jade scaffold]

The cable itself isn't the primary failure part it's the end that's the issue

[jade scaffold]

They have a liquid version of the astral

But it's 3.5k

[tidal birch]

Ya but the best way to check for an issue it to watch the amps going through the cable.

[mossy gull]

I mean not an AiO

[jade scaffold]

I don't get why Nvidia didn't make every expense for the connection safety on the FE card like no joke that cooler midy have cost tens of millions to develop

No one released water blocked cards anymore really

[tidal birch]

Cheaper = more profit

[jade scaffold]

They never sell well

[mossy gull]

@jade scaffold I mean a card like this

[jade scaffold]

They spent millions on god's gift to cooling and god's gift to compact PCBs

[mossy gull]

Probably because in their test chamber it works fine.

[jade scaffold]

It could not possibly have cost them much more to make the connector work as intended

[tidal birch]

That's only a one time cost. If they can save 1 cent on each card that's another cent in profit on each card they sell.

[jade scaffold]

They made like 2000 of them

[mossy gull]

Probably like €0.30 more during manufacture

[jade scaffold]

They have not made that rnd money back

[tidal birch]

It doesn't make sense to us. However, it does make sense for the stock holders.

[jade scaffold]

Well they have

But from the 50k USD super chips

Really sad we didn't get the new node in the 50 series or the multi chiplet design like the B200 got

[mossy gull]

"Too expensive"

[jade scaffold]

Also AMD not helping

I feel rtx 5000 was weirdly rushed in general

[mossy gull]

AMD should've released an 9900XTX

[tidal birch]

100%

[jade scaffold]

The 5080 overclocked is damn near the 4090 in performance

And it overclocks really well

Nvidia GPUs haven't overclocked that well since pascal

And I'm.not even increasing the power limit to do it

[tidal birch]

Do you think the WireView GPU Pro from Thermal Grizzly would help regarding these melting issues?

[jade scaffold]

They just didn't tune the GPUs which is weirdv

[tidal birch]

Becuase they are already reaching 600 watts.

[jade scaffold]

Maybe wait for a new version

[tidal birch]

Resonable

[jade scaffold]

I would like you to read like 2 lines up

[tidal birch]

Yeah I just saw that. Sorry, I am also currently in the middle of a dnd session.

[jade scaffold]

I didn't have to increase power to get nearly 4090 performance why didn't they ship like that people would have been much less disappointed

Would have forced AMD to price the 9070xt well too

Instead they've tossed it at basically the same as the 7900xtx

Also as for amd they will so be remerging their data centre and consumer architectures which might get us chiplet gpus and something atleast semi competitive with Nvidia

[mossy gull]

What if the 9900XTX would perform close to the 5090 for the price of the 5080, that would be a decent hit against Nvidia.

[jade scaffold]

The issue is AMDs efficiency

[tidal birch]

With 4 x 8 pin connectors not melting

[jade scaffold]

The GPU would cost more than the 5090 to make

[mossy gull]

Many people don't really care though.

[jade scaffold]

And and this is a big one have to pull a kw of energy to match that performance

You would be running like 7 8 pins

[mossy gull]

If people cared that much about power consumption then nobody would buy the 5090 and 14900K.

[mossy gull]

I am running 600w to 650w with a single 7900XTX shunt mod and it only has 3 8-pins.

[jade scaffold]

They also still haven't figured out cache coherency

So no multi chiplet gpus

In other semiconductor news Intel's 18A node is looking great

It's equivalent to TSMCs 2n performance wise it's less dense but allows for higher clock speeds

Intel also have a very real chance of releasing a multi chiplet gpus next GPU gen next gen may be the generation of multi chip gpus

[mossy gull]

Intel Arc C750?

[jade scaffold]

Hoping for a 780 next gen

[mossy gull]

C780?

[jade scaffold]

Intel have a few advantages over amd chiplet wise the main one being that they can build full silicon interposers for waaaaay cheaper than anyone else

[mossy gull]

To be honest, I hope Intel will get their own Frame Generation and their own High-end GPU.
Currently they compete with the 3060 Ti and 7600XT.

[jade scaffold]

They do have frame gen

Or atleast it's being worked on

Xess is really high quality compared to fsr which is something and are hoping to improve on either fsr 4

[mossy gull]

I mean driver based frame gen like AMD and Nvidia.

[jade scaffold]

Nvidia dont really have driver based fg

[mossy gull]

I thought they have with the 50 series

[jade scaffold]

Amd do but it looks horrific the in game fsr3 one looks way better than it and that still looks worse than nvidias

[mossy gull]

The reason I want driver based frame gen are old games that are fps locked to 30fps.

[jade scaffold]

They have smooth motion but it's not amazing tbh it's more interpolation than frame gen ATM

[tidal birch]

With how gpus have taken on so many new uses I wonder if it where possible for GPUs to offload Ray Tracing to a different processor. Like a Graphical Tracing Processing Unit. Doubt it would work with scheduling and word loads but who knows.

[jade scaffold]

Same for and fluid motion frames it's not really the same as the fsr frame gen more interpolation so it just kinda looks bad

[mossy gull]

It's something.

I mean it's either looking at a bad 30fps thing or framegen to 60fps.

[jade scaffold]

You could do it but there's no point tbh

[mossy gull]

I remember the days of running Raytracing on a Threadripper 😅

It was horrible but oddly possible

[tidal birch]

Actually, why not just a Secondary Processing Unit. It has the AI cores, Ray Tracing Cores, and encoding cores.

[jade scaffold]

For one you'd need to give the ai RT thing it's own ram which is expensive it'd need alot of bandwidth

And it wouldn't be instantly aware of what the GPU is doing due to latency and it would hold data the GPU is already holding so you are storing multiple copies of the same data

It's just inefficient

[tidal birch]

ohno, I can imagine one case in which it works.

[mossy gull]

Only useful in some render farms.

[jade scaffold]

For gaming it's not useful

For a AI farm we already have big boi so gpus

[tidal birch]

Rasterization card SLI bridged to the secondary card.

[jade scaffold]

But with raytracings use in rendering you need it to be with the render cores

[jade scaffold]

Yea no that doesn't really work either

It would increase the price of the GPU itself

A cache coherent nvlink bus would require an enormous substrate which in a top end GPU probably doesn't matter but for things under 1000 bucks it'd hugely inflate prices

[tidal birch]

Someday SLI may come back.

[jade scaffold]

I mean look at this abomination the chip on the right is bigger than the 5090 one one on the left is over 2x the size of the 5090 chip

These are nvidias ai chips

Made to be all interconnected

The super massive one on the left is actually a multi chip gpu

[vestal wind]

So should I be worried if, when trying to run the benchmark tool for Monster Hunter Wilds, my C Drive starts getting clogged until the game crashes and the drive only has less than 2MB free?

[mossy gull]

Monster Hunter Wilds is recommended to be on an SSD, and fully filling up an SSD is very bad for the SSD.

[vestal wind]

Oh I've installed it on the E Drive, which is an SSD, the C Drive should just be Windows and one or two other utility programs, so for some reason while it's doing the shader cache it's filling up the C Drive despite being installed on a different drive.

Thought a 128GB SSD for that basic stuff would've been enough. No idea how it's gotten this damn clogged.

[mossy gull]

Try to run storage sense and see if your system has many temporary files.

And also, your Documents also should be standard on your C drive.

[vestal wind]

I don't use my Documents folder much, unless that's where stuff like some save data and whatnot's been going. Also Storage Sense just cleared up a whopping 976KB so that wasn't much use. 😛

[mossy gull]

Games will save in documents.
And if you play Bethesda games it'll clog up fast and if you play Farming Simulator with mods, they will also go into the documents folder.

[vestal wind]

Okay, it's not my Documents - just checked and that's stored on the D Drive as well (which is nowhere near full).

Oh, the AppData folder's on C.

[vestal wind]

Okay, yeah; somehow the MHWilds benchmark test takes up nearly 7GB of space on the C Drive when running in temporary files.

That settles it; gonna have to get that drive upgraded properly before Wilds releases, getting sick of this.

[cinder lagoon]

Man, fuck microsoft.
I've just finished the re-setup my accidentally formatted laptop, bitlocker was enabled by default and I specifically chose to save the key to my microsoft account.
Now that pos wants me to enter the key and it's not in the microsoft account

Nvm that shit saved to the account that is not the admin account on that system but a secondary office account

REEEEEEEEEEEEEEEEEE