Tech Lab forum - General Discussions!

[cinder lagoon]

Have you tried verifying the game files yet

[fierce monolith]

Yes

[mossy gull]

I personally think it's Steam DRM being an ass

[fierce monolith]

I would agree

[fierce monolith]

How do I mod it with out steam

[fierce monolith]

Big news

I got base game too work now I can’t get foc. Too launch

[fierce monolith]

I did it I got it too work

And I don’t know how

[mossy gull]

There's modding tools and probably a website with mods for the game.

[fierce monolith]

I fixed it

Both work now

[mossy gull]

And now write the solution in case it happens again

[fierce monolith]

Um I got no idea what I did

I did a complete wipe of my computer

I guess that did it

And thank you all who helped me thank you so much

[mossy gull]

I thought you already did that before 😅

[fierce monolith]

So did I

I guess 2 time a charm

At least it work now lol

Honestly I think it my dual run program I made

[frozen igloo]

This pc is for sale in my area for 1600usd with peripherals, does that seem like a reasonable deal?

[mossy gull]

If you think it's worth it, I did this quickly
https://pcpartpicker.com/list/3vy9bL

[sacred seal]

Depends on the peripherals (monitor?) but I would say no overall.
Thinking this is it?

https://www.newegg.com/abs-sr7700x4070tis-stratos-ruby/p/N82E16883360477C

[mossy gull]

"AI Powered"

[frozen igloo]

That looks like the pictures, the peripherals are a 27 inch and 30 inch monitor both of 1080p 200hz, maono microphone plus a cheap mixer, logitech g pro super light, pro x tkl and pro x wireless headset going by the information given and pictures

[mossy gull]

ARM wants to compete with Nvidia and Intel with GPUs?
Intel can't even make proper GPU beating the RTX 3060
https://videocardz.com/newz/arm-reportedly-developing-gaming-gpu-to-compete-with-nvidia-and-intel

[surreal moss]

More options would be good though

[mossy gull]

It would be
I can't wait for more competition

[frozen igloo]

Got the pc down to 1300usd, is it a good deal then?

[jade scaffold]

1300 usd for that with peripherals is actually pretty good

[jade scaffold]

tbf intels first attempt at makeing desktop GPUs AMD and NVidia have a rather large experience advantage

Next gen is the gen for people to try catch up NVidia are gonne be in a league of their own amd are going to have stalled its a chance for any company who knows how to make a GPU wanting to make a decent desktop GPU to catch up a bit

[frozen igloo]

That’s without peripherals, still seems like a decent deal despite mobo and psu being ok

[jade scaffold]

1300us for a 4070ti build i think is pretty good

[frozen igloo]

He dropped it to 1200

[mossy gull]

But why compete with Intel?
Not only are they years behind, they compete with an older gen AMD card....

[jade scaffold]

Because intel are still a competitor

Infact intel are actually their main competitor

They are competing with intel to try and erode NVidias low end and mid end segment

[sonic meadow]

heck, nvidia seems perfectly happy to do that on their own 😦

[jade scaffold]

Yea

I mean not really

Nvidia clearly don't care about their entry level but the 4090 somehow outsold all of Rx 7000 regardless

[mossy gull]

AMD just should've unlocked the power limit on the card and given it a better cooler, as we see with a shunt modded 7900XTX, it is very very powerful

[sonic meadow]

which will likely fuck massively with the lifespan of the card. and amd doesnt choose the cooler on aib cards. the partners chose the cooler to meet the thermal requirements of the chip when running at spec

[mossy gull]

I seen my own 7900XTX at times hit 600w, but it doesn't do it continuously, even when it's 40c under load

I have a liquid cooled 7900XTX though

[jade scaffold]

yea but then they would have been laughed out of the GPU market for needing to hit 900w to match the performance of a card that runs at 400w

never mind the fact in its current config the 7900xtx is still less efficient than the 4080 is despite the 4080s increadibly dirty power curve

[mossy gull]

It still would have more potential and it would be between the 4080 and 4090.
As we seen, nobody really cares about power consumption or else they wouldn't buy a CPU that uses past 250w.

[jade scaffold]

well people do care because intels market share has been eating total shit over the past few years

and its not their performance being lack luster thats causing that one

because the 14900k is still as fast or faster than AMDs None 3D chips

[mossy gull]

And yet, in the pre-build market the 14700K and 14900K are way more popular.
And more people buy pre-builds than they build it themselves

[jade scaffold]

AMD have been trying to escape their reputation for having terrible efficiency for ages
making a card that gets its ass completely beat by a 4080 super pulling less than half of its power would be an embarrassment

[sonic meadow]

nevermind that massive increases in power only give minor increases in performance

that extra 100% power is what, 10% perf?

[jade scaffold]

AMD Vega failed because it was so comedically inefficient mixed in with the early driver issues

RX6000 did well because it was run at a lower power than NVidias cards

[mossy gull]

If Nvidia would've gone with Samsung chips, they also would've used over 500w.
But Nvidia went with TSMC at the last moment.
Which even shocked AIBs because they made the coolers for the heat output of the Samsung chips.

[jade scaffold]

no it was known they were going with TSMC from pretty much the beginning

They commisioned a custom node very publically

[mossy gull]

Well, a shunt modded 7900XTX did beat the 4090 OC in benchmarks 😅

[jade scaffold]

it didnt

it wonne in 1 benchmark that was already heavilly amd Scewed

it lost by enormous margins in everything else despite pulling double the power

[sonic meadow]

okay,. it was about 15% gains on the shunted card

which is an insanely shit gain for such a massive amount of power

[jade scaffold]

and 15% doesnt get you within the 50% increase you'd need to match a 4090 from a 4080 or 7900xtx

[mossy gull]

3Dmark Timespy.
The 7900XTX scored 18335 point, the 4090 scored 16645

[jade scaffold]

infact if NVidia wanted to just make the 4080 15% faster they literally could have

in the same power profile too

[mossy gull]

Last time I checked Timespy was focused mostly on Nvidia

[sonic meadow]

and you still lose out massively in any rt enabled scenario

[jade scaffold]

that things missing a shit tonne of cores

[mossy gull]

And yet, in Cyberpunk 2077, in path tracing, the 7900XTX can run 60fps, just like the 4090.
Ofcoarse with AMD optimised profiles as the game is focused on Nvidia.

[jade scaffold]

no it fucking cant 😂

not without consuming like 2x the power and running FSR

[sonic meadow]

it gets 10fps

[jade scaffold]

AMD out here needing 800w to do what NVidia can with a die that could run at 350

[mossy gull]

The 4090 can't do it without DLSS either, I seen it with my own eyes as my brother has a 4090.

[jade scaffold]

its not even funny tbh
the AD102 Die could be built with 5% more cores than it has and ran at a 15% higher clock speed and it would outperform a shunt modded 7900xtx at 350w

thats just sad on amds part

[mossy gull]

On a 4090, path tracing without DLSS, you're hovering around 30fps.

[jade scaffold]

its even sadder that NVidia didnt built them like that because it would have made the 4080 a half decent value

[mossy gull]

Still, people seem to barely care about cost and power usage.
The fast is, even though it would consume double the power, it still can come close.

[jade scaffold]

4090 in 1080p can hit 60fps with path tracing and no DLSS

[sonic meadow]

at 1440p too https://cdn.mos.cms.futurecdn.net/DPdzytqBz3wQ9CWeWdknEZ.png

[jade scaffold]

in 4k its only hitting 20 and in 1440p its getting 40

[sonic meadow]

oh whoops. missed the framegen

[mossy gull]

Who still plays on 1080p?

[sonic meadow]

most people

[mossy gull]

https://youtu.be/eIkx5JwplbQ

[sonic meadow]

literally 57% of people on steam

[mossy gull]

I lately seen 1440p being far more popular

[mossy gull]

Aka Asia, the people who play on a 3060

[sonic meadow]

no need to ping me every time

[mossy gull]

Asia has pretty much the most gamers, i don't think EU + US will topple China if it comes to gamers.

[jade scaffold]

the 4090 at 4k ie over double the resolution here is doing 60fps according to gamers nexus

[mossy gull]

Then why did my brother reach 30fps without DLSS and framegen in 3440x1440.....

[jade scaffold]

cos your brothers probably a bumbling idiot

[sonic meadow]

probably didnt have pt/rt on

[mossy gull]

Also, RT Overdrive isn't Path tracing

[sonic meadow]

Together with NVIDIA, we’re bringing a completely new, fully ray-traced, aka path-traced, rendering mode to the game with this patch – Ray Tracing: Overdrive Mode.

[mossy gull]

Path tracing is a separate setting in the game

[sacred seal]

We have had this coversation about your brother's cyberpunk preformance at least half a dozen times already. Something was very screwy.

[sonic meadow]

tbh thats half of what is going on with eq

[jade scaffold]

that video you linked
it FSR Quality with frame gen on at 1440p

The 4090 with DLSS quality and frame gen on it hitting only slightly less at 4K which is over double the resolution

[sonic meadow]

they get weird experiences and then generalize it to fuck and back

[mossy gull]

I guess it ain't impressive that AMD can reach above 60fps in Cyberpunk in path tracing, I guess that still makes it incapable of running path tracing.

[jade scaffold]

well no its not that its not impressive

its that they still arent comparable to NVidia

the 4090 in raw performance is only 50% faster
the fact that its running a load at over Double the difficulty to run and getting only slightly worse performance is a sign that AMDs Raytracing tech is Way behind even in 2077 which is remarkably well optimised on the raytracing front

[mossy gull]

Cyberpunk is an Nvidia sponsored title, ofcoarse it'll run better on Nvidia, even if AMD had a comparable card, it still would run better on Nvidia.

That's just the nature of sponsored titles.
That's why I still think it's impressive AMD comes close, even if the GPU uses more power.

[jade scaffold]

no its not

they dont come close

its not even remotley close to being close

AMD are beaten out by almost DOUBLE

when the difference should only be 50%

like at worst

[sonic meadow]

amd doesnt have a comparable card, their RT perf is massively far behind

[jade scaffold]

this also doesnt account for NVidias RT with DLSS looking better than AMDs with FSR

DLSS trys to assist and get rid off RT Artifacts which FSR because its not AI based just cant really do

[mossy gull]

Yeah, but in order to use the newest DLSS you need the newest card, which means Nvidia has to think about what, 10 cards?
FSR still works on more than 3 generations of cards from different manufacturers, I still think it's impressive.
But I wish FSR had the ability like XeSS, where on their own cards would use the the AI cores and on others it wouldn't.

[jade scaffold]

FSR works on anything

literally because its just a algorithm

Its not using any kind of AI hardware NVidias does though i dont really see why NVidia lock DLSS down to newest cards despite their cards for the last 3 gens having Tensor cores greedy bastards

[mossy gull]

So is XeSS on non Intel cards

[jade scaffold]

yea

nothing wrong with it being algorithm based just makes it easier to port and gives you a few limitations as to what you can do

[sonic meadow]

DLSS 3.0 makes use of a new generation Optical Flow Accelerator (OFA) included in Ada Lovelace generation RTX GPUs. The new OFA is faster and more accurate than the OFA already available in previous Turing and Ampere RTX GPUs.
aka its using new hardware

cant use features that dont exist

[jade scaffold]

yea The Tensor cores are newer gen to gen

but in reality that doesnt stop them from porting it back

[sonic meadow]

dunno what kind of perf cost there would be trying to emulate it on hardware without the feature

[jade scaffold]

it just makes it more complex to do and would harm the performance of DLSS slightly as you have to work arround missing instructions

[mossy gull]

And yet people made DLSS 3 work on a 3070, it's just Nvidia locking it down to the 40 series.

[sonic meadow]

the new hardware is used for the framegen

[mossy gull]

Frame gen also works on the 3070

[jade scaffold]

works is generous

ive seen the vid of that working its got worse artifacting than the 4000 serise does

could probs be mostly eliminated given enough time

[mossy gull]

They still can let it work on the 30 series, which still a ton of people have, even if it looks slightly worse.

[jade scaffold]

NVidia being as big and infinatly rich as they are COULD do it they just choose not to because they are dicks

but regardless of that that doesnt change the fact they hold a Generational soon to be 2 generational Efficiency and performance lead over AMD

[mossy gull]

I still think FSR should work like XeSS, where it would use AI on the cards from the manufacturer

[jade scaffold]

AMDs AI accelerators in their RX cards are kinda fucking useless tho

they are fundimentally identical to the ones used in Vega which means they loose out to Pascals tensor cores

its why their new AI chips have dedicated AI accelerators rather than just building them into the gpu

its why the CDNA Architecture is SO different to the RDNA Architectures

[mossy gull]

Does FSR use any kinda AI tech?
If so then I can understand it runs better on my AyaNeo than on my old laptop

[jade scaffold]

no

its purely algorithm based

[mossy gull]

Then they optimised to to hell and back for the 780m and not for the 6800M 😅

[jade scaffold]

Training AI models is also super expensive especially for something as complex as DLSS

Its not a surprise AMD dont want to spend the literal billions needed to make such a model

where as for NVidia a few billion isnt really anything to them

[mossy gull]

They have been buying AI companies .
I wonder what AMD will do with them.

[jade scaffold]

kinda same for intel too

[mossy gull]

I was reading about that, I hope Intel can recover from the Harakiri CPUs

[jade scaffold]

Intel can dev and build their own chips no need to pay TSMC for their ai accelerators as well as Intel having infinite silicon to build chips is probs why theres an AI Based XeSS

[mossy gull]

How much I dislike Intel, that's a pain to see your own products killing themselves

[jade scaffold]

yea Intel dun goofed with 13th and 14th gen

[mossy gull]

I am curious how Battlemage will run

[jade scaffold]

Shouldn't affect 15th Gen

thats built on TSMC

atleast Lunar lake is

[mossy gull]

1st Intel has the security issue, now Harakiri CPUs, hopefully 15th gen will have both fixed and not new issues

[sonic meadow]

the only thing I am finding is a fsr bridge. nothing that actually uses dlss3 fg

that is for stuff that actually works

there was a bunch of faked stuff, or ones that resulted in doubled (but not generated) frames

[mossy gull]

It was a while back I read about it, a mod that made DLSS3 work on all 30 series cards.
Close to launch of DLSS3.
I didn't know it used FSR as well

[sonic meadow]

the fsr one was this year. april february or so

there was some older stuff that didnt really work

[jade scaffold]

everyone has thoes issues

AMD had some not so long ago

[mossy gull]

It's been since December I played CP2077 last, so stuff may have changed since then😅

[jade scaffold]

pretty much all of these hardware security issues are a None issue

[mossy gull]

The difference was, Intel could be accessed over the web, while I believe I read that with AMD you had to be near the PC in order to exploid the issue

[jade scaffold]

they can all be accessed over the web

EVERY ONE OF THEM

you just need kernal level access

which mean

its a none issue because if you have kernel access you can already do ANYTHING

[sonic meadow]

the AMD one required that the pc basically already be compromised

at which point.. whoopdee doo

[jade scaffold]

thats pretty much all of them

[sonic meadow]

nah, some didnt

[jade scaffold]

people make these big deals out of these hardware security issues but they either require Physical access to a PC or they require Kernel acces at which point either way your data is fucked

[sonic meadow]

some could be done via inside a web browser iirc

though that might have been more minor ones

fuizzy memory and all that

[jade scaffold]

these are VERY and i mean VERY High level complex hacks right
these are not something you use against some dudes computer these are things you use against Google or something

all of these exploits are things that have to first Get onto a computer
sometimes that can be though a browser exploit but that browser exploit method needs to go

Browser > Malware Download > Malware Runs past anti virus > Gets kernel access > then can fuck with data

the one issue with these things is they work against desktop pcs BUT NOT AGAINST SERVER CPUS because server CPUs encrypt memory

[sonic meadow]

I think it might have been rowhammer 🤔
there was a proof of concept implemented in JS that could run in a webbrowser

[jade scaffold]

so unless theres an exploit in the memory Encryption you cant actually use these "dangerous exploits" against servers

[mossy gull]

I mean, I always said to my family, as long you don't anger a hacker, nothing really will happen 😅

[jade scaffold]

which means you have to get access to a high level employee with the exploit then you have to figure out how to mess with their servers from there

[mossy gull]

I mean, people have done it, so that says something 😅

[jade scaffold]

i mean it hasnt because its such a pain in the ass to do

you know whats easier

PDF with virus > i have a business proposition > PDF gets downloaded > Virus gaming

not even funny

[mossy gull]

Hacking a broken AMD laptop?
I agree

[jade scaffold]

the single most reliable way to get control of an account

[mossy gull]

It's sadly also the most common way

Fun fact: Toyota got hacked and 240GB got stolen 😃

[mossy gull]

@jade scaffold we get a new TSMC facility in Dresden, Germany

[iron rock]

I need electronics help.
I have a set of old nicad power tools, a drill and a circular saw. old 14v black and decker from the last 90s early 200s. they are over 20 years old and the nicad battery they came with has finally died on me... dont ask me how it lasted close to 25 years!
the tools are still in great shape, I have taken good care of them and dont want to throw them out. they were a gift and kind of a big deal, like one of the nicest tools I ever got.
trying to find a way to convert or make a new lithium battery for it but not having much luck figuring out the best way to do this

I dont have the tools (a spot welder) to make a custom pack from scratch. I cant figure out how many amps these things draw, it dosnt say anything about amps on the tools or the battery.
the easiest way is probably using RC 4S packs, I would need a 4s BMS chip but those are cheap. the problem is they are kind of exspensive and worse they are Lipo so they dont last very long. like 150-200 charge cycles?

[mossy gull]

The amps are not important, only the voltages.
The amps are the battery capacity.

Best would be to use Lithium-Ion batteries in this case but you need a circuit board that regulates that the battery doesn't full discharges or over charges.

You can do with a normal soldering iron as long it heats up very quickly.

[iron rock]

um... the current draw is really important

[sonic meadow]

yeah, cordless tools pull silly amounts of current

[iron rock]

and I have always been told that soldering on a lithium cell is a good way to have it blow up in your face or just ruin the cell

[sonic meadow]

probably looking at 2-300w full load, which would be around 20A

[mossy gull]

It isn't, I have 2 power drills, 1 16 amp and 1 12amp, both same voltages and both have the same strength, just 1 has a shorter battery life

[sonic meadow]

12 amp, or 12 amp hour?

[iron rock]

I have no idea what it might be. they are old tools, not like the new ones that are actually more powerful then there corded counter parts.

[mossy gull]

I don't know 😅

[iron rock]

that is probably them amp hour

[sonic meadow]

actually... 12Ah would be massive 🤔 12Wh would also be tiny

[iron rock]

hm... that is weird

[sonic meadow]

the "big" ones from my dads drill are 4Ah iirc (18v)

[mossy gull]

They're quick charge batteries and I have 4 of them in a charger.

[mossy gull]

They're 12v😅

[iron rock]

lol

[mossy gull]

Mine are I mean

But they're professional drills funny enough, their batteries are just tiny

[sonic meadow]

anywho, a 3s setup (for about 12v charged) would work fine for a 14v system. you would get slightly reduced power, but I dont know if it would be noticable

[mossy gull]

The problem is Nickel batteries are generally weaker, but drills back in the day needed stronger motors, these days the motors are pretty efficient

[sonic meadow]

decent 18650 cells should be able to handle 20A (or about 15c, depending on the specific cells you get) without too much trouble I dont think

[iron rock]

if I could make my own pack out of 18650s. there is also the problem that I cant fit 8 cells in the existing battery, I could fit a max of 6. almost every modern tool battery I have seen uses at a set in parallel and then how ever many in series to get to the right voltage. I dont know if this is done for the extra current draw these tools need or just for extra capacity.

[sonic meadow]

nickel-cadmium batteries were anything but weak. they might have had lower capacity than the later nickel-metal hydride batteries, but they had substantial current capability

[iron rock]

if I could run this on just a 4S setup of 18650s, using high discharge cells (I think they go up to like 30amps?) that would probably mean I have 1500-2000mah. the original pack was 1500

[sonic meadow]

we didnt really get back to that amount of current capability until lithium-ion, and those are much more persnickety

[iron rock]

these damn nickel cads lasted 25 years!

granted I wasnt using them every day or anything but still

[mossy gull]

I mean compared to Lithium they're pretty weak😅

[sonic meadow]

not as much as you think

[iron rock]

the real down side is memory

[mossy gull]

Advantage of nickel is you just put in storage completely empty and charge them when you needed them, sadly with lithium if you fully drain them they're dead.

[iron rock]

ok power density, last for ever, good current. but they have a horrible problem with memory. if you dont fully drain them before charging them they do a stupid

[sonic meadow]

NiCd you actually store full

misread the article

nevermind me

The capacity of a Ni–Cd battery is not significantly affected by very high discharge currents. Even with discharge rates as high as 50C
daaaaamn

[mossy gull]

That's how my dad's drill battery died, that's why we then went with Bosch Blue drills

[iron rock]

so do you think a single 4s 18650 setup would work?

[mossy gull]

Depends on how long you want the battery to last

[iron rock]

or are those batteries just going to over heat at the draw of a circular saw or drill

I assume the saw is the most power hungry

its the mini kind with the 6inch or so blade

[sonic meadow]

14.8v nominal, so voltage is completely fine.
its probably fine, maybe just monitor the battery temps for a bit. heck if you really want to you could throw a thermal fuze in the pack

[iron rock]

problem is I dont know how to do this. maybe if I can find a 2x2 18650 battery holder I wouldnt have to spot weld them but I can only find the 4 wide kind and that wont fit.

and I dont know how I would charge it

I am an electronics building noob

[sonic meadow]

pull the cells out and shove them in a charger

[iron rock]

that is a little inconvinent...

[sonic meadow]

I dont disagree

im not too familiar with custom battery charging solutions. but you should be able to find a 4s bms, and then just feed... some voltage in to the setup

[mossy gull]

My advise is to find a good YouTube video about it'

[sonic meadow]

I am sure there are entire communities out there about shoving lithium batteries in old power tools

[mossy gull]

There are

[sonic meadow]

some of them might even still be active on forums 😆

[mossy gull]

From them I learned how to put a lithium battery in my car instead of the lead acid

[sonic meadow]

stupid discord creating externally impenetrable silos...

hmmm. have you looked to see if lithium third party batteries are available?

[iron rock]

I have found a good bit but nothing that really fits what I am trying to do and they all have fancy things like spot welders to build there packs

[iron rock]

yeah, there is nothing. these were not popular enough and are too old for anything liek that to have been made

[sonic meadow]

oh, the age isnt as much of an issue as you think 😛

[iron rock]

can find them for some of the big brands liek makita or dewalt but not for a B&D

might be some of the nicest tools I have owned but they are still trash tier by most peoples standards

even back then when Black and Decker wasnt as bad as it is these days

sigh why cant this be easy. maybe I should just stick an RC/drone 4s lipo pouch in it

[frozen igloo]

Are my eyes playing tricks on me or is this more than one m.2 slot?

[surreal moss]

I want to see the other side but I think that’s just one m.2

[frozen igloo]

It’s supposed to be one but the height of the slot is throwing my eyes for a loop

[surreal moss]

Definitely taller but it looks like a one to me

[jade scaffold]

yea its just 1

its just the slot design thats throwing you off

[sonic meadow]

Yeah, 1.
What you are seeing is the contacts/pins continuing all the way down to where they solder in to the mainboard

That bend they do is really deceptive, especially with how tall the m.2 port is

[frozen igloo]

That's what I thought it was but it looked so strange

[mossy gull]

@jade scaffold This is pretty interesting
https://youtu.be/rsI08XloLKM?si=HtfOQql6Dm-79cKB

[jade scaffold]

Oh yea it's thoes air jets

Been waiting for the things to publicly releases

[frozen igloo]

LTT has some videos with them right?

[mossy gull]

I can't wait for them to come in mobile devices

[jade scaffold]

i cant wait for the bigger ones by the other company to come out

will be amazing for hand helds

[mossy gull]

Fun fact:
A similar type of "fan" has been tested on RC planes to see it it's viable for bigger planes.
Kinda funny how electrical current without moving blades can move air

https://youtu.be/IorDYGI1uqc

[jade scaffold]

thats not even remotely similar

[jade scaffold]

thoes air jets have moving parts

they physically vibrate to push

its only 1 solid moving part making it still kinda solid state but ion engines work very differently

[mossy gull]

I thought they were completely solid state like the Ion engines, brain not working

[jade scaffold]

Noh they just shake real angrily

[mossy gull]

I just hope they're more quiet then their rotatyboys alternatives

[jade scaffold]

They are

Their only issue is scalability

They only scale 2 dimensionally

But they do also get better with more advanced manufacturing nodes too

[mossy gull]

Even so, just stack them like 1 top and 1 bottom, and if air pressure is enough it can cool a laptop.
Rotatyboys are noisy once they ramp up, that plus air resistance makes them really loud.

[jade scaffold]

they arent fans

they are the heatsinks too

they go on the chip or on a vapor chamber

the reason they can be so small and flat is because they use ultra high pressure to blast away the boundry layre making no need for fins

[mossy gull]

Oh good, even better, then we don't need bulky ass cooling anymore, you bet if Asus would make these tiny things they're going to be bad😅

[cerulean helm]

Dammit. There is an open box 27 inch LG QD-OLED 1440p monitor at a local store for $552. Only downside is its a matte finish. I am very tempted.

[mossy gull]

LG doesn't sell QD-OLED displays

[cerulean helm]

Its the 27GS95QE-B that I am looking at.

[mossy gull]

That's a WOLED panel

LG themselves make WOLED panels, QD-OLED come from Samsung

[cerulean helm]

Ah gotcha

[mossy gull]

I personally prefer WOLED because black is true black, not brown-ish colour like QD-OLED

[mossy gull]

LG has amazing burn-in protection, I can recommend LG OLED any day

[cinder lagoon]

I don't see the downside? :D
it comes down to preference but I'd rather be unable to properly see the screen in sunlight reflection rather than being blinded by basically looking directly into it

[mossy gull]

Oh, LG is closer to semi-gloss than actually matte

[cerulean helm]

Ive been told a matte finish is less quality than a glossy finish. And I really don't have a need for a matte finish as I don't have a glare problem where my pc is setup.
But, this is the monitor. But I also just bought 2 new LG ultragear 27GP83B-B last year.

[mossy gull]

I have a drawing display that's more matte than 1 of my 45GR95QE-B

My other 45GR95QE-B is a custom order glossy panel

[cerulean helm]

So like, I'm tempted cause OLED. But my current monitors are a year old. And OLED is only going to get better and more affordable.

[mossy gull]

If you're in a dilemma that big, then don't do it

[cinder lagoon]

Oled is pretty much at it's peak - and selling a 1 year old monitor with warranty will get you much of it's price back

[jade scaffold]

The main thing improving with oled atm is the Life span

in terms of colour accuracy response time OLED is pretty much as good as it can possibly get

Oleds even starting to get really bright too

[mossy gull]

I never understood the brightness aspect.
Why do people want 2000+ nits when they're sitting in a dim room

[sonic meadow]

more brightness is needed for HDR.
also people dont know what they need, so become wrong about what they want

[cinder lagoon]

Only need those nits on a monitor if you have giant window fronts (i.E. Window seat in office building)

Or on a TV if it happens to face a window or it's generally pretty bright

[mossy gull]

With that I think there's a limit when it's noticeable or counter-intuitive.
While LCDs need like 2000+ nits for good HDR, OLED doesn't really need it because of the near infinite contrast.

If stuff gets too bright then the darks will disappear.

[sonic meadow]

you still need high emissive brightness for hdr, regardless of the "near infinite contrast"

[mossy gull]

But you don't need 4000 nits on a panel that can go as 0 nits

1000 nits is already very bright on an OLED panel, maybe I am just over sensitive to light

[sonic meadow]

nothing would surprise me with you

[mossy gull]

I have 2 45GR95QE-B and a LG C1, on both I had to turn down the max brightness 😅

[cerulean helm]

Well, I made the impulse buy. RT ratings had a pretty good review on the LG OLED and the open box price with full warranty was to good of a deal.

[cerulean helm]

Update after the setup. Sop far so good. The matte finish hasn't bothered me at all. Either that, or I'm just not sensative enough to notice the difference. HDR in windows looks great. Better than the LG nano ips ever did. Haven't played any games yet however.

[mossy gull]

Like I said, it's more like a semi-gloss.
It's not as bad as people think, I thought it was really bad that's why I have a custom ordered glossy 45GR95QE-B.
But once I got a 2nd normal 1 I barely see the difference in low light.
Only when my window are open I see the difference.

[mossy gull]

@jade scaffold I have a dupb question, can I properly cool a vapor chamber with another cooler😅

[jade scaffold]

theres no reason why not

[sacred seal]

A vapor chamber is basically an upgraded heatpipe, by itself it does not cool anything

[jade scaffold]

a vapour chamber is just a fancy heatspreadyer

frankly i think modern CPUs should just come with a vapor chamber slapped on them

especialyl the EPYCs and Threadripper sized CPUS

[sacred seal]

Wonder what the cost would be on that

[jade scaffold]

you are charging me 11 grand for a CPU the cost of a vapour chamber can really not be THAT Significant

on consumer ones even still 700 for a CPU is abominable

[sacred seal]

Well threadrippers go as cheap as 500, but point taken

[jade scaffold]

No?

the cheapest TR is 1.4k for a bunch of defective low quality ass silicon

[sacred seal]

Though there was a modern 12 core, but it seems like I was mistaken

[jade scaffold]

there is

its only the TR Pro one thoe

the 7945wx

and its not actually real

THANKS AMD

its like OEM only and non of the OEMs have them

[sacred seal]

Yeah, that explains my confusion. the 24 core goes for 1k, so 500 for the 12 core seems reasonable

[jade scaffold]

1.5 for the None pro 24 core

[sacred seal]

I think that is what the first gen tr went for

[jade scaffold]

its 2.5k for the PRO 24 core

the 12 core PRO chip would probably also go for 1.4k

[sacred seal]

https://pcpartpicker.com/product/kNyqqs/amd-ryzen-9-3960x-38-ghz-24-core-processor-100-100000010wof

[mossy gull]

Okay good, I am gonna try to OC my old laptop by water cooling it with my PC's cooler when the PC is turned off.

[sacred seal]

Which is it...

[jade scaffold]

thats 3rd gen

7960x

[mossy gull]

Yesn't

[mossy gull]

I have a boring 7970X 😅

[jade scaffold]

i really dont like Threadripper

[sacred seal]

I stand corrected, I thought they quit after 3rd gen

[jade scaffold]

Threadripper is one of the worst things AMD has made in recen times

Threadripper 1000 and 2000 had comedically bad Latency issues making them unuseable for many tasks but they were gonna keep supporting the platform

[mossy gull]

And yet, as soon it disappears everyone wants it back

[jade scaffold]

but wait no we fucked up the socket design so we are gonna re do it for 3000

but we promis this time we'll keep it arround

then they didnt

made a new socket for 5000 pro and didnt release a 5000 none pro line up

not to mention 5000s for like 3 years had a memory channel that just DIDNT work

then theres 7000

which Had a none functional chipset with revision 1 mobos

and is just too expensive

EPYC is now more HEDT than Threadripper WHAT THE FUCK AMD

[mossy gull]

But Epyc boards lack ports and PCIe slots, not to mention a lot of them are none standard motherboard sizes

[jade scaffold]

what

LMAO

USB ports yea

sure

[mossy gull]

That's why I don't have an Epyc, too many USB devices.
And not enough space for expansion cards.

[sacred seal]

I wish AMD would make a 5900x3d. Who on earth wants a 5500x3d?

[jade scaffold]

who wants a 5900x3D

the x900 chips should NEVER be bought by ANYONE

they use the worst silicon, worse yet they are inconsistently dogshit and they have 2 CCDs so you only get V-cache on some core s

sometimes you get it on 8 sometimes youll get it on 4 sometimes on 6

[sacred seal]

Someone who has a AM3 system and can use more then 8 cores?

[jade scaffold]

get a 5950x

decent qaulity silicon and you might atually get decent perfromance

[sacred seal]

Then you are missing out on the single core preformance

[jade scaffold]

NO YOU ARENT

the 5900x and any x900 chip from amd will have the worst single core clocks

will ALLWAYS loose to a x800x3D

[sacred seal]

Yes, but it will win against a 5800x, and will be faster overall then a 5900x

[jade scaffold]

no it wont

you are most likely to get 6 x3D cores which will perform the same as 8 none 3D cores in games that can use them

then get their ass beat in anything not cache sensitive

worst case you get 4 X3D cores

which has happened

the 5900x is fundimentally TRASH so is the 3900x and 7900x

they are each the worst CPUs in their line up due to their wildly inconsistent build and bordering totally defective silicon

[mossy gull]

Isn't this CPU in a 8+4 configuration?

[jade scaffold]

because they arent Epycs too they still have the same latency issues that prevent the 7950x3D from utilising cache from other CCDs

[jade scaffold]

no?

8+8

[mossy gull]

I actually meant the x900X(3D)😅

Sleepy mind no read😅

[jade scaffold]

yes the x900s can be 8+4 or 6+6(Most common)

[sacred seal]

Looking at benchmarks, 7900x3d is faster then the 7900x, and the 7900x is faster then the 7800x

[mossy gull]

Wait, it can be both, that's what I call a lottery 😅

[jade scaffold]

yea it can be both

in all core the 7900x is going to beat the 7700x DUH

[sacred seal]

Gaming benchmarks

[jade scaffold]

the 7900x3D is kinda all over the place due to being able to get only 4 cores with 3D cache

which in single core games not an issue

in multi core its problematic

x3D gives you ~30% more performance in games that can use it

the 7700x has 30% more cores on a single CCD

Games can use cores from multiple CCDs but if the game needs to kick data across them it does cause minims to drop

[mossy gull]

Today I found out my AyaNeo runs games better than an underclocked laptop with a 5900HX and 6800M 😅
Which is kinda funny as the AyaNeo has an upper limit without OC of 60w, and the other laptop is down clocked to 100w.

[sacred seal]

Seems pretty consistant across all the bencharks I am finding. You have a game in mind that demonstrates this issue?

[jade scaffold]

no but i trust the professional overclcokers who work with AMD

who tell me the 7900x is the single crappiest CPU amd make at the moment

[sacred seal]

You mean the 7900x3d?

[mossy gull]

My brother has 1 in a 6+6 config😅
He can apparently disable 6-cores to give the remaining 6-cores a power boost

[sacred seal]

Thats been a thing forever. Gaming mode or whatever

[jade scaffold]

oh no and the 7900x its also trash its silicon is low quality af fine enough if you are too cheap to buy the 7950x for productivity

Ryzen already has Bad silicon by amd standards the x900s just get the worst of it

[mossy gull]

That's a thing?
I never knew

[sacred seal]

Huge controversy with intel testing its CPUs against AMDs in gaming mode awhile back

Who was the 3rd party tester who did that...?

[jade scaffold]

it even looses in cyberpunk which CAN use 16 cores the 7800x3D and 7950x3D manage to hit the GPU limit by the looks of things but EVEN the 7600x3D is Winning over the 7900x3D

[mossy gull]

Cyberpunk can use more than 16 cores, but after 8 cores the performance increase is pretty minor

[jade scaffold]

even the 5800x3D can somehow weezle its way into being close to the 7900x

[sacred seal]

I'm not surprised that the 7900x3d looses out to the 7600x3d, but that is not what we are discussing. The 7900x3d is for the most part (Cs go apparently is an exception) faster then the 7900x, and the 7900x is faster in games then any of the other lower end x suffix parts.
I would not expect a 5900x3d to beat a 5800x3d, but it will beat the 5900x which is one of the few options for am3 if you want more then 8 cores

[jade scaffold]

or you go for the 16 core part

[sacred seal]

A 5950x3d of course would be even better, but that would be more expensive and might strain am5

[jade scaffold]

the x900s only saving grace is AMD can sell it for cheap because its made out of all the crap stuff that fails quality testing

[sacred seal]

But AMD does not have a 5950x3d ither. But they do have a 5500x3d, a 5600x3d, a 5700x3d, and a 5800x3d

I don't understand who is supposted to buy the 5500x3d

[jade scaffold]

their new TRs are much the Same the silicon in thoes things is appallingly bad its why i dont like them not to mention AMD will likely dump the platform and make a new even more expensive one next gen like the have for the past 3 TR gens in a row

budget gamers who want balling performance and efficiency

the above 8 core X3D parts imo are kinda silly if you want 12 or 16 cores you want them for profesional stuff

but you want to waste 15=20% of your performance in thoes getting a bit of extra gaming performance

[sacred seal]

You have the 5600x3d for that already

[jade scaffold]

thats a mid end part

the 5500 would be actual budget

[sacred seal]

The 5500x3d is a 5600x3d with a slightly lower clock speed

They are both 6 cores

[jade scaffold]

it uses that same dogshit silicon the 5900x uses

but people WILL ACTUALLY BUY IT

no one buys AMDs 12 core parts

they are super unpopular
Why? because they are bad and the x3D one REALLY doesnt make much sense

[mossy gull]

The only reason I have a 7970X is so I can run 2 16-core VMs with their own dedicated GPU

[jade scaffold]

yea thats what ever

but i wouldnt consider threadripper HEDT anymore

Its now purly work station

the MINIMUM cost for TR7000 will cost you 3k

[sacred seal]

Lots of builds out kicking around out there with the 5900x

[jade scaffold]

not compared to any of the others

[sacred seal]

No, but that is like saying there are not a lot of builds with the 5700x3d

[jade scaffold]

more than there are with the 5900x

its literally their worst selling part

same with the 7900x and 7900x3D

they sell awefully

[mossy gull]

I always seen them as workstation CPU, I never really understood the HEDT naming

[jade scaffold]

honestly i think they'd sell better if the AM4/AM5 IOD wasnt totall fucking ASSS

OG TR was HEDT

it was "Cheap" it had all sorts of weird features

[sacred seal]

Yeah, becuase they fill a wierd parts for a wierd niche. You want more then 8 cores, but what you are using your PC for is not generating you income.

[mossy gull]

The 7960X is roughly 1600.- overhere

[jade scaffold]

then you should be using the x900 not an x9003D

[mossy gull]

The 7970X is 1000 more expensive

[sacred seal]

Most people ither go for the 8 core, or the 16 core, and that predates the x3d lineup

[jade scaffold]

the x900x3D is a Scam of a part

[sacred seal]

So nothing to do with the preformance

[jade scaffold]

its basically for the consumer who isnt informed on what it actually is

the x900x sure if you use your pc for work and you cant afford a x950x sure i guess

[mossy gull]

I remember the days when 16-cores cost you like 6500.-

[jade scaffold]

For gaming it gets its ass beat by the 7800x3D the 7600x3D and 7950x3D

like not unsubstantially in most cases either

AMD STILL LACK BUDGET OPTIONS

the 5500x3D kinda fufills that

[mossy gull]

7500X3D🙃
Just imagine

[jade scaffold]

still havnet made a fucking quad core in like 3 gens now

no truly budget chip

a 9300x3D for like £65 would be pretty cewl

[mossy gull]

That would crush Intel

[jade scaffold]

well considering intel are somehow the only one offering a budget option yea

[mossy gull]

AMD needs to make X3D laptop APUs

[jade scaffold]

i want to be able to throw CPU Mobo ram PSU together for under 200 with brand new parts

no they fucking do not

X3D is awful for laptops

[mossy gull]

They actually did once, on an Asus Scar

[jade scaffold]

it raises Idle power so much

yea they did and it was just a desktop chip slammed in a laptop

which gets NO battery life

the way they are going with thoes "Ryzen AI" chips is a much better direction

[mossy gull]

For a gaming laptop I wouldn't mind, anything better than the the 14900HX

[sacred seal]

You are taking a modern innovation and projecting it backwards onto something to say its bad. x900x3d parts are not as good as the x800x3d parts on a per core basis, but they still out preform there non x3d counterparts.
5500 is going for sub 100 right now, I would say that is pretty budget. The 5700x3d is going for like 200 now, so you have to fit both the 5500x3d and the 5600x3d in the 100-200 dollar price point

[jade scaffold]

yea thats fine

a 5900x3D is useless

and is another one of thoes chips thats just a scam

[mossy gull]

Alienware Area-51m with a desktop 10900K, 128GB RAM and a desktop 2080 Super, damn, I love that half hour battery life.

[jade scaffold]

if you are using your PC for work get the work varient of the chip

the 5900x will do fine for that

[jade scaffold]

yea and we dont need MORE of that

[sacred seal]

Would it be it better then the 5900x? If yes, then its only a question of price point, and is an objective upgrade to users who already have a 12 core part

[jade scaffold]

No

it would be inferior to the 5900x

it'd be more expensive only to loose in your precious work performance

and then loose to all other x3D chips in gaming

[mossy gull]

You know what I love about insane laptops like the Area-51m and the Titan, they can run just fine on those portable battery stations, while desktops simply use too much.

[sacred seal]

I'm not seeing that with the 7900x vs 7900x3d. THe 3d seems to be pretty well objectively better

[jade scaffold]

except it allways looses in workstation

[sacred seal]

What workstation? Blender (because that is all that anyone seems to test)?

[jade scaffold]

and the 7900x3D can only beat the other 2 lower tier x3D chips if a game can use EVERY core

[jade scaffold]

literally everything none cache sensitive

so blender Video rendering

CAD

[mossy gull]

Maya3D, which is still 1 of the fewer programs which can utilise 2 GPUs without Crossfire or NV-Link

[jade scaffold]

the only thing it might edge out an advantage in is flow simulation

because 6 of its cores have extra cache

[mossy gull]

I love flow simulations.

[sacred seal]

Your talking like a 5% difference, if that

[jade scaffold]

5% difference for worse performance and class trailing gaming performance and a 15% higher cost

and higher idle power

and higher thermals in the cores

ya know the little things

its an inferior product

the x900s generally are

like i said Low quality silicon

[sacred seal]

I'm not disagreeing it is a superior product, but for the niche that the x900 parts serve, it is still a useful part

[jade scaffold]

the normal x900 sure

theres a use for it

if you cant afford a x950x

sure get one they are super cheap

[sacred seal]

No, the x900 class. The hobbiest. People who have tasks that take more horses, but are only done for fun.

[jade scaffold]

but a x900x3D doesnt fill that neich its not going to be Super cheap because cahce is EXPENSIVE

it could fill the niche of the x900 just worse than x900 does

[sacred seal]

5% loss in workstation for 10+% in gaming preformance
And Cache cant be that expensive because they are making a 5500x3d

[jade scaffold]

its pretty expensive

the difference between the 5900x3D and 5500x3D

is the 5500x3D will sell 3x the number of units

[sacred seal]

At a lower profit margin

[jade scaffold]

still tho

the 5900 sells so badly you can pick it up for sub 200 really often

[sacred seal]

That 3x more sales could very easily have 1/3rd the margin

[mossy gull]

What's wrong if someone gets a x900x3d?

[jade scaffold]

i dont think you can pick up any of the other 5000 chips for under half MSRP

[sacred seal]

Meaning you are just as good off

[jade scaffold]

well i doubt that

you are using 2x the silicon

for ever 5900x3D you make you could make 2 5500x3Ds just for the CCDs alone

thats alota extra silicon that could go into 5500s/5500x3Ds

[sacred seal]

The 5500x3d has to be under 200 usd, Lets assume they cost 150 and have a 50 dollar margin. That would mean a 5900x3d would cost about 200 dollars in parts, so you get 3x more profit as long as you are selling it for more then 350

[jade scaffold]

which they cant

they'd have to sell it for sub 300

[sacred seal]

5900xt is 350...
5900x is 250

[jade scaffold]

Like its not even funny how badly the 900 sold

if you sell a 5900x3D for more than 300 it WILL not sell

[sacred seal]

A 5800x3d is 400

[jade scaffold]

no its fucking not

i got one for 230

[sacred seal]

Thee 5700x3d is 200

[jade scaffold]

highest i can see is 300

at 350

you are at 7900x price

the 7900x will totally BODY a 5900x3D

[sacred seal]

With a higher cost of ownership

[jade scaffold]

i think a 7700x might

the 7700x manages to get about 5800x3D performance in games meaning its probs about 25% faster streight up

so it might actually manage to pull 5900x performance in everything

with better gaming performance

but also WHY WOULD YOU EVER GET A 5900x3D

if you want a cheap upgrade for your AM4 system

why not just the 5900x

which can be got off of Amazon DIRECTLY FROM AMD

for 240

[sacred seal]

And no, it would probably only be slightly faster. 5800x3d was fairly close in performance to a 7800x It would stand to reason that the difference between a 5800x3d and a 5900x3d would be similar to the 7800x3d vs the 7900x3d

[jade scaffold]

there is no 7800x

theres only the 7700x and 7800x3D

[sacred seal]

Sorry, typo

7700x*

[jade scaffold]

LMAO

[sacred seal]

What is that in?

[jade scaffold]

Blender

and then Adobe premier pro

and it anihilates the 5900x in Photoshop by an Osene margin

so does the 7600x

[sacred seal]

That is interesting, because the 5800x3d is getting creamed in those same benchmarks as well

[jade scaffold]

yea because it willdo

[jade scaffold]

the 7700x has a 13% IPC lead and like a 18% clock speed lead over the 5800x3D

the 5800x3D isnt a work chip

thats what the 5950x and 5900x are for

[sacred seal]

Nither is the 7700x

[jade scaffold]

and yet it bodies the 5900x and would boddy a 5900x3D

and a 5900x3D even for relativly cheap i could probs get AM5 with a 7700x going

[sacred seal]

Actually, I guess that makes sense, the 3d cache is only really good for gaming

[jade scaffold]

or fluid sim

theres a reason Epycs can be configed with like 1.1GB of cache

yea no the 7700x beats or matches the 5800x3D in gaming the VAST majority of the time according to gamers nexus when it looses its usually so close you wouldnt be able to tell

7700x is about the same price as the 5900x is which means if you already have AM4 5900x the way to go if you need all core work load

but if a 5900x3D is like 350

sorry 5900x3D the 7700x is better value

[sacred seal]

Yep, but the cost of platform ownership is so much lower with am4 that the 5800x3d was recomended over the 7700x at launch

[jade scaffold]

If you already have AM4

DDR5 WAS expensive

its not anymore

there were no budget boards at release

there are now

[sacred seal]

Its still up there, and the new boards cost more still

Its much much better, but I think it still comes out about equivalent

[jade scaffold]

you could get a 7700 a mobo and ram and it'd outperform a theoretical 5900x3D for 350 easy

[mossy gull]

I still remember the day we have 150.- costing GPUs with 512MB VRAM 😃

[sacred seal]

Wow, ddr5 really has come down in price a lot more then I realized

[jade scaffold]

from AMDs side of things
using 2x as many CCDs for a product that you KNOW will not sell well
VS using half as many for one that will sell well because its cheap

DDR5 is at where DDR4 was at DDR5s release

[mossy gull]

How do you think I could afford 256GB of it😅

Not by stealing out of a broken server

[jade scaffold]

DDR5 ECC however

[sacred seal]

It pretty well halved

[jade scaffold]

not cheap not cheap atall

[sacred seal]

I concede defeat that a 7700x build could be made for less then a hypothetical 5900x3d build

[jade scaffold]

also 7700x because its Zen4 gives you access to some actual work features

AVX512 babeeeeee

I need an ITX SP6 Mobo

[sacred seal]

AVX 512 is not really used in business. If your buisness can use AVX512, you are probably throwing a GPU at it

[jade scaffold]

its just nice to have as a consumer

for an actual buisness you'd never use a X3D chip either

not unless your using one of the Epyc 3D chips for that sick fluid sim action

This chips insane

I love these 8004 Epycs

you can get 3ghz all core at 180w which is the same power as a 7950x needs to get 5ghz all core but with 4x as many mores

would make for an awesome M4 Mac competitor PC line of chips

[sacred seal]

So a little over 2x the efficiency if my napkin math is right

[jade scaffold]

i really like them as lil HEDT/Home lab chips the epyc 8004 chips start at a mere £400

i feel like they could be used in a similar price bracket to Apples M chip PCs

1.8k for a 32 Core model
500 for the mobo
540 for RAM
1.2K on a RTX A4000 or B4000 when that releases

and boom 32 Core PC with insane efficiency 20gb of VRAM 192gb of DDR5 system memory for about the exact same price as the max configuration M serise mac studio Not bad not bad atall

Pc would only pull like 300w under load max

[cerulean helm]

Well, I do not like the forced image cleaning at the 4 hour mark for the LG OLED. Hoping it doesn't try that while in the middle of a game.

[jade scaffold]

My Samsung one just does it in the middle of the night when I go to sleep

QD OLED is also much more resistant to burn in

[mossy gull]

Yet currently most QD-OLED panels burn-in faster than an LG C1

[cerulean helm]

Uh, no? At least with LGs WOLEDs. RTings own 3 month testing shows the LG C2 and G2 show no signs of any burn in vs Samsung and Sonys QD-OLEDS. Their findings indicate it may be due to QD-OLEDS having no white subpixels.

https://www.rtings.com/tv/learn/longevity-investigations-update-3-month

[mossy gull]

I can confirm, I had an C6 or C7 Curved, neither showed burn-in, but the display died, probably PSU issue, I simply bought the C1.

[jade scaffold]

Gen 1 did

[jade scaffold]

Most OLEDs generally dont experience burn in

its not actually really a thing on new oled pannels not unless you leave an immage on them for EXTREME periods of time

most burn in tests are these extreme time periods without pannel cleaning which means pretty much all OLEDs begin to experience burn in

[mossy gull]

It's usually image retention, and it's usually fixable

[jade scaffold]

most OLEDs do cleaning any oled that does cleaning will never actually have burn in

not unless its like a really badly designed monitor

My G9 OLED ive had for 10 months now i think

and it gets used atleast 8 hours a day and its not got the slightest hint of burn in or any immage retention despite my discord window being in the exact same spot for the last 11 months

[mossy gull]

I love how blue lights, especially blue LEDs have been a pain in our backs for probably 100 years

https://www.pcgamer.com/hardware/gaming-monitors/lg-says-its-new-dream-oled-tech-cracks-the-pesky-problems-that-prevent-oled-gaming-panels-being-brighter-and-lasting-longer/

[fierce monolith]

Thank you guys for the help

[mossy gull]

What help?

[fierce monolith]

You guys help me fix met game remember

[mossy gull]

Oh yeah, that's what we're here for

[fierce monolith]

Well
Thanks you guys you didn’t have too but ya did so thank you

I’ll name my kid after you if I ever have one

[mossy gull]

I doubt you want to call a kid Aruthymysu🤣

[fierce monolith]

It a great idea

If I ever get a girlfriend she hate me for that but I love it

[mossy gull]

https://youtu.be/1KsTsODkTJ4

[cinder lagoon]

Man that must suck xD

Tldw tho?

[mossy gull]

Yeah, that's how bad Intel is currently.

[jade scaffold]

watching the video its really odd

no one went to epyc because of spectre/meltdown because spectre/meltdown are a problem with branch prediction ANY CPU with branch predition or speculative execution is vunerable to it

Spectre specifically just refers to the method of training the branch prediction to fetch data to CPU cache it shouldnt then cleverly getting data out of that cache because CPUs dont clear their Cache untill it wants to overwite something in the cache

it works on AMDs chips too

seems AMD arent even upgrading GPU memory next gen which is unfortunate a lil extra bandwidth is allways nice though with next gen not careing about being fast, and GDDR6 being easier and cheaper to get than GDDR7 will really help with the value of next gen cards

[mossy gull]

Could be cost and demand related, if Nvidia takes all GDDR7 and 6X, then AMD has less.
And them in this case it's better going with GDDR6.
There currently is a huge issue with memory development where they simply can't make enough.

[jade scaffold]

yea not to mention AMDs next gen isnt even gonna be much if any faster

so faster memory probably wont help that much really

[mossy gull]

I mean, Nvidia did release a 4070 GDDR6 due to supply shortages

[jade scaffold]

thats because of a weird thing with GDDR6X

but also GDDR6X unlike GDDR5X isnt particularly impressive

sure GDDR6X is faster then GDDR6 but not by an ammount that matters that much

with GDDR6 you still get like 95% of the memory performance which makes me wonder why even bother with 6X

[mossy gull]

I personally think that GPUs just should start using HBM by now 😅

[jade scaffold]

1st there isnt enough HBM for that

2nd

GPUs arent fast enough For HBM in gameing

only maybe the 4090 and even then

[mossy gull]

Still, if GDDR is gonna face a supply demand, we need to eventually look at alternatives.

[jade scaffold]

you know what faces a supply demand 30x that of GDDR

HBM

[mossy gull]

We can't keep hanging onto GDDR

Same that X86 eventually goes over to ARM

[jade scaffold]

HBM is not only expensive to Produce
its expensive to impliment you need a more advanced memory controler which will usually be physically larger taking up space thats batter used for things like more cores

x86 wont go over to ARM

thats silly

i dont know why i keep having to explain this to you

[mossy gull]

Then make something that's a replacement for HBM😅

[jade scaffold]

ARM IS NOT MORE EFFICIENT THAN X86

[jade scaffold]

what you mean just other gddr

making a new mem standard will just encounter the same issue

instead of taking up fab space making only 2 types of memory now you want to make 3

[mossy gull]

I am not talking about efficiency, I am talking about how the CPUs do their calculation stuff.

[jade scaffold]

the same thing

x86 wont transition to less instructions like ARM

it certainly wont be dropping execution units

The closest X86 will get to being ARM like is the advent of shrunk down more compact cores for mobile devices

but fundimentally in terms of basic math X86 does math the exact same way ARM does

multiplication and addition are still multiplication and addition

[sonic meadow]

what x86 has is all the multimedia and stupidly wide vector stuff, which is a substantial amount of its die space

and is getting bigger

re gddr: you cant really get any simpler than ddr, so you cant really get any cheaper and easier to manufacture

[jade scaffold]

yea the Vector and floating point units of X86 Cores accounts for a whole bunch of space

Cache too

[sonic meadow]

a bit (ie memory cell) in dram is a single transistor.

[jade scaffold]

X86 cores also generally target higher clock speeds so you have to make cores phsycially bigger to avoid cross talk its how X86 CPUs are capable of pulling 50% higher clocks than ARM Cores being bigger thoe does hurt efficiency but theres nothing stopping you from compacting down an X86 core knock off some of that enormous cache sacrifice some of that clockspeed to squash it all together and then you get a really efficient core

[sonic meadow]

clock rate is just a measure of clock rate 😛

[jade scaffold]

also yea DRAM is just 1 transistor

its 1 transistor and 1 capacitor

gods most teeny tiny capacitor

[sonic meadow]

techinically yeah, but they dont bother with the capacitor. the transistor itself has enough capacitance for it to work

they cheated

[jade scaffold]

but yea you can crush down an X86 core to be real small then they end up more efficient than ARM cores in a large number of aplications

Zen4C is pretty good evidence of that

[sonic meadow]

low power laptop (ie sub 10w) and tablet/etc chips are a great example of that.

[jade scaffold]

32 cores at 150w all fixed at 3.2ghz is ALOT of performance

X86 laptops untill this gen have pretty much just used identical designs to desktop cores

so they are actually not amazingly efficient

going full tilt they are very efficient

but if all you're doing is browsing watching YT what ever they have a whole bunch of extra crap thats just not needed

hell for APUs the GPUs so slow that even for gaming all that extra high performance crap isnt needed

[sonic meadow]

intels were distinct-ish for a bit, but when you run them that far along the efficiency curve things get great.

[jade scaffold]

The fuck you actually doing with thoes hyper threads
what the hell you needing 32MB of L3 Cache for in a laptop isnt doing shit other than consuming power

[sonic meadow]

not the mid or high power laptop chips, but the low power stuff

[jade scaffold]

its why i found the Zen4 C cores and Zen5C cores so interesting

for a normal laptop for ANY gaming APU they are far more than fast enough and they SIP power

intels E cores are even more efficient but they have a caveat or 2 if you want a purly E core CPU as they dont have all the extensions AMDs zen4C cores still have all the extensions

[mossy gull]

Calculate faster, that's why you needmore cache

[sonic meadow]

its honestly not a bad idea to cut all the fat crap out of your little cores if they are always going to be with BIG cores. since you shouldnt be using avx512 on your little cores anyway

[jade scaffold]

no

if you arent hitting the chip hard

that cache is literally not doing anything helpfull

its just sat there consuming power and its not an inconsequential ammount of power either

[sonic meadow]

L5 when

[jade scaffold]

thats just an SSD being used as a cache for another drive

or some weird CPUs have in built L4 cache

Some chips just skip L3 entierly and just dedicate like a 3rd of their die space to a big fat chunk of L2

[sonic meadow]

stick a pga/lga socket on the back of the motherboard behind the cpu for an upgradable external cache. say... 128-512MB/core

[jade scaffold]

cache isnt really done on a per core basis anymore i think only L1 and L2 is L3 is just unified cache so a big external cache would also probably be unified

the issue with cache is its REALLY REALLY Expensive

a L4 type cache might be less so

[sonic meadow]

depends on the speific type of storage you use

[jade scaffold]

part of the reason NVidias 4090 gpus cost so much is the fact that like a quartar of their volume is this obsenely large 100MB chunk of L2 cache

[sonic meadow]

cache is great

[jade scaffold]

infact Nvidia have been doing this for a while on their GPUs

if you look at the internals of alot of NVidias GPUs you'll see its actually kinda 2 GPUs connected to eachother by a fat block of L2, in their confrences they talked about this being them basically trying to prototype their multi GPU sollution for blackwell

[sonic meadow]

especially when dealing with the massive throughput of modern vram, and the increasing complexty of shader programs

[jade scaffold]

look at this silly thing

heres Hopper being even more silly

the L2 Cache is split in 2 and connected via an interconnect

[sonic meadow]

probably means you have compute groups stepping on each other for cache access less often

[jade scaffold]

as far as im aware the 2 caches in there dont actually know they are 2 caches

[sonic meadow]

though I am no semiconductor designer

[jade scaffold]

NVidia recently been pulling off some borerline magical shit with interconnects

This thing is Phycially 2 chips, not even the chip knows its 2 chips

as far as its concerned its just one really big chip

then they somehow got NVLink fast enough to trick an entire server rack into thinking its a single GPU instance

sad that it doesnt seem that its going to be making it into the consumer space with RTX 5000 woulda honestly grabbed a dual die 5090 in a heartbeat

[sonic meadow]

rtx 5950 gtx 😛

[jade scaffold]

Plz no

There's rumors of a titan comming that's gonna be called the and Im gonna puke

The "Titan AI"

Would it fucking kill you to call it something cool like Titan Blackwell or something

[sonic meadow]

AI is all the rage right now. both kinds of rage

[jade scaffold]

the titans not even using the new cool tech found in B200

its just the same as the 5090 and but with all the cores and cache actually active

[mossy gull]

ARC RXRGTX A6969 Ti Super Titan

[sonic meadow]

with knuckles

[mossy gull]

Let me edit it

That would happen if AMD, Intel and Nvidia would cooperate

[mossy gull]

12-pin can't deliver enough power?
https://wccftech.com/msi-dedicated-8-pin-connector-all-x870e-motherboards-next-gen-nvidia-rtx-50-gpus/

[sonic meadow]

Did you read the article?

Because your comment suggests you didn't. It also suggests you don't know about the slot power limit

It also isn't a new idea. I've seen Molex, 6pin PCIe, and I think 4pin ATX/8pin EPS used for this purpose before.
The purpose being to supplement the 12v coming across the 20+4 connector because PCIe devices are allowed to draw 75w through the slot

[mossy gull]

It's so the PCIe can provide more power so it puts less strain on the 12-pin

Wild guesstimation

Now they should put it on the back of the board

Looks nicer

[sacred seal]

Its to put less strain on the 24 pin from my understanding

[sonic meadow]

That's a very wild and completely stupid guesstimation

[sacred seal]

If for some reason you want to run multiple 40 series cards

[mossy gull]

Sounds right up to my alley 🙃

[sacred seal]

I remember my A+ textbook talking about molex on the motherboard, though I have only seen it in person once

[mossy gull]

It's how a video explained it.

[sonic meadow]

Remember that the 4pin p4 connector came out when CPU power was pushing like.. 50w or so. Which means they were worried about that little power over the 20 pin at the time (I think it predates the 24/20+4). And PCIe is rated for 75w/slot

[mossy gull]

I use to build PCs where HDDs were powered by Molex, and even fan controllers were powered by it, damn I miss those fancy fan controllers

[sonic meadow]

It helps that when you make a bold claim that the evidence you post actually supports it

[sacred seal]

I used to build PCs where the drives had ide cables

[surreal moss]

So did I

[sonic meadow]

As I

[mossy gull]

From what I understand is that the extra 8-pin is so it can push more power through the PCIe for the GPU.
That's what I understand from 2 videos.
So instead of 75w it can push 225w.

[surreal moss]

SATA feels so fancy in comparison

[mossy gull]

I remember the days of eSATA drives, those were the drives to get if you wanted fast external storage.

[sonic meadow]

The very article you posted (again, supporting evidence) mentioned multigpu

[mossy gull]

I missed the multi GPU part, my bad'

But hey, better news for people like me so we can have more GPUs.

More GPU, More POWA

[sonic meadow]

Actually I was mistaken. It's only 66w at 12v. The remaining 9.9w is at 3.3v

And when you also consider other PCIe devices connected to the motherboard like SSDs and the plethora of 2.5/5/10g networking, let alone.. USB 3/C and their power potential

[mossy gull]

USB-C 4 pass through you mean?
That I can completely understand.

[cerulean helm]

HomeLabs finally has a non rack mount NAS/server box. A bit expensive but glad to see more NAS/server chasiss that are not rack mount.
https://store.45homelab.com/presale/hl8

[jade scaffold]

no

what the fuck are you rambling about

why are you like this

its like that moores law vid you posted before you clearly didnt watch that either

the PCIE Power plane mobo side can only really do like 100w sustained it cant actually do much more than that for anything other than a temporary ammount of time

mobos with lots of PCIE slots have been running extra power to them for longer than you have been alive

but also this doesnt increase the power to single slots

the pins in the PCIE slot itself arent rated for more than 75W and will be damaged if much more than that is put through them

[sonic meadow]

I was wondering when you would respond to that😆

[frozen igloo]

Any recommendations for replacement m.2 screw and stand offs or are just about any of them online fine?

[sacred seal]

A screw is a screw, though m.2s are not standardized

Though on mobos most of them use the same

[sonic meadow]

Likely m1.5 or m2.. by about 3-4mm

For the screw, the standoff could be anything at the motherboard end

[jade scaffold]

i hate finding M.2 standoffs that work with mobos

server mobos never come with M.2 screws so you have to buy the screws for them and its also a pain in the ass because noboddy will sell you 2 m.2 screws so you end up with a bag of like 3000 of them

[sonic meadow]

I was doing some upgrade work on my brothers pc about a year back (had built it a year or two prior), and wanted to use the second m.2 slot... luckily we never threw out any of the packaging and found the clear plastic bag with the tiniest screw mankind has ever seen in it.

(yes I know watch screws go even smaller)

[mossy gull]

It's more annoying that they aren't interchangeable, I have many M.2 screws, salvaged from broken boards, and it's always an annoying fit to see if it fits.

[sonic meadow]

im just annoyed they didnt start with a toolless design.

[mossy gull]

https://tenor.com/view/arbiter-were-it-so-easy-halo-difficult-gif-16171194

[sonic meadow]

"oh, lets upgrade from a toolless design (sata/pcie) and force you to use the fiddliest screws imaginable, while also needing to use said screw to push the ssd in to the spring loaded connector"

[jade scaffold]

m.2s are so simple to do tooless too

they dont have to be pressed down to even work so its not even like thats the actual concern

you could just press them down with what is in effect a TV remote battery cover and that'd be fine

[mossy gull]

I love those little clips you see on those god forsaken expensive boards that are not even a €0.05 add on

[sonic meadow]

dont need the bottom contact/spring, but the actual latching part is perfect.
though given that m.2 mounts are variable length you would still need the ability to move it around

[jade scaffold]

yea the tv remote battery cover i think works best

just a metal one of thoes with a thermal pad will work on any size of SSD

[mossy gull]

I love when only 1 of the 4 M.2 slots are toolless

[sonic meadow]

I am not familiar with whatever kind you are meaning, since all the ones I know use AA/AAA

[jade scaffold]

like these kind of things

i just mean the battery cover

[sonic meadow]

oh

hahaha

[jade scaffold]

think about it it'd work so well

[mossy gull]

SSDs should be a kinda click in like what SD cards or CF cards have

[jade scaffold]

its tooles its easy to make holds the ssd down can be made to look pretty nice

[mossy gull]

But that's not what they want, they want the janky mechanism they have now

[jade scaffold]

no i think they are actually just stupid

its one of thoes things thats so simple it causes oversight

[mossy gull]

Cameras have a "click in" solution, basically an SSD in a casing that fits the annoying form factor of the camera, but it's easy to use and it is on some an hot swappable NVMe SSD.

[jade scaffold]

theres no advantage monotary or otherwise to making M.2s a pain to install for no reason

[sonic meadow]

I do think the form factor itself is rubbish, but the mounting itself just feels like unnecessary cost saving

[mossy gull]

What if, just hear me out, what if SSDs have the mounting pins like CAMM2 has, so at the bottom, and then you push the SSD that's in a casing down like it's a battery?

That would make it even possible on laptops to have an SSD with heatsink as the connectors lay much lower.

[jade scaffold]

i dont mind M.2s form factor tbh

its super compact for how obscenely fast it can get

13000 MB/s is alotta drive

my main issues with M.2s is no one is making low latency ones which matters more than you'd think its how the Gen 4 optanes can load things faster than the top end Gen 5 SSDs

Solidigm have some low latency drives but they are enterprise only and are super expensive

[sonic meadow]

its more that it encourages (and was obviously designed for) being put between PCIe slots, which means it gets obscured by graphics cards, which also impedes cooling ability and ability to remove/replace/install on a commissioned PC

[jade scaffold]

the origional M.2 slots used to stick streight up out of boards

like a silly lil PCB skyscraper

[mossy gull]

Maybe too little consumers as for 1.

[jade scaffold]

the thing is low latency NVME SSDs are the only correct way to go about SSD design

[mossy gull]

It still is possible on some boards I believe, or it's a joke from YouTubers

[jade scaffold]

having silly transfer speeds doesnt do SSDs much good if the access latency is dogshit

[jade scaffold]

you dont have to screw down NVME SSDs and they'll stick up at an angle but otherwise work fine

[mossy gull]

But how do we know the latency if all they benchmark them on are Mbps and IOPS?

[jade scaffold]

because A they state it and B you can just test access latency

[mossy gull]

The question is, would low latency a good thing for consumers.

All I know between the 970 Evo Plus and the MP600, that the MP600 feels slightly faster

[jade scaffold]

the access latency for optane at its peak was between 1.3 and 0.7 nano seconds

the PCIE Gen 5 aorus 10000 2tb has a access latency of 0.4 Milliseconds in some cases over 100x higher latency

[jade scaffold]

yes it is

its like asking is RAM a good thing for consumers

you want minimal latency

[sonic meadow]

eh, you want good enough latency

[jade scaffold]

solidigms data centre SSDs arent quite as low latency as optane however they have managed to achive 13 nano seconds

which is still in excess of 30x better than a normal PCIE Gen 5 SSDs

[mossy gull]

I didn't know, is that why MP600 feels snappier than 970 Evos?

[jade scaffold]

Optane consistiently dances on the graves of Gen 5 SSDs despite being theoretically Half as fast in raw data throughput

[mossy gull]

I don't really care about Mbps, I care personally more about IOPS

[jade scaffold]

less latency is allways better

the main thing we want is a good balence between low latency and high transfer speeds for an affordable price

[mossy gull]

I seen SSDs with 7 or 8 Gbps, but lose from a slower SSD just because the slower SSD has higher IOPS

[jade scaffold]

the PCIE Gen 4 optane outperforms the theoretically Twice as fast Gen 5 990 Pro 2tb in IOPs

which it shouldnt

but it does anyway

because modern Drive Design asside from solidigm and intel is DOG SHIT

[mossy gull]

My experience with Samsung often been like it's fast at the start, but slows down a lot afterwards.

I don't know if it's something with all Samsung drives or just mine, I only had 2, 1 OEM and 1 970 Evo Plus

[jade scaffold]

but if you want the same overall bandwith the samsung drive has you can just get 2 optanes and then end up with well over double the IOPs

No samsung drives are just mid as fuck

Blazingly fast™️ sure

but basically only because of their overkill Read Write speeds

They have high latency and their controler is simply bad

and their value proposition just isnt there

[mossy gull]

What about Corsair SSDs?

[jade scaffold]

not much better tbh

much of the same

[mossy gull]

I mean, my 970 Evo just felt slow compared to my MP600

[jade scaffold]

just usually found cheaper

[mossy gull]

What about Sabrent?

Intel Optane are pretty expensive here, like 250.- for just 1TB

[jade scaffold]

yea Optane is expensive thats the issue with it

Solidigm are doing some promising things with low latency dives not as low latency as optane but still 30x lower latency than the competition

still arent going to be cheap drives but the thing is with both these solidigm dives and optane is they will NEVER die

they are rated to be written to for 5 years straight and still not die

[mossy gull]

As long they aren't Intel expensive

[jade scaffold]

you try that with a samsung drive and it'd be dead within 2 days

thats the appeal of Optane and Solidigms new drives they make great cache drives because they are so hard to kill

[mossy gull]

I haven't had Samsung drives after an entire series decided to Harakiri in less than a week.

[jade scaffold]

its the same for most other drives tho

they arent intended for constant writes

and for consumers they wont be experiencing constant full speed writes

[mossy gull]

I mean it was a firmware issue on the 990 which made them literally Harakiri themselves

[jade scaffold]

so what we need is a middleground low latency drive for cosnumers much lower latency maybe only slightly better life span and even slower speeds as long as the latency is low enough the speed doesnt matter that much

[jade scaffold]

yea im saying even when they do work they arent intended to be written to at full speed constantly solidigm and intels low latency drives are intended for that and its why they are so expensive

[mossy gull]

150.- for a 2TB Solidigm is doable, that's like 30.- more expensive than normal 2TB SSDs in my area.

But Intel are 250.- or more

[jade scaffold]

Solidigm chill

only certain solidigm SSDs will be the low latency type

i dont even think this one is the low latency type 60us is pretty speedy tho

[mossy gull]

What about the the P41 Plus?

[jade scaffold]

none of the M.2 drives are i dont think

[mossy gull]

Yeah I don't have space for none M.2 drives😅

[jade scaffold]

sadly solidigm dont list their specs as in depth for consumer drives as they do for datacentre ones

[jade scaffold]

might try and nab one of these actually

i have more PCIE gen 5 than i know what to do with anyway

thats fukin 14.5gb/s at only 60us

this isnt even a specifically low latency drive

[mossy gull]

I am looking at their M.2, Corsair compared to Solidigm
MP600 Pro vs P44 Pro(they're only 10.- difference in the Netherlands)

• 650K IOPS vs 1.4m IOPS (Solidigm wins by double the IOPS)
• 7 GBps on both
• 1400 TBW vs 1200 TBW (Corsair has more TBW)
• 1.7 million hours vs 1.6 million hours (Corsair having 100k hours more between failures)

I personally would do it for the latency for programs that need it, but the MP600 being sold with 4TB would make it better suited for games that don't really care about the IOPS.

[sonic meadow]

the TBW and MTBF have basically the same figures between them

[mossy gull]

Corsair does state their SSD is 3D TLC, Solidigm doesn't say anything about it.

[mossy gull]

I doubt I'll ever reach it, so I don't really care about TBW, I did however reach the TBW from my Samsung OEM drive, all I know is it had a data speed of 2.5 GBps and 300K IOPS, so it wasn't a fast OEM drive.

But now I am curious how a Solidigm would perform in Star Citizen compared to a 4TB MP600 Pro, if for me it's worth it'

I am replacing my SSDs anyways after 4 years, usually because they come in double the size for the same price as the old ones

@jade scaffold any ideas?
Shall I go Solidigm or stick with Corsair?

Or a mix?

[jade scaffold]

go with which ever

i like solidigm as a company

they are literally just Intels SSD devision

they actually still sell intel branded SSDs like their old ones

though i think intel the company still sadly holds the rights to optane

[mossy gull]

I guess I'll mix

[mossy gull]

I already thought the 660P sounded familiar

[jade scaffold]

yea they were split off from intel as their own company a couple of years ago

oh fucking hell

Gonna be waiting on a Used one for that shit lmao

[mossy gull]

Yeah, I'll wait till they pass this onto cheaper drives😅

[jade scaffold]

i have other priorities anyway

i have my Micro PC and my GPU upgrade to go before that

[mossy gull]

I wish more companies made 2230 SSDs

Sabrent, Corsair and WD are the only ones making 2TB, but nobody makes 4TB ones

[jade scaffold]

I want a tiny ultra fast SSD

My idea for my micro pc

Is to include a lil battery that when the pc is unplugged will give the pc enough power to go into hibernation mode

Then I want the pc to be able to boot back up really fast so the drive only needs to be small but I want to to be as fast as possible

[mossy gull]

I just spoke someone who wants a GPS inside their laptop so they'll never lose it or so they can erase data when it gets stolen or to use Google Earth for navigation.

I never heard something more dumb than that.

[surreal moss]

That’s built into Windows and MacOS/iOS by default

And isn’t that stupid

Or dumb

Why do you think it’s dumb?

[mossy gull]

GPS doesn't travel through walls.
You better can use stuff like Bluetooth or WiFi to track a device like a laptop.

[surreal moss]

The signal degrades but gps does work in doors. That aside it send like the person wants what is readily available

[mossy gull]

GPS works very very poorly when you live in an house with weaponised concrete walls.
I often get no GPS signal with my GPS as long I stay inside my home.

Yes I still use a TomTom.

[surreal moss]

I mean GPS can’t transmit and provide the functionality they want anyway. That said it sounds like they are looking for features that are already available

[mossy gull]

They're looking for accurate location tracking and the ability to erase a laptop when stolen.
But that's already available without GPS.

[surreal moss]

That’s what I said

Guess we’re fixating on GPS tonight

[mossy gull]

That person was, they don't believe you can track a laptop, or pretty much any device without GPS😅

I want to build a cyberdeck now😐

[jade scaffold]

Every time I see a cyber deck it's a fat bastard

It should be possible to do one at single GPU slot thickness

Gonna have to use an easy to cool CPU maybe a 7600 then heat pipe the cooler off to the side

[mossy gull]

I am thinking of using a Latte Panda for it, not some laptop component

[jade scaffold]

I just wish the panda was more compact

Not enough silly tiny mobos around

I kinda want to learn PCB design just for the purposes of making a super mini pc

Not gonna be doing stuff like ram not gonna need a chipset either only usb and 2 m.2 slots my main worry is vrm

That way I could make a truly compact boi

[mossy gull]

I want to make a cyberdeck that's like a 7" tablet, so I am pretty much space constrained.

I hate that every powerful tablet these days start at a whopping 11", which is huge.

And I want mine to last for a very long time, so if I am gonna use the Latte Panda I have to underclock it to hell and back to make it last the whole day.

I personally also dislike that tablets these days have to be as thin as possible, but by making them insanely thin also makes then not comfortable to hold.

Like thin tablets are suppose to feel "premium", but for me a ToughBook feels more premium

[balmy flicker]

shoot I want to down load modded Minecraft on to my steam deck but I there are separate options for linux versions but arch is not listed they are split unbuntu/debian and fedora

[mossy gull]

As someone who uses Debian and has used Arch, Arch is a pain to use, set-up is easier, but that's about it.

[sonic meadow]

shouldnt you just be able to use an ubuntu/debian build anyway?

[mossy gull]

Well, in theory Debian even supports Windows programs and Android apps.
But it doesn't currently.

Arch on the other hand was made to be easier to understand, but it's a bigger pain to work with.

Their fundamentals are the same, it's just the programmers from Arch are in the massive minority as Debian is more versatile.