off-topic-tech

[cold tide]

also likely to snap the foot eventually when connecting stuff, or strip the threads, it is alu by the looks of it

[languid gulch]

true, i just like when someone can manage to make things match so well and have an open stand to show it off

[cold tide]

yeah but I like things to be practical aswell

and for example, the bearings in fans arent made to be ran at an angle like that

[languid gulch]

with a tiny stand like that, i'd love to put something like an Assassin 120 on the cpu, and then a 280 rad on the GPU

but yea, the angle is a bit wonky

[night girder]

Reddit is weighing identity verification methods to combat its bot problem
The platform's CEO mentioned Face ID and Touch ID as ways to verify if a human is using Reddit.

Step 1. Flood the internet with AI garbage as tech companies.

Step 2. Tell the users, you have to fight the flood you created yourself, with privacy invasive measurements.

Step 3. Humans are idiots.

www.engadget.com/social-media/reddit-is-weighing-identity-verification-methods-to-combat-its-bot-problem-195814671.html

[languid gulch]

greedy fucks manufacturing problems to sell a solution, whether or not the solution actually works

mmo players know the bullshit well

also known: once it's crept in, even on one thing, it always ends up being ruined forever

so prepare to have reddit be garbage within a year

[night girder]

Not a big reddit user. Don't have account.

[mental oriole]

Same

[cold tide]

its only one of the ways they rationalize it

the other one is the classic "think of the children"

for example

ultimately they want a face next to anti-semitisms on the internet, since thats all they care about, really. They couldnt care less if anything or anyone else was hurt of foul-mouthed.

look at your own PM right now, vs. liedewij.

[twin dew]

Stupid me...
Just spend hours trying to force my phone to get specific IP via DHCP.
Problem was that I was trying to tell NetworkD to assign IP from different subnet than that interface...
Missed edit when combining two previously different subnets devices into new combined subnet...

As NetworkD DHCP server has basically no debug logging available.

And it wasn't complaining about the assignment, just ignoring it.

Thought it was Connman DHCP client on the phone ignoring the suggested IP and server not forcing.

I was poking deep into DBUS on the phone via SSH.

[night girder]

Shit happens Baldur 😄

I call it "chasing my own tail"

trying to solve a problem, I don't undestand, and eventually it turns out I fucked it up in the code or config 🤣
But once you start to deep into the depth of the issue, you lose sight of the culprit sometimes. Hard to explain.

Logging is so important in my opinion. Exactly because of these situations.

[twin dew]

For most obvious errors like that NetworkD does emit error about ignoring invalid value.
Not in this case... Equivalent of assigning "192.168.2.2" instead of correct "192.168.1.2" as the IP to be assigned to specific MAC address via DHCP...
So it was silenty ignoring that assignment, as the IP wasn't in right subnet.

[mental oriole]

Hate silent ignore.

[twin dew]

Sound of silence.

Not in here, this is off-topic.

Spamming same image on all channels is just rude anyways.

[soft bloom]

This picture broke my brain
is this title a clickbait?

[mental oriole]

What title?

[soft bloom]

This picture broke my brain
that

[twin dew]

Yes.

If seen without the pic directly attached, without clicking in.

[soft bloom]

I don't think in this case picture is any less clickbait

[twin dew]

With the pic the title references in view at same time, you have much more data to see if you want to hear the talk about the pic or not.

If the pic isn't shown, you have no idea what the title might refer to.

[mental oriole]

Still don't know the context of the title nor the picture.

Assuming that youtube video?

[twin dew]

Seems to be hypothetical.
That there is link with that "title" somewhere, including YouTube video where the thumbnail pic isn't the picture being discussed.
Is it clickbait headlining or not.

IMHO yes, if the picture being referenced in the title isn't visible in at least thumbnail form with the title.

Needing to click the link to see the "picture that break brains" makes it clickbait.
With the pic visible with the title, you can make some judgement about the contents without ever opening the link, even when the thumbnail isn't full size.

[soft bloom]

that from 3b1b

[mental oriole]

Clickbait if you ask me because my brain didnt get broken.

I did watch it yesterday.

[soft bloom]

I think that clickbait extends further.
it's not limited to obstructing information,
it is about introducing ambiguity in what is the topic of interest.
I would roughly define it as "giving viewer/reader enough room to think themselves what is interesting about it, instead of straight up telling"

in that case, it could reference ]spoiler] and [spoiler] but instead it was generic enough that anyone without intrest in math or art could think "oh that sounds extreme, I want to know more"
But it is in that kind of middleground where it isn't missleading at least and is used for good purpose

[soft bloom]

because it was gently squeezed throug hthe hoops

[mental oriole]

I was also semi familiar with the concept before hand from 3d modelling. The same concept applies to uv maps. Texture coords are mapped from 0->1, but it wraps around if you go above 1 (or below 0), so 1.5 is the same as 0.5. Even though the video example is very different, the same generic principle applies.

[alpine cape]

I overlclked my ram from5600 to 6000 using memory try it. I booted into windows it felt unstable(couldn't interact with taskbar using mouse,mouse in general was tweaky) but memtest86 no errors??(1 pass all tests)

[twin dew]

And memtest isn't be-all end-all test, especially at just single pass.

Memory Try it and like auto-OC almost never works stable.

It can give base from which to tune manually, but shouldn't be used as is.

[tribal kraken]

D5 pumps arrived excatly same day. Alphacool VPP from Finnish retailer and Bykski one from AliExpress

Now just to wait if one the ones in use dies

[dire igloo]

how is creating ambiguity not a part of withholding information?
you're intentionally withholding the disambiguation

[night girder]

Clickbait is just the act of misleading viewers, atleast that's how I see it.

If you have a thumbnail with pictures that don't show up in the video itself (happens a lot) or you have a title that's misleading, I would call it clickbait.

And a "broken brain" is already pretty ambiguous. How do you check for "broken brain"? Brain scans? Psychiatry?

Or do they mean optical illusions? Which also "break/trick" the brain.

The fact I have to think so much what a title could mean, it's ambigious and high chances of clickbait.

[dire igloo]

I mean, it is about the maths behind one of M.C. Escher's art pieces, so there definitely is an aspect of "breaking the brain"

but it's very bait-y indeed, tho I would not call it dishonest

[night girder]

What is "breaking the brain". Like I explain my reasoning a bit.

"Why the M.C. Escher's art piece is actual an optical illusion" - could be a title. If that's the case. Less ambigious.

[night girder]

You make a point there.

[pure karma]

i mean did your brain actually break?

[night girder]

One half walked out of the room.

[pure karma]

then its clickbait cause it dident break you brain lol

[dire igloo]

But also misleading because I wouldn't call it an illusion

[night girder]

I haven't seen the video. Was just one of my hypothesis.

[slow flint]

My new things :steamhappy:

[visual tree]

Ubiquiti ❤️

[slow flint]

10Gbe uplink that I may be able to fully use in 10 years
I ❤️ NBN 🇦🇺

Max we cab get right now in AUS is 2Gb

But I'm also NOT paying $180-230 per month for 2Gb internet

[visual tree]

Private customers can also get up to 2 Gbit/s here but I think 1 gbit/s is more than enough

If I downloaded games from steam/epic, I am not sure I would be able to utilize 2 gbit

Even if there was no hardware bottleneck on my side

[jagged snow]

New Intel chips seem promisingly competative?

That's good to see, it's been years

[sharp matrix]

Yeah the plus chips seem good for the price and performance so actually something competitive from intel this time around, lets hope it can keep this up going forward with nova lake to

[jagged snow]

Great too see, amd is already getting too comfortable after a few years on top

[sharp matrix]

Yes and the last thing we want is either one becoming comfortable, as we know what happens, intel was at the top for so many years and stagnation happened without much innovation

Why im happy to see intel stepping up like this and being competitive with amd

[jagged snow]

Yep, absolutely~

[languid gulch]

i wish that HDDs were designed so that if it wasn't the platters or heads that went bad, you could yank them out in a module, and throw them into another drive case

at least in a way that'd let you get the data off it yourself, even at a crawl, and onto a good drive

[languid gulch]

i lost an 8TB drive last year

thankfully it wasn't full of anything critical, still sucked

[languid gulch]

i still have a couple moving boxes full of CDs 🤣

[cold tide]

That said what game is cpu bound currently to any degree where it matters. (ie not talking about cs2 running 500fps whilst being cpu bound).

It will be 2 years into AMD Magnus powered consoles before that starts to become an issue again. Due to increeased AI and RT stuff.

[sharp matrix]

Insert any number of factory building game here

[twin dew]

Also lot of badly coded recent AAA games.

Doing stupid stuff each frame on CPU, causing FPS issues if not on X3D CPU.

In cases leading to me with 7800X3D and RX 6600 getting better and more importantly much more stable FPS than people with much better GPUs, but non-X3D CPUs.

[sharp matrix]

Some good examples I can think of include monster hunter wilds, which from what I understand isnt very well optimized and is heavily cpu bound because of it

[twin dew]

Lot of weird and unnecessary CPU to GPU dependencies each frame.

Common with Japanese game coding, still not broken off from old-school console coding conventions.

[sharp matrix]

In either case regardless of the code coding practices. It's clear the need for better cpu's isnt going anywhere anytime soon, at least until some theoretical future in which we aren't cpu bound in any use case, but that probably wont ever happen

[mental oriole]

I cant be the only one who hates the new way to find new releases on artists you follow in spotify...

[cold tide]

this is often single threaded limited, currently both intel and AMD are moving horizontally (threads) more than they are vertically (ipc/clocks). Though zen6/7 are rumored to increase this quite a bit.

[cold tide]

i think this is people misinterpreting lots of cpu cycles wasted on cycling memory from and to vram as cpu overhead

that game really, really doesnt like 8gb gpus

[cold tide]

they still havent fixed pajeets uploading their crap onto verified artists page via that exploit. I think they really dont care.

[twin dew]

It does lot of synchronous to GPU feeding shit on CPU for no real reason.

Stuff not related to feeding the GPU at all.

[cold tide]

which gpu did you check this on, and with what tooling?

[twin dew]

I didn't, the people coding VLK3D did some reverse engineering to find out why stuff was slow.

All roasting like that I do is based on VLK3D, Wine or Proton devs reverse engineering when trying to find why stuff is broken or slow.

Or causing hitches.

Like this case with Death Stranding 2 doing insane stuff as latest.

That personal experinence was with one of the recent Star Wars Jedi games with max details.
Internet full of complains about stuttering and low FPS randomly.
No issues for me, but don't remember the technical details about what it was doing in background causing that with most CPUs.

[twin dew]

IIRC Elden Ring originally ran CPU and GPU things one after another on each frame.
No starting processing game logic on CPU for next frame until previous frame had finished rendering on GPU and displayed..

And then GPU in turn waited until all the CPU work was finished, and then fed in as one unit for the GPU.

FromSoftware is manic about still getting frame perfect controls.
And IIRC Monster Hunter devs are close.
Leading to weird stuff from need to sync CPU and GPU processing so closely.

[waxen pendant]

i agree with this, I think elden ring is still locked at 60fps on pc

[twin dew]

When we don't live in NES/SNES era anymore, where that was the way you had to implement things.

Don't remember when console coding for most part moved away from that, around 2000 +-5 years.
But several large japanese devs still try to do it.

[twin dew]

In turn from Korea:

crimson desert checks for wave32 support
i'm sure this game will age well and continue to be playable on hardware for generations to come
just like, say, sims 2

Where that wave32 vs. anything else as how the shader code is exactly parallelized is GPU driver internal thing...

And same game on start with Intel IGP/GPU just gives paraphased dialog: "We don't support Intel".
And Intel had been in contact with the developers early, trying to offer free hardware and engineer & programmer support, and was declined.

Sims 2 had excessive hardware compatibility checking, that only supported GPUs that were available when it was launched.

And for anything newer, you had to hack the game.

And the stupid thing was, that was just to determine default settings it started with when first ran...

No fallback path at all to just set some default if the GPU wasn't explicitly in the list.

[waxen pendant]

this is fun

[twin dew]

I wonder if those Pearl Abyss devs wrote the game to send everything to GPU in explicit 32-wide batches, instead of the normal and faster "send everything using same shader at once and let GPU driver handle the splitting to GPU execution units".
Which would explain that "No Intel" as current Intel IGPs and GPUs have 16 wide units.
Even when that wouldn't matter, as when that Intel driver gets 32-wide set, it will just execute it on two 16 wide units instead of one 32 wide unit...

If you want to run a shader on every pixel, you don't split it up into batches, you send all 2M+ in one command, and the GPU driver will schedule that stuff on the GPU execution units, interleaving with other shaders if you don't fill the whole GPU with one command, or if the last set of the command doesn't fill the whole GPU, starting with the next set on free units instead.

I have two computers and 6 monitors, I can do multiple things at same time, pausing games etc. if necessary to momentarely write on Discord or like.

[twin dew]

Underfilling is bad, sending instruction bunches that don't fill execution units, as that makes some of the GPU just idle.
Overfilling is not and is how everything is supposed to be done, with GPU driver then dividing the work to the GPU, and interleaving multiple instruction bunches as needed.

Now I'll head to get that new as warranty replacement switch and grocery store, then return to gaming on main monitor.
While having IRC open on one monitor, Youtube on another, and Discord on couple more.

[pure karma]

This i what i meant by the new firefox searchbar, its not officially in use yet because it caused compatibility isues so it was disabled last minute but you can still access it if you have modded your firefox to not use the address bar since its obviously still in the code just not in use yet

[twin dew]

On plus side, got the newer V2 HW for the switch as factory fresh new package, and not refurb or like.
On minus side, it uses different model wall-wart, so need to replace that too with all the cable management.

running expect-assume over crimson desert now, so far, local array OOB, groupshared OOB, derivatives in non-uniform control, texture() in nonuniform quad control
jesus christ this game is so broken, how does it even run on native

Game devs just keep on coding to work on current drivers, and not to the actual spec.
Using various undefined and straight up illegal by spec functionality that just works with current drivers for whatever reason.

[flat ether]

I have a question for some tech possibilities in game. Currently, NPU/Blackwell enabled players have the ability to develop their own AI Agents. What I want to know, are we capable of interfacing with Satisfactory, to have an AI Agent that can learn how to make the game as efficient as possible with building?

[umbral marten]

Official Modding Discord

Learn all about modding and join the community here!
<3 @twin dew

[mental oriole]

Wtf did I just read...

[alpine cape]

My pc takes unusually long to boot and the VGA light is on shud I be worried?(4060;b760m)

[twin dew]

Lot of jargon about running some kind of testing on the game, and then complaining about stuff it does against DX12 spec in shorthand.

[alpine cape]

..help?

[twin dew]

Not much we can say that your MB manual wouldn't.
That if it is stuck with VGA light on, it cannot see/start your GPU for whatever reason.

[alpine cape]

can i somehow fix it without opening my pc and clearing the cmos

[twin dew]

And this isn't any kind of tech-support channel, so someone might jump in and help, or not.

[alpine cape]

yea js panicked sry

[twin dew]

Which I wouldn't expect to help, unless you did some tweaking just before and then you should know which changes probably caused that.

[alpine cape]

i used memory try it 2 days ago to overclock ran from 5600cl40 to 6000cl40 then tdy i installed a harddrive (it worked perfectly the next few restarts)

[twin dew]

Which should cause RAM error, not VGA error, if related.
Unless something weird going on.

Unless it set so high voltages it fried stuff on your CPU.

[alpine cape]

thats what i thought too...

[twin dew]

Which I forgot to mention as common issue, that those Auto-OCs use way too high voltages.

[alpine cape]

the only thing that sint supposed to be is the fac tthat my cpu is rated for 4800ddr5(i512400f) whcih again, isnt related to gpu

[alpine cape]

right. il go revert it now

[alpine cape]

Is there a chance it magically fixes itself or do I need to wait 2 min everytime I boot?

Yaay now it says memory overclock fail eventhough i resetted it all to default in the bios

;-;

[twin dew]

So good chance you MagicZ:t your computer, aka majorly damaged/killed part(s) with bad OC attempt.

[alpine cape]

it wasnt taht bad(legit js used memory try it! to make 5600 6000 same timings)

[twin dew]

And point is that that is a MB maker Auto-OC feature, and those do often use way too high voltages.

[alpine cape]

oh

it was on for like.. max 8-9h

is that enough for it to susteain permenant dmg?

[twin dew]

Even second on too high voltage causes permanent damage.
How long that damage takes to proceed to be bad enough to be a problem depends.
And once that starting damage has been caused, it will continue taking damage over time even when that initial high voltage is removed.

Higher the initial voltage, or the continuing voltage, faster the process.
Up to that instant dead.

[alpine cape]

OH FU-

welp

[twin dew]

Question mainly comes, what did that feature use in your case.

But cannot be answered anymore.

[alpine cape]

as per the same mb manu(msi) everything was under normal limits except sm cpu vddq thingy whch was at 1.4v and was highlighted rred

[twin dew]

Basically active silicon is always degrading while in use.
Just that the speed in normal proper operation is very slow, and designed usually to make the worst sold units last 10 years of constant use.
OC without raising voltages is using the operating headroom left for that long term stability while the silicon suffers the expected degradation.
When raising voltages, you are in turn almost always shortening the lifespan.
With "safe" level OC raised voltages, not to extent it would become issue with normal gaming computer CPU replacement rate, but sometimes does even then.
If you make a mistake, you can dramatically shorten the life until instability at stock sets in, even if that mistake is only for short while.

Reason why I had to warranty replace my first 7800X3D, as it got hit by that AMD VSOC issue.
Set RAM to 6000, booted to OS, saw that VSOC was high, went back to BIOS, lowered it, still had time to take enough damage to get very accelerated degradation in IOD.
This was before the VSOC issue was known, got replacement later once it was known issue, as I had monitored things and noticed FCLK max starting to degrade, to about -200MHz headroom in 3 months.

And why Intel had to extend the 13th/14th gen warranties, as their firmware had issues where it caused momentary VCore spikes, causing damage, that will continue to affect the CPUs that got hit for long time, shortening their lifespan even if stable now with "fixed" firmware versions.

[alpine cape]

hmm i did run memtest 86 for about 2h before doing anything and it didnt report any errors. i didnt however take the time to check voltages

[twin dew]

And it might have been stable initially, until the damage progressed more.

[alpine cape]

hmm

so

should i run emtest nowagain?

to see if the oc damaged my ram

[twin dew]

That there is good chance your CPU is now toast and not stable even at previous stock.

[alpine cape]

gods. waht have i done

[twin dew]

If you got that OC failed message at previous stock now.

And not while trying to revert the changes.

[alpine cape]

nvm it wasnt fully stock i accidentaly turned on sm memory module mode i think it was called, and set it to performance(after removin all oc) it finally booted tho it took the same 2min of communicating with the gpu

[alpine cape]

the oc fail was due to said memory module mode

which was previously disabled(when i oc'ed to 6000)

[jagged snow]

What is this timeline 9to5mac.com/2026/03/24/your-wireless-router-is-now-banned-from-sale-in-the-us-but-you-can-still-use-it/

Helium shortage is also about to hit hard, 2020s chip shortage v3

[twin dew]

And basically no-one will start making those in US.
As putting up a production line even for just final assembly to level necessary to pass that "Made in US" would take year or two, minimum.
So US is fucked.

As the expectation almost certainly is that that ban will not last.

[jagged snow]

Yep, this is honestly insane to me

It seems to only cover residential routers

That's fairly broad

Is that only covering switch/router/wap combo units or does it also cover all those 4 port unmanaged switches, wifi extenders, and other residential internet products?

[visual tree]

"It’s important to stress that the ban applies to the importation and sale rather than the use of new foreign-made routers. You can continue to use your existing router."

So, you are not allowed to buy a router from canadian amazon and ship it to USA if I understood since importation is illegal?

[pure karma]

amazon dosent stock in canada so no but yes

[visual tree]

You could also buy a non-american router from virtually any store abroad but the router would probably get confiscated during import

This will also result in ISP's increasing prices for their services for sure

[twin dew]

They cannot get more routers at all, for multiple years in minimum, after their in-country stock runs dry.
So I would expect them to order gigantic loads of current models while they still can, before that ban goes into effect.

[visual tree]

In short, someone didn't think of the consequences or consult with advisors before they made a stupid decision

[twin dew]

Aka Trump had another bad idea, and the yes-men went to implement it, as part of the attempt to bring manufacturing back into US via all the wrong ways.

[visual tree]

As someone who works for an ISP, I don't even think we have a single router or network equipment manufactured by a company in US

Correction: equipment for customers

We probably have cisco equipment for ISP central offices

[twin dew]

Not even Apple does the final assembly in US.
And for the lower level parts, not available in quantity.
Not even casings or like, no large scale plastic injection factories etc. left.
Outside in-house ones for some large companies.

And then much smaller shops.

And don't even think about most of the electronics parts.

Somehow magically a US built router wouldn't have software vulnerabilities XD

[visual tree]

Bet it's going to be a shitshow once the law is passed

[twin dew]

As one of the basis in that actual doc:

Malicious actors have exploited security gaps in foreign-made routers to attack American households, disrupt networks, enable espionage, and facilitate intellectual property theft. Foreign made routers were also involved in the Volt, Flax, and Salt Typhoon cyberattacks targeting vital U.S. infrastructure.

https://docs.fcc.gov/public/attachments/DOC-420034A1.pdf

https://www.fcc.gov/supplychain/coveredlist

So some China-linked telecoms (equipment) companies, couple of Chinese surveillance camera companies with cameras that have cloud functionality in-built.
Kaspersky from Russia.
And now two completely random bans on drones and routers.

All previous from Biden era, two latest from Trump.

And basically it all boils down to this:

President Trump’s 2025 National Security Strategy stated: “the United States must never be dependent on any outside power for core components—from raw materials to parts to finished products—necessary to the nation’s defense or economy. We must re-secure our own independent and reliable access to the goods we need to defend ourselves and preserve our way of life.”
In theory good idea, but trying to implement it ass backwards.

From that same FCC PDF.

Lot of the tariff stuff was from the same source basically.
Trump having no idea how stuff works, trying to force rapid change to US manufacturing in everything.

Instead causing loss of manufacturing jobs from supply chain disruptions.

[visual tree]

We have a saying here: "Paper will put up with anything that is written on it"

Meaning ideas can sound good on paper but very difficult to implement in practice

[twin dew]

Even US changing to war production during WW2 took usually about a year per factory.
And that was with the factories, machines and operators already existing, just reconfiguring and repurposing stuff.

Now they don't have buildings, machines, materials or workers.
And no US company is going to start insourcing when the policy changes monthly even with Trump, and if Trump goes out, it will all be reset, just when the first project started right now would near serial production.

As said in many sources, only way to do switch like that would be to make clear 10+ year import tax schedule, as actual law, going up and up by the year.
With clear political support that it will not be just torn down later.

[visual tree]

Making insourcing decisions right now in USA must be a nighmare for companies since the orange buffoon keeps making and abandoning decisions based on how he feels every day

[twin dew]

Of course this is now much faster when you know what you are doing.
Which China does, but no-one else does.
Which is why starting new product production in China is so much faster and cheaper than anywhere else.

Taking weeks to months, instead of years.

When the actual factory building doesn't have to be built, that adds couple of months.

But retooling with the "generic" factories around there to make YOUR product.
With various independent factories doing different parts, plastic molding in one, metal stampings in another, PCBs in another, final assembly in another etc.
With fast switches between products being manufactured in each factory when each run is complete.

The complain from companies in US has been that it takes more time for initial design or manufacture of injection molding dies (one part only, without the final tuning that too has to be done as third step on the actual production line with initial production) in US, than it takes to get from 0 to finished run in China.
And the factories to actually use that die for major production runs just don't exist at all.

That experience and capability needs to be brought back to western countries, but it doesn't happen with what Trump is doing.

[visual tree]

Topic change

Does this solar panel setup mean 1 string?

From the numbering, I think it's 1 string connected in series

Asking because I need to know the diameter of a conduit from a solar inverter to breaker box

Oh, dumb me. Didn't read the instructions carefully 🤦‍♂️

It says 32mm conduit no matter how many strings you have for connecting inverter to breaker box

[twin dew]

If that is with one string, then AFAIK someone is messing up.
Because AFAIK all the panels in a string must get about same amount of sunlight.
Even branches etc. obstructing some of the panels can cause issues.
And that kind of two-direction set in same string...

But I might be wrong. Or they might be doing two strings.

[visual tree]

Panels will have optimizers btw because of potential shading which is caused by a chimney and different roof orientation

They are used in situations when you don't have the same amount of sunlight everywhere afaik

I will have this model: solar.huawei.com/en/products/sun2000-450w-p2-600w-p/

Also getting AIKO full black solar panels which look more aesthetically pleasing compared to standard solar panels

Was supposed to get TrinaSolar at first

I am so glad we will use AC for heating/cooling after installing solar panels. Natural gas price here has drastically increased and middle east situation atm is only making it worse....

Not to mention gas combi boiler is very expensive

How do you heat your home in Finland btw?

People in cities here who have district heating laugh at people who use natural gas because of high prices without realizing natural gas prices also affect district heating prices too

They don't know most thermal plants here use natural gas for district heating

They are like "I have no gas boiler in my apartment so I don't use natural gas at all"

Boy, are they wrong

[languid gulch]

https://www.reddit.com/r/SoraAi/comments/1s2ol8j/openai_set_to_discontinue_sora_video_platform_app/

[visual tree]

OpenAI will soon go bust

https://www.youtube.com/shorts/Df0uu3Kzvc0

[languid gulch]

lasted a whopping 15 months

[echo elk]

"not a bubble"

[oblique hamlet]

Good

[agile sun]

Hon hon hon.

I am here to try and figure out what part of my rig is slowing down my SF experience

Kinda want to be able to run the game Stabley before 1.2 leaves exp branch

For your information I do have cats

[pure karma]

start spitting specs quick or else your gona get boo hood since this is off topic

[woven vault]

the CPU utilization stat can be misleading since it takes all your cores into account. i would look at the single-core performance to see how it is really doing

[agile sun]

Hold On a hot sec

[woven vault]

sometimes reducing graphics settings actually makes performance worse by shifting work to the CPU instead while the GPU idles

[pure karma]

it shouldent get worse only stay the same

[agile sun]

My GPU Probably is a piece of crap since this is a pre-built and I never upgraded it

[languid gulch]

ouchie

openAI might not exist tomorrow 🤣

[pure karma]

good

[agile sun]

Good

[echo elk]

Good

[agile sun]

Net positive for the human race

[languid gulch]

yup

[pure karma]

fake money results in fake companies

[languid gulch]

now i have to see if NFTs lived longer

oh yea, by a long shot

[fickle gust]

Yes find system specs in task manager

Gives us the numbers along with ur ram speed I wonder if xmp ever got turned on

[pure karma]

XMP wont matter

its like at worse +-10% on ddr5

[fickle gust]

Xmp certainly matters

[echo elk]

Assuming ddr5 lol

[agile sun]

Right. you wanted specs i got you specs

AMD Ryzen 5 5500
AMD RX 6500-XT
32 GB DDR4 RAM

[echo elk]

Probs your cpu

[fickle gust]

Yeah thinking same

[echo elk]

It's fine but shouldn't make you get low fps

[fickle gust]

Also xmp would matter here so whats ur ram speed in task manager

[agile sun]

says 2400 MT/s

[fickle gust]

How much did u pay for this just curious

[languid gulch]

cpu is fine, 4gb 6500XT is ouchie

[fickle gust]

Oof

[echo elk]

Slow!

[fickle gust]

Yeah don't think xmp is on

[agile sun]

Like 900-ish

[pure karma]

yea that 6500XT is garbage these days but that 5500 although low end for end game should hold up fine

[fickle gust]

Okay not bad actually

Prolly still on the high side

[pure karma]

900 for a low end cpu and e waste GPU is not good

[agile sun]

it is my second Ever Desktop. I'n still trying to figure this out

[fickle gust]

Yeah but with ram prices and ssd prices

Ram and an SSD alone are like 400$

[echo elk]

You definitely want to check whether your ram goes higher than 2400MT/s, and if on bios you turned on DOCP

[pure karma]

probably 4 sticks of ram cause if it was 2 it should be at 2933 even with XMP off

[languid gulch]

SF without much construction stuff will suck down 10-12gb vram at 1440p max settings

[agile sun]

just bought a RX 7900XT of amazon as an upgrade

[pure karma]

that will help loads then

[agile sun]

sweet

[fickle gust]

Woah

Get either a 7900xtx or a 9070xt

[pure karma]

but yea a cpu upgrade wont hurt either if you wanna throw the money at it

[fickle gust]

Should be roughly same price

[echo elk]

Not that I've noticed honestly. I run a 9070xt but don't see that much

[languid gulch]

green?

[pure karma]

https://tenor.com/view/elmo-fire-elmo-green-elmo-mscelmo-gif-24557102

[agile sun]

my bro has xtx. dad didn't want to spend the extra for what he thought was just more VRAM

[pure karma]

iv seen 17GB of vram usage on my save

[fickle gust]

Ah what about 9070xt

Cuz that should be same if not cheaper

[pure karma]

and i only have a 12GB GPU so late game 12GB is not a stretch much

[languid gulch]

oh. i can change mine

[fickle gust]

It's like 500-600 for me and would be better

[agile sun]

no shit.

your kidding

[fickle gust]

I see one for 620

[agile sun]

you gotta be shiting me

might not be to late to cancel the order on the other card

[fickle gust]

I would not be basically preforms the same as 7900xtx

I think 4k wise it could be worse this is ofc assuming no modding the 7900xtx

Tho I've heard the 9070xt is a monster power hog too if u let it

So I'm hoping next gen amd is solid

[agile sun]

I checked the XTX vs the XT, it's just an 11% above the other

[fickle gust]

And does the same in terms of eating power and boosting too hell

[pure karma]

its not crazy especially at anything less than 4K

[agile sun]

I did also buy a new power supply

[pure karma]

not particularly worth canceling over unless you want to nitpick at that point

[agile sun]

yeah. I won't bug my dad about it

[pure karma]

that would have been a good sugestion BEFORE ordering not anymore really

[languid gulch]

there, better?

[agile sun]

This is the shit im not in the know about dude

im still a pleb

[pure karma]

im not either i never bothered memorising AMD's specs for anything

[fickle gust]

I wonder if the 9070xt beats a 7900xtx actually even in 4k boosted

[languid gulch]

depends on the vram needs

[fickle gust]

Shit true

Fuck I can't go down too 16gb dammit

[pure karma]

not in any reassonable scenario probably

[fickle gust]

Cuz a 7900xtx with 700-800w is so much fun

And a 9070xt at around 500w boost pretty good I heard

Me I bought my 7900xtx in 2021 when it came out

[agile sun]

well, I got a 1050 supply

[fickle gust]

Crazy too think 7900xtx is 4 years old it feels so old

[fickle gust]

Oh you'll be fine I'm talking modded where u get an unlimited power limit assuming u can cool it

[agile sun]

ah ok

[fickle gust]

If u mean do I beat 4090s in raster yes

[languid gulch]

i still have my $600 6600XT

[echo elk]

According to GamersNexus benchmarks it's usually a close tie between them. Depends on the game

[fickle gust]

So close

Really have high hopes for next gen

I think I'll be able too hit 3000mhz on my vram with fast timings

Which means 1:1 with soc which should have some nice latency gains

[languid gulch]

meanwhile i'm running my 4090 with a 175W cap 🤣

[fickle gust]

I like maxing out my 165hz display tho plus the power efficiency improvements have made me really hyped for next gen

So my issue is I like higher fps so I don't like capping my card

[languid gulch]

oh SF is capped on my monitor even at that wattage

[pure karma]

dang im jello

[fickle gust]

What

[pure karma]

im peaking at 50fps from a ram bottleneck of all things

[fickle gust]

But yeah I max out like 230 fps at 300w

[languid gulch]

ayup

[fickle gust]

Am using fsr4 tho

[languid gulch]

everything maxxed except RT

1440p

[fickle gust]

Oh I have everything maxed

4k

That's why ur 1440p no wonder u don't need power r

[languid gulch]

4090 is some kinda beast, even when you throttle it

[pure karma]

everything maxed with DLSS Quality but no FG i get 50FPS on my 4070ti which is playable but not ideal

granted it isent the problem but balming it is funier

[languid gulch]

it's also not a FPS so idc that much

atm i'm the one trying not to melt

[fickle gust]

God I remember when I limited it too 30 fps and it was horrible

[fickle gust]

Yeah when I water loop I some how wanna put the exhaust out side

[languid gulch]

redragon M913

[fickle gust]

Why? But the super light pro x2 or whatever from Logitech

[languid gulch]

i'm too cheap

plus my mouse has the 12 side buttons

for like $35

[fickle gust]

Only games u need more than 4 in is poe2

Or ones like it

[languid gulch]

$30 rn

lol

this one has the pinky shelf on it

[fickle gust]

I can't wait too upgrade if only my ram wasn't 1k that'd be cool

This build use too only be 4k too upgrade

[languid gulch]

i want a 3d printer just so i could make my own peripherals

if i'm gonna be spending hundreds on premium products, might as well ditch them, grab the good switches and keycaps, and make my own shell

lol no shit

i have big hands so it works for me

i ditched logitech because they swapped out their switches for shit ones

[pure karma]

i play with less than 15FPS AFTER FG on my laptop so thats nothing

scoria pro gaming mouse

[fickle gust]

Ur poor eyes

[pure karma]

i cant find the original recording without the crt effect but yea the framerate is the actual recordings framerate

its about double the fps in reality but obviously the encoder was shitting the bed

you kinda just get use to it

SF has no input delay up until less than 10FPS so its really not all that insuferable

what is anoying is it dosent register inputs from between frames like some games so thats very tedious

when 1/3 of your frames are real

[fickle gust]

That's gotta feel horrible too oof

[fickle gust]

I mean when u decide too have more than 15 real fps so to more than half the frames are real this doesn't happen

Also why so much static

[pure karma]

i think native dropped to about 12FPS before i started using FG to make it at least less of an eyesore

but now native is probably 8FPS or less

[fickle gust]

Do u have the first PC know too man or?

[pure karma]

nah 13620H 16GB of ram and 4050 in my laptop

my save is just a bomb

[fickle gust]

I just feel bad holy

[pure karma]

although just like my main pc its once agaim a RAM choke that is the performance limiter

simply too much data going in and out of ram to manage

[fickle gust]

It shouldn't be limiting u that much tho

Even at 4k max settings I'm only at like 20gb of total ram usage

[pure karma]

when the game is using over 16GB of RAM and managing some RAM as VRAM too shit starts to fail

[fickle gust]

Pagefile?

[fickle gust]

And even at 4k pretty sure I don't use over 16gb of vram

Oh u prolly have less

[pure karma]

iv peaked at 17GB at 1440P and i have a 12GB card

[fickle gust]

Maybe it was more I'll watch it but wack and if so turn settings down

[fickle gust]

Or wait not here

https://tenor.com/view/sealion-sea-lion-sea-lion-nod-sealnod-gif-13740382386571380611

Or it's slow mode and I'm dumb

[fickle gust]

Wait what

[pure karma]

shared memory

[fickle gust]

Yeah ik that but like u said ur running out of ram too so like lower settings

[pure karma]

ram has nothing to do with settings

this ram bandwith issue happens when the game starts to load item models but not the hitboxes and the ram bandwith gets bloated by constantly updating hitboxes instead of important stuff and in turn performance goes to shit and so does the ram usage

[fickle gust]

Ram and vram has everything too do with settings

The higher they're the more vram and ram ur gonna use

So ur case u should be playing all low

[fickle gust]

Unc what

[pure karma]

above Uobject limit shit you probably havent had to deal with dont worry about it but its a real thing

[fickle gust]

Oh I've gone above the object limit so I just leave it at max for every save

Only took me 200 hours

Kinda annoying I figured out there was one

[pure karma]

well to be fair its unreals fault and not the game at least

[fickle gust]

Oh ik it's kinda unreal how dogshit the unreal engine is now and devs who use it mostly

Most devs can't even do HDR properly

And that's easy

[pure karma]

unreal is the best tool for absolute experts being used in the worst way and its really starting to grab the wrong attention

[fickle gust]

Speak yo facts twin 🗣️

[pure karma]

unreal never has and never will be for beginners no matter how much ai slop and guides you throw at it especially with whatever blueprint system they aded that probably gave me brain damage the first time i saw it

[fickle gust]

I'm just hoping css does HDR well if they add

It cuz I talked too someone. Who said they'd look into it

Css seems too be pretty good with ue5 as it's fairly well optimized and it's really pretty

Lod could be better

[pure karma]

its VERY well optimised

[fickle gust]

I agree

[pure karma]

i call it the most optimised game ever even if its probably not true but the optimisation to content ratio is the highest iv personally ever seen

[fickle gust]

I can agree there

Honestly arc raiders did it pretty well too I think

I'm on 7900xtx and was rocking like 200-300 fps with raytracing on

[pure karma]

thats pretty impresive

[fickle gust]

I still think it was cuz of a mod

[pure karma]

it took me like 600 but then again i spent probably half that building trains so whos to say i was trying

[fickle gust]

I had powered foundations and I feel like that caused it

Idk for sure but I feel like I didn't truly hit the limit

[sharp matrix]

UE blueprints is a abomination

I remember when i first saw the system. Its so much worse than dealing with c++ literally any day

Because beyond very simple usage, the graphs become a bloated mess that is increasingly harder to follow

Aka those blueprints can go back to the shadow realm thou came from

[fickle gust]

Sometimes I wish u could like make black on an OLED not pitch black

[echo elk]

Isn't the point of oled to have ultra dark black

[slow flint]

Yeah lol

[slow flint]

That's kinda OLEDs thing though
If you're finding stuff is TOO dark when it should be brighter then maybe mess around with your monitors brightness and colour settings

The blacks are a big part why you get OLEDs but if it doesn't seem right, probably a factory defaults issue
Like how TVs have the most dog shit settings enabled by default like motion smoothing

[languid gulch]

i wish that miniLED had taken off more than it has by now

[fickle gust]

No it's just like having a black background u can see the monitor like smears so much worse

Like cuz cleaning it is a bitch

It's so annoying gets dirty so fast and then if u clean it wrong it smears a tad and it just bothers me

This isn't really real

I had a LG c1 I abused and no burn in I gave it too my friend and he's been chilling on it for almost a year now and I got my LG g5 which is pretty peak

And on newer panels it really isn't real unless u just leave it on 24/7

Plus I mean mines comes with a 5 year burn in warranty so after 5 years I get a brand new TV too give too my friend when I upgrade

[slow flint]

Your manufacturer will have recommended cleaning instructions online
If you don't wanna follow those I'm sorry but it's a skill issue

[fickle gust]

Yeah I just didn't let it cooldown cuz I forgor

[twin dew]

And even that is misleading with most tools.
As Windows moves threads around multiple times a second, and almost all the tools that show per core CPU usage have longer averaging periods.
There are specialist tools that monitor per thread CPU usage, that show the actual situation, if any single thread hits 100% single-core or not.

[twin dew]

If talking about Satisfactory, wrong place.
But if you get stable 5-15 FPS, you hit Unreal Engine internal (total) VRAM usage limiters.
With 4GB card, you cannot have any HW accelerated programs in the background when running the game.
Includes Discord, Steam, Spotify etc. if that HW acceleration is on in them.

Using DirectX11 just makes random textures go to blurry mess instead of dropping performance, but has other minor issues.

[fickle gust]

Speaking of what's best for this game amd wise vulkan or dx12?

I've left it on dx12 mainly cuz it says it's forced and doesn't let me change it but I wonder if vulkan is better cuz it's newer right

I'm assuming I added that too the engine ini too fix something

[twin dew]

Vulkan is marked as Experimental for a reason.
DX12 is the only fully supported renderer, that Vulkan is just autocompile in UE set to enabled, not maintained.

[twin dew]

You have -dx12 launch argument.

[fickle gust]

In the steam options I don't think I do I swear I got rid of it

[fickle gust]

Didn't realize vulkan was experimental I just know it better than dx12

[twin dew]

No, it is just OpenGL version of DX12, sidegrade.

OpenGL vs DX11.
Vulkan vs DX12.

[fickle gust]

I see

[twin dew]

Depends on specific GPU driver and specific game which is "better".

[fickle gust]

And it's better than dx11 atleast that's what star citizen switched too

[fickle gust]

I see I'll test it out

[twin dew]

With Satisfactory, almost never better, usually not worth even testing.

DX11 is the last DirectX version with lot of CPU overhead from being "easy to use" as it abstracted lot away.
Vulkan and DX12 are both based on AMD Mantle API, which moved lot of the responsibility from OS graphics API layer to developers, with much lower level access to the HW.

Just that DX12 is MS API.
Vulkan is Khronos Group API.

Based on same basic principles from that AMD Mantle base, but DirectX 12 as continuation of that MS DirectX API line.
And Vulkan as continuation of OpenGL API line.

[fickle gust]

I ain't gonna lie I don't really understand most of this I mean I kinda do but not really but cool

Was curious about it does explain it a little tho

I just know sc switched too it cuz of dx11 restrictions

[twin dew]

Yeah, but they could have switched to DX12 and not Vulkan.
But Vulkan runs native in both Windows and Linux.
And for long time, AMD Windows drivers were faster with Vulkan than DX12, to level that if you had CPU headroom, using VLK3D in Windows to convert games DX12 output to Vulkan, allowed you to get more FPS in many games with AMD GPUs.
Same thing wasn't true with Nvidia.

And might not be true anymore, or with newer AMD GPUs, only older.

Vulkan isn't better or worse than DX12, just different.
It came later, as getting the spec development through multi-company Chronos Group took longer than Microsoft alone doing DX12.

[fickle gust]

I don't necessarily see it as better but as it has tools dx11 doesn't idk about dx12

But I see

I joined in during 6900xtxh that was my first GPU

So the whole dx9 and vulkan etc was confusing

[twin dew]

DX9, DX11, OpenGL versions are high-level graphics APIs.
That abstract lot of stuff away from the devs.

DX12 and Vulkan are low-level graphics APIs, where developer is in full control, including ability to shoot themselves in the foot.

And for most part DX12 and Vulkan are very close, from that common heritage in AMD Mantle API.

In capabilities.

[fickle gust]

I see this made sense too me ty

That makes a lot of sense then as sc needs a lot of control

Especially for when they actually optimize it

[twin dew]

And that abstraction on higher level graphics APIs uses significant amount of CPU.

[fickle gust]

That also makes sense then sc has too be the most demanding cpu game imo

I mean hell I'm on a 12700k and use 100% of my cpu

[fickle gust]

<@&387163995947270144>

Idk who takes care of this but

[white kraken]

I was too slow 😉

[fickle gust]

Indeed ty it's all good

I'm just glad I pinged the right role

[white kraken]

@fallen temple is the speedy Mod 😉

[fallen temple]

zoom zoom

[fickle gust]

Noted for future use

[white kraken]

Nono, please don‘t individual Mods, always the Mod role 🙂

[cold tide]

where is the amd_fidelityfx_upscaler_dx12.dll file located for Satisfactory?

[twin dew]

Probably doesn't exist as such.
Satisfactory uses the source code plugin version, not precompiled.

[cold tide]

mehh

so you cant drop-in upgrade

optiscaler team released 4.0.2b

so now works on RDNA2

[twin dew]

The FSR3 files the game uses are under Satisfactory\FactoryGame\Plugins\FSR3\Binaries\Win64

[cold tide]

thats too bad i guess

just dropped it in Horizon Forbidden West and it works, crazy that a bunch of modders do what AMD doesnt.

[verbal raft]

its SDK V1 version is in Satisfactory\Engine\Binaries\Win64

with the name amd_fidelityfx_dx12

its 3.1.0 by default
in any case though i recomend installing optiscaler

[verbal raft]

thing is its gonna be mostly broken in this game anyway since the upscaling ratios are totally diferent to the normal ones
which breaks FSR 4 int8 even the b version

though the command r.screenpercentage set to 66.6 for quality mode, 58.8 for balanced or 50 for performance mode should fix it

[twin dew]

https://www.xda-developers.com/wine-11-rewrites-linux-runs-windows-games-speed-gains/

Been in Proton/GE-Proton in some form for a while already.

Ok, only GE-Proton, not mainline Proton.

[cold tide]

oh boy, let me get right on that

[ripe garden]

Sarcasm I hope?

Your so right

Holl up lemme mention da big yahu

@twin dew should I switch to Linux? I mainly play Roblox and steam games

Do I get better performance than windows with it?

[cold tide]

no way man, that inspect element scam looked so legit, i totally went for it

[dire igloo]

In Destiny 2 definitely not

[cold tide]

AMD's is a lot more consistent though

[pure karma]

Roblox isent supported on linux

[dire igloo]

In the big 2026, Nvidia has good drivers, let alone good Linux ones?
That's a new one for sure

Too much fat is unhealthy

[pure karma]

Arc is down like 30% yea sadly

[cold tide]

i just dont see the reason/appeal

no switching to linux

i mean its not lazyness, if it could genuinely do stuff better, stuff id want to do, i would teach myself but i can get windows to do whatever i need a computer to do

so...yeah

and anyone bringing up the spyware angle of windows

that is baked into the hardware, it really doesnt matter what os youre using

yeah, isps too

[wanton orchid]

@twin dew do you think latest x570d4u-2l2t bmc/bios fixes the x550-2l chips single port update bug ?

[sharp matrix]

In theory nvidia has drivers that arent total garbage, you just have to find them

Basic like factory drivers

Dont you dare blame my delta flyers!

https://tenor.com/view/star-trek-voyager-race-warp10-impulse-thrusters-warp10-delta-flyer-gif-26938307

[cold tide]

i dont like how they did my boy dukat dirty in the later seasons

but its still enjoyable

its funny how they did away with kes beyond season3 to introduce a bombshell, but she turned out to be the most likeable character lol.

fun will now commence

he was a bit of a coomer in tng, but they redeemed him

[pure karma]

same and if they hold up to there recent statements for as long as its activatable for free il probably stay on windows but linux is fine i dont really have beef with it much outside of its dependecy on using the terminal

[cold tide]

yeah not countersignalling, hence saying it doesnt appeal to me personally.

and using rufus to make the bootable drive, making an offline account is super easy

imagine more people knew

and didnt throw out their pcs because of muh tpm2.0 requirement

genuinely made me lose faith will all tech reviewers

they had one chance to stick it to the greedy hardware vendors they rail on daily

and they let it go

[pure karma]

same

seing so many people not know rufus exists was such a realsiation that no one gives a shit

[languid gulch]

delta flyer #2,768

more than i care to admit 🤣

[cold tide]

Yeah impotently ranting on screen, but then when oppertunities present themselves be utterly ignorant like the rest.

Even in the comments you could see people saying "dude.. rufus?"

but they were ignored.

the windows 10 thing, when people were throwing away computers

their (GN) 150iq take was to put linux on it

like normies would be able to use that

low impulse control is hardly amazing

no offense

nope

the US almost spoke my language, never forget that

https://tenor.com/view/gekoloniseerd-voc-nederland-kolonisatie-verenigde-oostindische-compagnie-gif-13241235

https://tenor.com/view/star-trek-patrick-stewart-captain-jean-luc-picard-face-palm-disappointed-gif-4780258

[jagged snow]

This is pretty incredible
https://youtu.be/9kae-UAME1U?si=MozRqvc_VSM2HsbR

[cold tide]

it seemed to fuck up at certain points but pretty cool nontheless

[pure karma]

thats cool that it dosent seem to have too many issues balencing

[languid gulch]

i see that microslop is back to spamming Defender notices about onedrive

[twin dew]

Cannot, as that is because of the physical implementation on MB PCB not being same as with the Intels own PCB.
Certain pin is passively pulled wrong way and makes the controller think it is in debug-mode, and allows firmware update to flash regions that shouldn't be updated normally.

Someone on the PCB design team at AsRock Rack fucked up and didn't follow the Intel design spec for the chip.
Same issue with at least that another same timeframe MB, possibly others too.

So that 3.15 firmware is the latest that was publicly available.
And not even freely available on the site anymore.

But checked and there are no new features from 3.15 to latest 3.70 based on Intel changelog feature matrix.
Intel doesn't tell what has changed.

Security level identifier was bumped to prevent downgrades, so there has been some kind of security fix at least, and then other things.

Not that that downgrade thing works, people have downgraded from broken 3.60 to that AsRock Rack 3.15 with edited .cfg to make the update utility think that that 3.15 is newer than installed 3.60.

Probably because of that pin misconfig allowing also downgrades without checks.

[tall forum]

new profile?

[wanton orchid]

ok thank you
I wasn't expecting the whole story
tldr only 3.15 customized will ever work
noted

[twin dew]

Unless AsRock Rack gives out another version, or someone bothers to do suitable mashup of that AsRock 3.15 and Intel 3.70 or like, if the submodule memory regions haven't changed.

Which they shouldn't have, as normally those are not programmable.

I would need to see about that, by downloading the Intel 3.15 and comparing the specific firmware file to that AsRock Rack 3.15.
They too the Intel set and just replaced one specific firmware file with their own, didn't even remove the "useless" files that are for different cards etc.

[night girder]

Trying to figure out how much space it takes to record in stereo 48KHz and 32-bit float... so I know what SD to buy 😅

Should be enough.

[twin dew]

48kHz, so 48000 samples per second, each sample taking 32 bits, so 1.5Mb per second per channel.

[night girder]

That would be 10,800Mb per hour?

[twin dew]

Mistake in there, Mb, I had written MB.
So 8 times less.

[night girder]

right ... 😄

[twin dew]

11 520 000 B per minute.
So about 11MB per minute per channel.

No per minute.

Brain not braining.

[night girder]

Huh, can't find model number for lexar 🤣 Found it.

32GB microSDHC (TF) C10.

not being sold since 2016 =_=

fuck it sanddisk it is.

[twin dew]

But 48 000 Hz * 32b = 1536000 b/s * ( 1B / 8b ) = 192 000 B/s (per channel) (192 kB/s or 187.5 KiB/s)
And you can just input any other sample rate and bit depth.

[night girder]

Sanddisk expensive.

Ah, was looking in the wrong manual. Of the essential. While I have H1n. And there is also H1.

No wondere I got all these weird old ass SDs 😛

Anyone used Transcend SDs in here?

Whaaat?

19 to 175.99 euro.

[wanton orchid]

tbh, I've seen all of that years ago already
and I didn't put time nor energy to disassemble it yet
because I don't have experience, tools, and priority to do so unfortunately
I'm really thankful that you are willing to put effort in digging in these things
I can't really support you much apart from talking here, even though I'd wish I could much much more (not specifically for you, but for a lot of people)
my energy budget yet reduced again lately
light is shaking
at least I can tell you it's seen

it's like I'm aging at high rate
can't capitalize on my recovery capabilities because I have no much to recover with

[twin dew]

Doing a check right now with hex editor.
But seems that some of the code blocks are common, but are at different offsets between Intel 3.15 and AsRock Rack 3.15.
But need to check those more later, now just mapping sections that are unused in Intel file but used in AsRock file as offsets.

[wanton orchid]

what I suspected is they locked the debug path in the firmware
if it can be stably identified
it would basically be a one or two byte swap, that would be handy

[twin dew]

Lot of other things related to using one of the 10G ports as IPMI port, that the default will not have.

[wanton orchid]

step 1 get generic 3.15 and compare with the custom one

[twin dew]

Yeah, doing that right now.
Also got Intel 3.10 and Intel 3.50 to compare if the sections that AsRock Rack has different are same between Intel versions.

[wanton orchid]

what scared me from trying again is I lost the exact notes on ID rewriting for downgrade
so I will not have the pass for patcher to trigger rollback

I may find it again if I succeed in recovering data volumes a nasty corrupted kernel bio path wiped 2 months ago

with a suspiciously old patch that didn't get merged

every step, mountains grows in front of me

[twin dew]

In the Asrock set nvmupdate.cfg, under header dSGVL_B0_AT2_NCSI, with NVM IMAGE: ASRR_X550_0_0_002_0.bin
Add into the REPLACES: section the newer versions ID.

That is in the new updaters same file, same section, as EEPID,

And run as usual.

[wanton orchid]

I just can't remember the ID
because that's not how my memory works
I have noted it somewhere, but may have lost that media

[twin dew]

Point was that you get it from whatever you used to install, first screenshot was supposed to be from Intel 3.70 file and not Intel 3.15 file:

Note how that REPLACES section contains 3.15 EEPID on the last line. and then couple of even newer firmware versions.
And the that sets EEPID couple of lines earlier.

[wanton orchid]

you mean it's 1:1 ?
it's always in the new replace ??
or you added these ids manually ?

[twin dew]

So if you updated to that 3.70, and stuff broke, you would open that releases config file, find that EEPID: 8000184D line, copy that 8000184D to AsRock Rack 3.15 updaters config at the end of that last REPLACES line.

So the updater thinks that that 3.15 is newer than that 3.70 that is installed, as the detection works on those EEPID at that stage.

And will happily overwrite the current EEPROM contents with that 3.15, as the current EEPROM contents EEPID is marked as allowed to be replaced with that AsRock Rack 3.15 EEPID.

[wanton orchid]

oooh it's version IDs ..
so it's an apply list
only have to put the newer (wrong) id on old list and it updates happily ?

So ids are stable

[twin dew]

Yes.

[wanton orchid]

I overcomplicated it in my head sorry

So whatever new version get it
I put on the old "patch this" list

[twin dew]

So all the Intel released firmware versions for that specific dSGVL_B0_AT2_NCSI are 80000526 8000059F 80000603 8000069C 800007F6 8000049A 800005D0 800005DE 800005EE 800005F8 80000634 80000651 80000688 8000069D 800007B5 8000084F 80000948 800009BA 800009E7 800009F9 80000A29 80000A73 80000AEE 80000D03 80000E3E 80000D37 80000D78 80000F49 80001112 80000F98 80000F9B 80000F9C 80001050 80001138 80001180 800011FA 8000134A (3.10) 80001373 (3.15) 80001578 800016DC (3.50) 8000172D (3.60) and 8000184D (3.70)

And then for the same chip, there are other firmware lines for different feature sets on the actual card.

Like dSGVL_B0_AT2_MCTP that has different set of firmware EEPIDs.

[wanton orchid]

I really thought new id was tied to old-new version pair
not only to new version

So it's really a "dumb" firmware flash
firmware defined device

the whole thing is refreshed on firmware update

[twin dew]

Yes, just checks if the card has right vendor and device ID, and then matches based on previous firmware versions EEPID to see which firmware file for that chip to flash.

As same chip can be used in multiple different configurations, with different firmware, but stame chip IDs.

AsRock Rack should have changed those chip IDs, but they didn't.
Most other Intel network chip users do on their MBs.

Which wouldn't have allowed to even try flashing the "wrong" firmware onto the MBs without editing that file.

[wanton orchid]

yep, better opt in mistakes than opt it out

[twin dew]

Changed that EEPID for their own custom firmware versions I mean, not the chip IDs.

As the chip IDs are also used to select drivers etc.

[wanton orchid]

I do think having to override debug path have security pros
but only if one can update to latest security patches

[twin dew]

Point is that they are mistakenly ALLOWING debug mode all the time.

AsRock Rack, normally it is closed out on retail products, unless you mod the PCB or like.

[wanton orchid]

what a mess

[twin dew]

In these few PCBs, the leg is held passively in state that ENABLES debug mode on chip boot.

[wanton orchid]

I meant once you patched the firmware
the debug mode can't even be physically enabled

[night girder]

So I know that Sandisk has this marketing stuff for microSDs. Ultra. Extreme.

But the fuckers also have Ultra PRO. Extreme PRO.

Infuriating marketing bullshit.

[wanton orchid]

it basically is same but more expensive with specific features on
which can also be less well supported
it's less fool proof
basically got an ssd which would crash my system

[twin dew]

But someone had updated to Intel 3.60 with that common failure, and then downgraded to AsRock Rack 3.15.
But IIRC hadn't ever installed that AsRock Rack 3.15 before 3.60.

[wanton orchid]

now PRO means "I have actually checked it would work out of the factory" it seems

[twin dew]

Too large differences from those code blocks moving around that I would be doing anything.
As would need to then see about possible jumps between the AsRock Rack code and the Intel code, and fix any offsets to match the possible frankenfirmware.

And cannot even try updating that Intel 3.70 in until next fall at earliers myself.
But very unlikely to work, because of that BMC/IPMI shared LAN port functionality on that first X550 10G port.

Which is probably the one that stops working when updating to Intel firmware.
And not the "standard" second port.

Why AsRock Rack shared that 10G port with BMC is just weird.
And they corrected that idiocy with the Broadcom 10G version of the board, sharing one of the 1G ports, like on the 1G only version of the board.

[wanton orchid]

I fully agree with this, never really understood why, especially since it does not even works well with the bmc they had baked in

[twin dew]

Cannot keep up with 10G

[wanton orchid]

here I see 1.79B is improving memory performance significantly, because of better mem buffer and secondary timing/syncing than before probably

[twin dew]

Or if testing in OS, various OS side vulnerability patches might be turned off as unneeded, allowing for better performance.

With the newer AGESA with fixes incorporated.

[wanton orchid]

I think it's both

I still see common security issues as mitigated

but maybe the mitigation list is shorter

[twin dew]

Ah, I reinstalled the main OS after updating the BIOS, so cannot check what the older boot logs say about applied mitigations.
But Spectres etc. cannot be fully fixed with firmware on AM4 CPUs.
Needed HW side changes

[wanton orchid]

yep

but iirc am4 software mitigations have less overhead than for Intel ones at the time

overall more consistent arch

maybe firmware mitigations are more optimized in latest microcodes too, may be that

[twin dew]

Will see if that BMC share slows the first 10G port down even when it is set as "not bonded, don't use at all" in BMC settings...

[waxen pendant]

How do Nvidia GPU users screen capture?

On my desktop i just record clips using the AMD control panel.

but my nvidia Laptop has no such feature

[night girder]

?

I think nvidia experience?

[twin dew]

Or Nvidia App as the new replacement in development.

[pure karma]

Xbox game bar?

[night girder]

Yeah, so software side of Nvidia.

Or

[pure karma]

Wtf else would i use

[night girder]

...

[waxen pendant]

thats a thing i guess? i only see it when i use a controller which is seldom

[pure karma]

Why would i get bloat to do something baked into the existing bloat

[night girder]

Can screen record too.

[waxen pendant]

i cant find any capture options in the nvidia experience or any other related software.

[waxen pendant]

that works!

[pure karma]

Windows + G then record last 30 seconds or whatever duration you set it to

[waxen pendant]

i just need a ten second clip lol

[waxen pendant]

thanks folks appreciate the input

[pure karma]

Its not the most reliable but it works

[waxen pendant]

I could download other programs but i dont wana so all this works

[twin dew]

And you might need to install full version then.
Like with AMD drivers having three different levels of install, where only the biggest comes with recording functionality etc., Nvidia also might have similar levels and the OEM didn't include ShadowPlay or whatever.

AMD has drivers only, no Radeon Software at all.
Minimal Radeon Software with just "normal" features.
And full Radeon Software with capture, OC features etc.

[pure karma]

Nvidia has that one clip feature but idk how it works and if i recall its automatic

[waxen pendant]

whatever Version i have installed came broken when this Laptop was bought new. I had to completely remove all GPU drivers and reinstall from scratch. since then The Nvidia software completely reinstalls itself whenever there is a update. its a odd experience.

in anycase these options work and i thank you

[night girder]

Thinking to start testing out Librewolf.
I have firefox updates blocked. So no AI yet. But I don't want to download AI or features that I am not going to use.
It's just all bloat. But I can't keep an old firefox build either for security reasons.

Other options are Brave and DuckDuckGo browser. Both that I used and wasn't fan off.

Oh there is also mullvad browser, the hell?

[waxen pendant]

I currently have mint (still) seems to be a nice balance of power user capabilities and ease of use

although its jsut side loaded im still mostly using Windows.

[night girder]

mullvad.net/en/browser

That's why we partnered with the Tor Project to develop Mullvad Browser – a browser designed to minimize tracking and fingerprints.

This is getting interesting ...

You still use it? Also, you tested browser?

[twin dew]

Yes, as VPN when needed.

[night girder]

Ah, well. They have browser now 😄

[twin dew]

For browser, no.

[night girder]

Which they claim to be open source.

without a github link ... 😫

[twin dew]

https://github.com/mullvad/mullvad-browser

Their VPN client is under same user.

[night girder]

yeah yeah I found it.

But I shouldn't have to google it darn it.

If you say "we are open source" LINK to project, why hide it makes no sense.

One of the purposes of "we are open source" is so users can check code and get more confident in product.

Also no code.

changelog=$(grep -B 1000 -m1 '^$' ./tor-browser-build/ChangeLog-MB.txt | sed -e '1d' -e 's/^\s** Build System\s*$/## Build System/')

Ok, so it seems to be tor-browser.

[twin dew]

They had general "Open Source" link at the bottom of all pages to general page, with link to that GitHub:
https://mullvad.net/en/open-source

And the Github page says to switch branch to to get source.

[night girder]

url="://dist.torproject.org/mullvadbrowser/$VERSION"
update_response_commit=$(curl '//gitlab.torproject.org/tpo/applications/mullvad-browser-update-responses/-/refs/main/logs_tree/?format=json&offset=0' 2>/dev/null | jq -r '.[0] | .commit .id')

[night girder]

Oh no.

chrome 😒

[twin dew]

What?
No, Firefox based.

[night girder]

Right ...

[twin dew]

Last touched 2 years ago for that subdirectory.

[night girder]

So?

Don't touch anything that doesn't need fixing or updating or refactoring.

But I think it's scripts to import from google chrome to browser. Or something like that. So for users to switch from browser and important things from google chrome to Mullvad browser.

Also the whole project isn't updated for what seems a year or two.

the branches itself are active.

[twin dew]

That dropdown shows old versions, highest number is mullvad-browser-148.0a1-16.0-2 and latest updated is mullvad-browser-140.9.0esr-15.0-1

[night girder]

So Mullvad seems to use Tor.

No clue what Tor uses underneath.

[twin dew]

Firefox.

[night girder]

Mullvad Browser is developed together with the Tor Project. You could say that Mullvad Browser is a Tor Browser without the Tor Network. Instead you can use the Mullvad Browser with (or without) a VPN.

So Mullvad is Tor Browser, without Onion networking.

[twin dew]

140.9.0esr for example is Firefox version, used as the base of that specific version.

[night girder]

I hope Tor nuked Firefox features though.

I am actually curious how this works. Like if Firefox updates, does Tor pull in all changes or do they cherry pick stuff from firefox over to their browser 🤔

[twin dew]

Forker rebases to newer base software version at points.
Like that Mullvad Browser 15.0.8 rebasing to be based on Firefox 140.9.0esr (esr = Extended Support Release)

[night girder]

Tor Browser 15.0a4, which is the latest alpha release, is available now, and one of the biggest changes is that it removes various AI features that Mozilla has been integrating into Firefox. This is a great change from the trend of AI chatbots and machine learning being baked into your browser.

This is what I was looking for.

[twin dew]

And that current FireFox ESR doesn't yet contain AI features at all.
Which both TOR and MullVad Browser Stables are based on.

ESR only gets feature updates once a year, only security updates between those releases.

[night girder]

what?

[twin dew]

https://support.mozilla.org/en-US/kb/choosing-firefox-update-channel

[night girder]

www.firefox.com/en-US/firefox/140.0esr/releasenotes/

You can now enable the updated Firefox sidebar in Settings > General > Browser Layout to quickly access multiple tools in one click, without leaving your main view. Sidebar tools include an AI chatbot of your choice, bookmarks, history, and tabs from devices you sync with your Mozilla account.

[twin dew]

Ah, ok.

My bad.

[night girder]

Mhm. I have a feeling I am overthinking it.

AMD Chat just made me realize, AI takes up diskspace.

So if Firefox adds AI, how much extra diskspace does it take? 🤔 And if it's a few GBs (which I highly doubt it is) and you not use any of the features, isn't that kinda silly?

Firefox should have done what AMD does. Give option to download AI tools for those that want.

For once, I have to say. Good job AMD.

Anyone here have firefox 148? Can you check diskspace?

So interesting, someone said to me their 148 version is about 330MB.

18MB extra for AI features 🤔

[waxen pendant]

all AI features are turned off. no clue if that effects anything

[twin dew]

No, as the code is still installed, just not actively used.

[night girder]

So either it's very lightweight. It was already included in previous builds.

Or it's parked somewhere else on the disk and not calculated in total size of firefox.

Because the size is only for ""C:\Program Files\Mozilla Firefox""

but I do find other references too. "C:\Program Files (x86)\Mozilla Maintenance Service". This one is not included in total size I believe.

[twin dew]

Which is the (auto)updater, so that you can update that main install without admin rights.

[night girder]

Interesting.

I still find it weird that the best way I found to block firefox updates.

C:\Program Files\Mozilla Firefox\distribution -> policies.json.

{
"policies": {
"DisableAppUpdate": true
}
}

The other methods kept nagging me, or updating firefox 🤷‍♂️

[night girder]

YT it crapping out on my end again.

Can't load the first row of videos, the thumbnails. Loads everything else perfect. Even after page refresh.

[waxen pendant]

This happens to me all the time on windows

[mental oriole]

I had that ysterday

[night girder]

maybe the first three videos are ads and get blocked 🙈

[twin dew]

No. They have CG-NAT that isn't in some anti-spam DB, so it gets mistakenly detected as proxy.

[twin dew]

Proxy needs active configuration from the user.
And with HTTPS, useless without MITM, which then also needs custom CA certificate on each users computer.

And for ISP, CG-NAT today is common, more common newer that ISP is.
HTTP(S) proxies are almost unheard of today.

Except in specific corporate environments, where they mostly exist for that HTTP(S) MITM checking, not for caching anymore.

But to anyone outside looking at the traffic cannot differentiate between those.
But most ISPs have proactively reported their CG-NAT endpoint IPs to those various protection block list providers.

Pilsfree would need to do that too.

I could have 1Gbit/1Gbit/s connection, but I'm not willing to pay 41e/month for it.
Used to have 50/50Mbit/s, but that was discontinued and I'm now on more expensive 100/100Mbit/s one.

As the steps are 10/10, 100/100, 200/200, 500/500 and 1000/1000.

The prices for internet access or mobile phone access aren't really set by costs, but what the market will bear.

That 10/10 and 1000/1000 have basically the same cost to the ISP.

Unfortunately all the entrenched ISPs with infrastructure in place could just lower their prices closer to their actual costs, kill the new competitor who will have higher costs because of infrastructure buildup costs/loan repayments, and then raise prices again.

Telia can charge almost anything, as they are the only company with fiber to the building, and possibly for the whole area.
Anyone else would need to build a line on their own dime.

ISPs got that killed to a level where any level of public funds aren't allowed if the area is already server by anyone.

To keep their monopolies in areas they already served.

On EU level.

So (partly) public funded fiber project cannot build in areas with existing ISPs.
And any so build networks need to be open to all comers at cost for rent the line.

Regulatory capture at its finest.

The original regulation only required that second point.
So that public funds could be used to build alternative common fiber networks even in cities etc. to compete with existing ISP networks.
But lobbying killed that, and limited to only unserved areas, to level that if ISP is planning to build a fiber network on their own dime in same area, any public funded projects are dead.

Cannot have actual competition...

[cold tide]

if you dont own anything, its not piracy

[twin dew]

I would like to have, but wouldn't be right for the cat.

[night girder]

No more odd question 🤣 - message got deleted 🙁

[jagged snow]

Indeed 🤣

I couldn't figure out how to word it without it sounding like a whiny reddit post

[night girder]

Were those resolutions of iPads too?

[jagged snow]

Iphone, iPad, and Android

[night girder]

Ah, makes sense 😄

[jagged snow]

Working on some promo images for an app I'm contracted on

Designer keeps sending the wrong image dimensions

As far as the sizes, I created the files in illustrator to match your dimensions. They are saved with 300 DPI resolution. Is it because my resolution is too high that it’s making them the wrong size?

[night girder]

www.photopea.com (free, photoshop like editor in browser).
Maybe you can fix it yourself?

[jagged snow]

Oh I have an illustrator license, I just asked for the raw files to re-export

Was just super odd to me

[night girder]

Smart smart.

[jagged snow]

And not my job

I specifically told them I don't do design 💀

[night girder]

I understand. But sometimes, it's better to do yourself. Then going back and forth for a few days.

[jagged snow]

Yeah, for sure

[night girder]

Because, sometimes, corporation communication is slow AF (mails etc).

[jagged snow]

These were the dimensions they sent

[jagged snow]

Thankfully I'm directly in touch with the designer so I don't have to go through an intermediary

[night girder]

Haha, that's funny. Because the name. Clearly. They tried. But the actually resolution is not what the name suggests it should be 🤣

[jagged snow]

Yep 💀

The name has the dimensions I asked for

I also needed/wanted jpg rather than png because apple won't accept any images with an alpha channel/transparency

So I'm just gonna do it myself when I get back home

[night girder]

Did you let the Illustrator/Designer person know?

[jagged snow]

Yeah, I did
There was also a few design issues I needed them to work out too

[night girder]

Maybe something is wrong @ their end and they don't know.
Or they can't learn from this, if they don't know. Can be they are overloaded with work. Not paying attention etc.

[jagged snow]

My understanding is they're a freelancer that's just getting started in this portion of the industry

Seems like they've mostly done graphic design for print media and brand campaigns until now

[night girder]

Ah, so learning 😊

[visual tree]

People are now using AI to "enhance" their home when trying to sell a property 🤣

https://www.reddit.com/r/realtors/s/Gbr5LCnc79

Do you seriously believe someone is going to buy a property without actually visiting it?

If I found a listing with nice property and decent price, I would visit it first to make sure the images are not fake

Although I could also use online tools to detect edited images....

Top image is AI

[mental oriole]

It even made the room bigger wtf

[pure karma]

yea thats been a thing for a while

[sharp matrix]

oh god now the ai slop property listings have arrived to have they

[sharp matrix]

yeah there is alot thats really bad about this, making the room bigger, leading to even more misleading listings is one of them....

[visual tree]

I really hate emergency alarm sounds on phone, they sound like alarms from doomsday movies

Woke up to this today

Also, I don't understand the point of emergency alarms if you send it too late

People received it an hour late after it was declared there would be no school for children

[languid gulch]

https://www.youtube.com/watch?v=xFgBgtRqJS4

[night girder]

@twin dew Do you use the Mullvad client?

[twin dew]

Yes.

[night girder]

Cool!

I did use proton VPN (free) for a bit, but the client wasn't very stable.

It crashed randomly (and quietly), so sometimes I noticed the VPN wasn't running because the icon dissapeared from taskbar.

[twin dew]

Or can use stock WireGuard or OpenVPN clients, and not Mullvads own.

[pure karma]

Some of them are so unserious i love it

ours is just annoying especially at night honestly

[night girder]

Yeah, that's what I am pondering about.

[twin dew]

But that MullVads own client has been stable, and is very usable.

Doing it with stock clients needs some extra work.

And they have done lot of extra work to make stuff as-leak-proof as possible, including stuff like Kill Switch (if their client stops without actual quitting, net will not work) and Lockdown mode (net will not work without the client, at all, ever).

And other optional stuff:

[night girder]

So one "locks down" the VPN tunnel, no VPN no internet connection. And the other one prevents of leaks when that connection is lost. Cool stuff.

[twin dew]

Both prevent data leaks if tunnel is lost to the client not shutting down properly.
But the latter prevents it always when client isn't running.

So if the client crashes or like, that first protects from data leaks.
But not if you quit the client, and not kill it.

[twin dew]

Basically in this case, with that Kill Switch on, net would have stopped working.

[night girder]

Have you used DNS content blockers?

[twin dew]

No.

Same idea as Pi-hole etc.
Just with their DNS servers in control.

[night girder]

Wonder how well it works.

[twin dew]

What is used etc:
https://github.com/mullvad/dns-blocklists

[night girder]

Cheers.

[waxen pendant]

Nvidia why TF is my driver update 1gb

makes me wonder what kind of bloat is being forced on me now

[mental oriole]

They're adding AI...

[waxen pendant]

likely to bad i have a 3060 mobile lol

[sharp matrix]

1gb of ai bloatware

[pure karma]

forget the size of the driver maybe they should start by fixing whatever storage medium drivers are stored on so my drivers stop corupting

[night girder]

Haha 🤣 Told you AMD was doing it "better". They don't dump it all on your drive, they ask you if they can dump AI on the drive.

AMD probably didn't invest that much into AI as Nvidia.

So they don't care if people download it, less agressive tactics/marketing.

[pure karma]

they dident dump as much because there busy lobbying and loosing

[waxen pendant]

All publicly traded companies have a legal mandate to be evil

[pure karma]

yea like the fact that loosing money for your investors is a crime which is realistically the #1 cause behind a lot of enshittification

[cold tide]

Notincing a lot more traversal stuttering after switching to the experimental build. Is that common?

[jagged snow]

Iserve

Apple internal use only apparently

[twin dew]

Wonder what that being named as BMC actually is.
Because no normal BMC is 8 core with 16GB of RAM and 1TB of storage...
AST2600 is 3-core ARM with basic 2D graphics. And supports max 1GB of DDR4 and firmware SPI flash chip seems to be around 128MB range for latest MBs.

With SuperMicro currently using about 64MB of that, based on the compression ratio.

For their Intel Xeon 6xxx Granite Rapids MBs.

[sharp matrix]

Its odd this BMC is with memory and storage here

[twin dew]

Yeah, seems to be another complete system, not BMC.

[sharp matrix]

I dont understand this usage hmmm

What even is 1tb(bmc) hmmm

[twin dew]

BMC = Baseboard Management Controller.
But in this case, based on the specs there is one full system with 8-core CPU, 16GB of RAM and 1TB of storage as "BMC", and then the "actual" system with 4 CPUs etc.

Basically normally server specs specify that they have IPMI/BMC, and possibly which controller it uses.
But it is just small specialized controller with smallish upgradable firmware on SPI FLASH chip, like BIOS/UEFI for the main system.
Not full on system with full generic OS like that spec suggests.

[night girder]

Aslo read Apple stopped Mac Pro production and no plans for new models.

[bronze jasper]

So last week, i figured out that windows had been constantly powering down the usb dongle for my wireless headphones. Kraken pros. Anytime there's no audio for a second, it powers off then powers back on when theres audio again. You wouldn't even know it was happening.

but it was causing some interesting crashes that i knew where audio stack related, because the whole pc would freeze up except audio would keep playing which is kinda sus. so i looked into it and fixed it. but that's just regular troubleshooting and discovering the problem. no big deal. Razor should've made it so the dongle doesn't do this with windows power management shenanigans.

Anyways, we're a week later and i'm realising that my headphone's battery life is way longer now. Where before, it would last 2-3 hours and i'd have to charge them again , but at least they operate while charging. Now though, it's lasting all day! ALL DAY AUDIO! i haven't charged it since i got home and started using them last night, and i'm still using them like 6 hours of wireless time.

my guess is that the dongle powering on and off constantly was killing the battery of the headphones. more bad design by razor since that shouldn't happen.

[pure karma]

The bad design was you thinking anything made by razor was good

[bronze jasper]

random insult out of no where. no wonder you're blocked.

[dire igloo]

Razer does have a few good products, but those are mice

[bronze jasper]

Yeah I wouldn't even seek out one of their mice. Dude doesn't even know how i got these headphones in the first place. He just has it out for me personally.

[civic gulch]

🫃

[pure karma]

you bought a razer product you had it coming

they cant be taken seriously when you need 4 pieces of software and have to pay a 200$ upcharge for a dongle to have fully functional headphones from them

[jagged snow]

PS5 is getting a price hike(100-150usd). I believe this is the second time they've done that. Kinda wild that we're seeing significant price increases on consoles years after release.

[pure karma]

isent it the 4th?

[jagged snow]

Oh it very well could be

I'm honestly surprised that it's still selling th is well too with that going on

[pure karma]

i know its atleast 3 im not fully confident on 4 tho

[pure karma]

well if you think about it no one wants the switch 2 cause they hate nintendo, xbox cemented themselfs as garbage for the foreseable future and steam cant stock anything

[bronze jasper]

the switch 2 is the fastest selling console of all time. i doubt they've got to market saturation with it yet either.

your hate for nintendo might be biasing you. unsurprising really.

[pure karma]

i literally own one

[bronze jasper]

yup. standard nintendo hater.

[visual tree]

We use the UK version here which sounds like this: https://www.youtube.com/watch?v=2N_pDWXEBCU

[pure karma]

thats somehow worst than ours in terms of being more annoying than useful https://www.youtube.com/watch?v=BPkApDupM-k

[visual tree]

At least we don't use the australian version which would wake up the dead from their grave lol https://www.youtube.com/watch?v=Y-AnfWOfODk

[pure karma]

honestly i think the astralian one is kinda cute at first and then it turns to hell

[visual tree]

Bet they play it every single time a random animal is trying to kill you in Australia

Oh no, it's daylight saving time change this week

[tall forum]

whats this?

[visual tree]

Emergency alarm sound

[velvet jasper]

Sometimes when I'm playing satisfactory white text pops up saying:
"Your PC has run into a problem"
Then my computer crashes.
Idk what it could be, pls help

[pure karma]

well step 1 would be finding out whats giving the message

if its windows ods are a good place to get a general idea of where to look is event viewer

[glossy glacier]

at the bottom of the screen there should be a stop code given, can you tell us which one it is?

[velvet jasper]

Sadly I have no idea

I'll check next time my pc crashes

Also it's not like a normal crash, it just shows the message and restarts

No bluescreen or anything

[glossy glacier]

you mean this screen?
that is a crash

[pure karma]

yea that is the new bluescreen

[glossy glacier]

BlueBlack Screen of Death

[velvet jasper]

No, just white text on frozen satisfactory and then instant restart

[pure karma]

so its probably exactly that but its so fucked the bluescreen isent working properly

i had this before were it would be my desktop instead of black/blue cause it couldent load properly

[velvet jasper]

I'll try to take photo of it

Next time it happens

[languid gulch]

yup, had that twice this week

[slow flint]

Just had a annoying linux issue
Booting into my system after updating last night and I have NO internet
No WIFI or Eth, only Loopback shows on ip link show

No Eth driver loaded or installed its just gone
Ended up realizing that my system tried updating to kernal 6.19.9-arch1-1 from 6.19.6-arch1-1 but it moved all modules to the 16.9 folder but then failed to actually install 19.9 and stayed on 19.6
So when it loaded into kernal 19.6 there was just no directory for 19.6 modules

Booted into Arch installer and reinstalled the linux kernal and we're back to normal

Now I just have to figure out why arch just wont boot while my Eth is connected

[twin dew]

Have you let it try to boot for 5+ minutes?
Whatever is causing the stop in booting might time out and let it continue.

So you can see what was the problem.
Or looking into previous boot with journalctl -b -1 might let you see where it got stuck.

With network interface being part of the problem, I would expect that something is pulling in systemd-networkd-wait-online.service or systemd-udev-settle.service, and one of those gets stuck waiting for something, until timeout is triggered.

For first, the cause is then .network file that misconfigures what is considered "network is up".
For second, just masking the unit is usually the right option, but would need to see what is pulling it into use, it isn't usually used anymore, because of causing major issues to boot time.

For services that cannot start without some piece of HW being available and configured, they should depend on the exact HW, not that generic legacy settle.

[slow flint]

Its been a while since I have but I did leave it for like 30 mins once
Monitor just stops receiving a signal and nothing ever comes through

[slow flint]

True I should give that a go
Need to learn more about linux troubleshooting

[twin dew]

If the monitor goes to sleep, then waiting will probably not help, but those previous boot logs can.

[night girder]

fucking AMD

Was playing a game that was always stable 60 FPS. DIdn't play it this week.

So I updatred adrenalin.

Then today playing the game and I got 40 FPS with a high frame time response.
I disable that ML-based FSR bullshit on adrenalin and voila. Back to 60 FPS.

So if anyone notices loss in frames. try disabling it. For global:

[wanton orchid]

what, it is supposed to be active when game settings activate it

did you check your game graphic settings ?

[twin dew]

That is the generic that works somewhat without game support.