off-topic-tech

[pure karma]

yea seriously

[twin dew]

Ah, reboot day again.

[edgy hazel]

"Reboot day" smh just shut off your PC when ur done

[pure karma]

yea i don't understand why some people insist on leaving their computers on...

[dire igloo]

https://videocardz.com/newz/pocket-sized-foldable-keyboard-is-a-full-pc-equipped-with-amd-ryzen-7-8840u

[languid gulch]

ok now THAT is creative

would be even more interesting if they stuffed an ARM in there

[pure karma]

its unique but it probably wouldent pop off

because at the end of the day its practically just a laptop whit no screen

[dire igloo]

That fits into a pocket

Also tiny trackpad

[wanton orchid]

it's right in the idea, but poorly thought out

[dire igloo]

That sits in the middle of the typing area

[wanton orchid]

I was planning to do something along these lines for myself, but noticed it would be not commercially successful anyway

[pure karma]

its not just that i think it would probably also pull a rog ally and become uncomfortably warm to type on

[languid gulch]

now i'm thinking about having something like that keyboard pc thingy, but have a "phone", that can just be a regular phone if you want it to be, but also a lower power mode that you can wire in to the keyboard & have it be a monitor for it

[grim panther]

with HDMI you easily can any TV into a screen. I don't see a problem there

also there are great portable screens available nowadays. The combination would still be smaller than a regular Laptop

[night girder]

I would rather have a phone that can do that in the future.

Connect phone wireless to: 1. Keyboard. 2. Mouse. 3 Monitor.

[grim panther]

typing on a phone sucks.

[night girder]

#off-topic-tech message

[grim panther]

then you have to carry a keyboard anyway

[night girder]

Yeah, but still have a phone 😄

And you don't have to carry it. You just leave it at work.

[grim panther]

which you can't use during the time

[night girder]

Get at work, phone connects and voila you got PC.

[grim panther]

only disadvantages

[night girder]

Why would you want to use your phone 🤦‍♂️

When you got PC. You can still call with it.

My iPhone connects to my mac, I can receive calls on my Macbook.

Contacts, on my Macbook pro. Music. On my my Macbook. Like, there is NOTHING that my phone can do that my macbook can't do.

Only thing they need to solve is battery life, CPU and GPU power. But they are working towards that. Phones do get better.

[grim panther]

oh well, but Apple just sucks

[night girder]

Continuity Camera, just made it so that your iPhone is basically a webcam.

[grim panther]

closed system, does not use establisherd standards, just crap.

[night girder]

Still more innovating that Windows adding adds to start menu

Or: www.forbes.com/sites/quickerbettertech/2024/06/09/business-tech-roundup-expert-microsoft-windows-ai-recall-a-security-disaster/

Good going Windows 🤣

[grim panther]

nobody was talking about windows

[night girder]

I point out, Apple sucks. But better than Windows imo. Or Linux.

[grim panther]

not in my world

[night girder]

What else you going to get other than those 3?

[grim panther]

depends on the use case

but a closed ecosystem like Apple provides would never be my choice

[night girder]

Linux does have the problem of the peer review failing etc. Few examples:

arstechnica.com/security/2024/03/backdoor-found-in-widely-used-linux-utility-breaks-encrypted-ssh-connections/
www.theverge.com/2021/4/30/22410164/linux-kernel-university-of-minnesota-banned-open-source

[grim panther]

guess what? Apple does not have a peer review

[night girder]

Don't worry EC is working at it 😉

[grim panther]

even worse

[night girder]

I know of almost no big companies, who use Linux as their main operating system for development and getting work done.

For home use, sure, do whatever you want. But it's more like pick your poison.

[grim panther]

99% of "the cloud" runs on linux

[night girder]

That's not for human use.

[grim panther]

even the Apple stuff

[night girder]

Hence I said: getting work done.

[grim panther]

satisfactory runs on Linux

all major software developers with an impact in the last 30 years did use linux

(or bsd)

[night girder]

At a bank I worked. literally 0% used Linux. You will never, ever get that sold there. When I worked for the Dutch goverment, we used Windows. Same case, you can't convince management to deploy Linux on a large scale for human use.

[grim panther]

but almost every apple cultist is not about technical facts, it's about prestige.

[night girder]

Meh, I don't care honestly. The discussion is beaten to a pulp in this channel.

Most of us know it's not black and white and more grayish. I know Apple sucks.

[grim panther]

agreed.

[night girder]

But it does have it's stong points too.

And especially when you look at context.

[grim panther]

design. Definitly

[twin dew]

For the workers?
Or for the servers?

[night girder]

Is it for work? How important is security? How important is privacy (bank, goverment)? How easy is it to use?

[twin dew]

Lot of banks still use old mainframes with the mainframe specific unix variant.

[night girder]

👆

Like I said: dev and getting work done. aka not servers.

I know servers use it. But that's not what I was talking about.

[wanton orchid]

microsoft shit is a security fuck fest

[night girder]

While I agree, try convince management in a company.

[grim panther]

the thing is: Apple is not more secure than windows. This is an urban myth. Marketing about this is just better.

[night girder]

They just don't trust Linux + their employees. That combination 😂

[wanton orchid]

everyone needs an egg so they can get a cock

[night girder]

cook? 😂

[grim panther]

i never have seen a cook hatching out of an egg

[twin dew]

Made a SystemD unit that sees if the Pi 1:s Ethernet can get IP via DHCP in 60 seconds during boot, and if not, disables the DHCP client network profile and switches to the DHCP server network profile.

60 seconds from enabling the DHCP client profile.

Would like to plug it to be also run when link status changes, but would need extra software or "hacks".

[maiden coyote]

This looks neat www.navidrome.org/about/ I'm thinking about setting up a container for it this weekend

Steam my own personal libary rather than youtube or spotify

[tough owl]

Would probably work

Is it Foss?

[stray badger]

says open source

GPL

[maiden coyote]

Says it supports an api I've never heard of.. I'd be pretty happy with it if it has a decent web interface that can be loaded on a browser. Firefox has a media in background extention that I can use on my phone

[proud thicket]

How to fix the little hover helicopter thing

Blade won’t spin

[winter path]

I’m building new pc mid to low range
Pc part picker list: https://pcpartpicker.com/list/kpmwWt

Lmk any changes I should make pls and thank you

[jagged snow]

You don't need or want kryonaut, that mem is a bit expensive, that psu is massively overkill, and you don't have a gpu

[winter path]

I already have a 1660s and hope to upgrade. If not kryonaut what thermal paste?

[jagged snow]

If you want thermal grizzly, arrownaut or hydronaut

And, if possible, make the jump to a 5600

[winter path]

Base 5600?

[jagged snow]

Yes

[winter path]

Ight also what psu would you recommend?

[jagged snow]

What gpu are you planning to upgrade to?

[winter path]

Idk tbh

[jagged snow]

Then it's hard to say

With the system you have, you only need a 550w

[winter path]

Yea so a 650w?

Give me some room

[jagged snow]

It's very difficult to say

[winter path]

Can you give it your best shot?

Anyways thank you for the advice

[tough owl]

Wendell my beloved

[languid gulch]

any last-minute steam sales worth grabbing?

[grim panther]

Pony Island

[languid gulch]

that sounds either super innocent or super dirty

[grim panther]

it's super fun. Give it a try for the 1$ or whatever it is now

[tough owl]

74 pennies

[grim panther]

don't watch videos about it, this will spoil the fun

[soft bloom]

www.youtube.com/watch?v=BKorP55Aqvg
"Expert"
Classic needs to be remembered

[soft bloom]

oh i love this
youtube.com/watch?v=OUNXFHpUhu8
its about parking in US and bs "science"

[soft bloom]

i know 1 walking sim you didn't buy, but should have

A Short Hike

[languid gulch]

also i'm yet to find a parking lot that isn't an absolute nightmare to navigate in

[grim panther]

12 hours to go

[soft bloom]

i can only speak as pedestrian, and i agree

[languid gulch]

i'm super lucky that i have a smallish car, but also one with no back end & a fantastic rear camera, & it still sucks in general to park

[soft bloom]

i hate when parking lot occupies a lot of space so it heats up on sunny day and you have to somehow walk around it or through it

but also, here they exist only near gigantic malls, with some exceptions. here, in city, its more of a problem to find a place to park. but i prefer to develop public transport. when its done good, it can do more than a car does (within a city).

also, i remember when my ex-gf was discussing that we need a car because she was planning a baby... and now i look back at it as crazy fantasy of someone who tried to play in adult games

aka "on a totally unrelated note"

our world is shaped by ideas, visions that drive people to invest their resources, dreams that slowly come true. so what ideas are we going to implement today?

[stray badger]

if he is using the stock cooler, it comes with pre applied paste. better paste wont really make a difference at this point

[jagged snow]

That's also true

[mental oriole]

But does it work?

[soft bloom]

define non-working state

oh, i can tell you that ongrund transport has some problems when roads are occupied with hundreds of individual transportation vehicles

but we also have e-scooters, bycicles, subway, city train, tram
and roads are roaming with maaany private-enterprenour busses, regular busses, trolley buses...

perhaps this can be viewed as leftover from soviet union, one that had to be fixed with help of western countries like Poland (some trams roll ads in polska while here)

but in the end - i would prefer being in AC cooled tram for 10 minutes longer, than to worry about a car

its a different style of life, one that needs different conditions

[mental oriole]

When the bus has 💩 ac...

You melt.

[pure karma]

thats called a school bus

[languid gulch]

the big yellow school bus is 100% a real thing in the US

just to confuse the danes

https://www.youtube.com/watch?v=fs037FpNgrM

even the color is specific

i had no idea it wasn't a regular thing outside the US

at least until fairly recently

🤣

and yes, they are definitely as uncomfortable as they look

all leaf springs

it's basically 1 step past a covered wagon

my high school was also a designated nuclear bomb shelter 🤣

the entire first floor is below ground level

not underground, but it'll survive a blast wave

[mental oriole]

Because trains are not cars.

[languid gulch]

was sweden one of the countries that switched lane direction?

ha

about once a month we have some idiot who likes to think that the light rail around here is just a big truck they can beat

it's dumb enough to play chicken with any utility truck

let's up that to 70 tonnes

we studied doing an underground, but passed because of the cost

it was going to be $2B/km

oh we have a lot, but there's also 350m people, which means plenty of idiots

we also looked at doing elevated tracks, but that would have been almost as expensive

our ground is large rocks, sand, and clay, and just destroys normal excavation equipment

so whether it's digging tunnels or digging support structures, it's slow & destructive

oh if we had decent mass transit here i probably wouldn't own a car

but here you can tell what city a bus operates in based on how terrible it sounds

we're used to no water where i live 😛

i mean, it is a desert

more water would actually be climate disruptive

that's really only about 3 months out of the year

the rest of the time it's fine

it might drop below freezing twice in the winter, & we haven't had snow in 30 years

lol

it's usually about 15cm of rain a year

chicago's decent

i'd pick denver

oh not remotely 🤣

chicago's 2300km from me

🤣

it's a beautiful city

i mean, yes, but also that's a cartoon

ah 🤣 yea very different cities

good 🤣

it's a town on both sides of the river, & the missouri side grew faster

most of the "social" stuff is on the south side, with industry on the north

[twin dew]

In Finland there can be "school" bus routes and times, but they are run by the normal bus company and are open to everyone who pays.
As in extra times on normal bus routes or extra routes that only run on school start and stop times.

[languid gulch]

makes sense to me

both systems make sense to me if they both work fine

[twin dew]

The US one is very wasteful in a way.
Very large fleet of busses that only do 2-4 runs each day, with separate drivers etc.

But as US doesn't have alternative source of busses and drivers...

And the school busses in US have much stricker crash requirements etc. than the normal busses there.

So product of the environment.

[wanton orchid]

I don't get the shape
all the load seems to be on rear wheel

[languid gulch]

it's a glorified army truck

[wanton orchid]

yet it's front wheel driven

[wanton orchid]

must be why

[languid gulch]

more like the old 6x6

[wanton orchid]

its made like a big big simple car

[languid gulch]

lil bit 🤣

[twin dew]

But question becomes, why are the normal US busses allowed to be death traps in crashes 😅

[languid gulch]

their rollovers are awful

but front/side impacts are usually ok because of how high up the passengers are

when schools do a long road trip they'll usually charter a "real" bus for that, which is a lot safer

?

oh

also i have no idea where the bus funding comes from, but if it's from individual cities/school districts, that might explain why they don't really get upgraded

i suspect we all have it hard, just in different ways

ehhhh

depends on how far back you go

i think 1860 sucked pretty much everywhere

right before our civil war

cats and decent food helps

i try to avoid them

[twin dew]

Yes, school/school disctrict fund are used for the whole "system".

[languid gulch]

our tax system is incredibly byzantine, and schools aren't well funded, so idk if busses are part of that or not

i'd have to move to ireland since i don't speak any european languages 🤣

i mean, a fair bit of spanish, but i haven't used it in a long time

[twin dew]

In most cases you have 5-10 years to learn the language after moving in?

To very basic level.

With cheap/free evening classes available.

[languid gulch]

sadly no

seeing him hack a mothership with a 30 year old mac is just too much for me now

oh jeff without a doubt, he's having the best life near as i can tell

also some amazing non-john williams music

i did appreciate that they kept the avengers theme throughout the movies, but that's really the only really memorable music i can think of lately

ah right

he's done some great stuff

forrest gump

i just worry, a lot of the movie music composers are old now

he's up near 90

[twin dew]

For "some" reason, this chipset heatsink isn't that good at keeping the AMD X570 chipset temperatures down 🥲

But as the first slot will contain SAS card that has official minimum airflow requirement of 200 linear feet per minute of air (trying to convert to something more normal atm.)...

Ok, less in reality as the full requirement is "Minimum airflow: 200 linear feet per minute at 55 °C inlet", nice for LSI/Broadcom to mix units.

So 1 meter/second "wind" speed required at that 55C inlet air temperature.

But will see if I have to do some kind of airflow guide, as the chipset sink is "below" the case can location.

[twin dew]

Need to see if the temperature goes up or down once in OS and actually using the chipset connected things vs. BIOS.

[twin dew]

But at least the MB seems to be getting BIOS update to AGESA 1.2.0.C with LogoFAIL fix, currently in open beta.

When the previous BIOS is from end of 2022 with AGESA 1.2.0.7, for 5800X3D support.

[wanton orchid]

why are you noticing just now
I should clean my fan btw

[twin dew]

Nah, not just noticing now.
Knew before I bought the MB.

[wanton orchid]

it gets between 75 and 85 C with good airflow

[wanton orchid]

it goes down

but not a lot

[twin dew]

Less than 85C with almost no airflow when I was tuning the fans and was running them all at 20% PWM.
So about 400 RPM for the case fans.

[wanton orchid]

In bios is usually 80-89C

with 25 C air

[twin dew]

But there is a reason why I switched the stock paste to Heilos for that chipset sink and the LAN sink.

Question just is if that + the 65% fan speed that I set as minimum are enough, or if I need to make air guide to direct one intake fans air down.

[wanton orchid]

I struggled with mine
because of the clip things

it's hard to get it clapped down without swinging

I wonder if I should re do mine

seen a ~5C improvement last time

I keep it around 80C for safety

I'm scared it will wear fast above that

even if it's supposed to be bigger size node

it's not very sensitive so I have very steep curve

maybe I should do the other fan control logic things

[twin dew]

94C in the IPMI for the warning temp for that.

[wanton orchid]

but translated menus are not very intuitive

[twin dew]

Why not use the English menus then?

[wanton orchid]

I'm talking about the English menu

it's probably translated from Taiwanese or similar idk

[twin dew]

Just normal server talk.

[wanton orchid]

and I've seen no doc for the "open loop" cooling thing

[twin dew]

Open loop: Temperature to PWM direct correlation, no feedback.

Closed loop: Target temperature with closed loop control.

[wanton orchid]

I still don't get how both rules interact with each other's when you set two tables

[twin dew]

Highest wanted PWM value is used.

[wanton orchid]

but isn't that infinit (100% always) when in closed loop ?

[twin dew]

So I set Open loop table with just 10C for 65% PWM, and then set the real settings in closed loop per temperature sensor with target temperature, and then the faster reacting temperature above that.

Used the open loop to set minimum, and closed loops then to work between that set minimum and 100% depending on sensor temperatures.

[wanton orchid]

I should set target as 80C then

I just don't get the stepping direction thing

[twin dew]

So Closed Loop Table 3 is for CPU temperature for me.
Set it to Ramp temperature: 75C (try to keep the CPU at 75C), with ramp rate 1% (if over that temp, raise PWM in 1-percent-point steps), ramp delay 5s (5s between steps).
And then Slew 85C, Slew rate 1%, Slew Delay 1s, so if the CPU hits 85C, the fan ramping moves to 1% and 1s steps.

For faster reactions.

And once you have tables set.
You go to the Temperature Sensor and Corresponding Fan Table page.
Click one sensor, select 0-1 open loop tables, 0-1 closed loop tables, then check the fans (right checkbox, not the left button) you want that sensor to direct based on those tables, and save.
And repeat for all sensors you want to use.

Clicking on fan button shows which sensors are used for control.

So I have dual-chamber case, I have the second HDD chamber set to 50% base PWM, and control based on extra temperature sensor on open table with 10C, 50%, and then closed loop table for fans 1 and 4 (the input and output fan headers for that chambers fans).
And Fan 3 is CPU, so it is controlled by just CPU sensor, on Open loop for 65% and closed loop that I specified.
And then main chamber input and output case fans are controlled by MB, cards side, CPU, LAN, chipset and DDR temps, with 65% PWM minimum via open loop table for all, and then sensor specific closed loop tables.

Classic PC MB fan control is the open loop type.
As closed loop can cause oscillation in the speed if you don't tune it properly, which is annoying.

Se on Open Loop side, you can just specify temperature in steps, with corresponding PWM signal, and then the PWM value for the highest temp reached on the table is used.

But on sensor level:
With open loop table, PWM wanted is what the table says based on absolute temperature.
With closed loop table, PWM is moved in steps with specific delay between steps to keep the sensor temperature at specific value.

When multiple tables per sensor and/or multiple sensors per fan are in use, actual PWM is highest of the wanted values.

[wanton orchid]

so it virtually keeps track of the closed loop table behind the open table value when the open loop pwm is higher

ok

All the selecting fan and matching tables and sensor I got it right dw

[twin dew]

The closed loop probably goes to 0% PWM wanted in that case.

[wanton orchid]

it's the closed loop and mixing I was confused about

[twin dew]

Until the temperature goes below the closed loop set value.

I mean above.

[wanton orchid]

I mean if temp above and still in open loop value, it ramps up virtually until it catch up

[twin dew]

Yes.

[wanton orchid]

ok nice thank you

[twin dew]

Delays the response, but it will respond once the temp is above the closed loop set value.

[wanton orchid]

it's actually the open loop that was causing me annoying oscilation
because it is 1C step sensor, and the chipset temp is not sensitive

so it was 79 -> 50%, 80 -> 65%, 79 -> 50%

[twin dew]

Yeah, but you can change the tables, so the steps are smaller.

[wanton orchid]

I'm going to setup the closed loop now that you helped me decode the slew and ramp things

I was like, wtf is slew if there already ramp

[twin dew]

Or optimize one table for specific sensor etc.

[wanton orchid]

no becuase then the chipset temp will not get constrained as much

the correct choice is closed loop in this case

slow ramp up, at same temp

if I select the temp correctly it will ramp up and down so slowly it will not be a problem

I guess ?

I will check anyway

[twin dew]

But yeah, that is the point of closed loop control.
Often harder to tune, but better when working right.

Which is I set the minimums to where the fans are still inaudible when the machine was next to me on floor.
And then set closed loop control with setpoints.

80C Ramp, 85C Slew targets for the chipset for example.

Sitting currently at 76C when in BIOS.

[wanton orchid]

that's probably what I will go with

[wanton orchid]

maybe I should do repaste too, though I suspect you have cooler air intake

mine is behind a radiator

currently sitting 85C as we talk

[twin dew]

60/65 for MB, 60/65 for Card, 75/85 for CPU, 60/65 for LAN, 80/85 for X570, 45/50 for the second chamber thermal sensor set between the case halves of the PSU (might need to move), 60/70 for the DDR sensors.

But currently at 50 MB, 35 Card, 40 CPU, 76 X570, 29 extra sensor, 31 all 4 DIMMs.

[twin dew]

With 1000RPM for HDD side 120mm intakes, 800RPM for 140mm out, 1300 for main side 120mm intakes and out, and 1000 RPM for 120mm Noctua on CPU.

[wanton orchid]

my fan are 1000rpm and 1200 rpm respectively

the other 2 fans are aio controled

I have complex setup

[twin dew]

But 50% for the HDD side, and 65% for main side and CPU as the base, which they are in now.

[wanton orchid]

my goal is the chipset temp also serv feedback for radiator temp

the best imo would be a external fan control board

with the capabilities I'd like

I have note on this

[twin dew]

This thing, now with two HDDs in aft cage.

And the SSDs on the bottom of main side that the cables are in place for.

[wanton orchid]

went for compactness

[wanton orchid]

radiator was not fitting top because of ram mount

I also got an apc

[twin dew]

The two network cables not managed are extras until I get that "new" server into use, and can get rid of the old one.

[wanton orchid]

with a switch behind the define mini c

man ngl It makes me smile how similar the setups are

"slew" still confuses me a little bit

[twin dew]

Ramp = slow control, slew = fast control, which can easily overdrive.

[wanton orchid]

do you agree that it slow down when the temp match, and speed up when the temp is match+1 ?

[twin dew]

Leading to ringing (control signal oscillating over and below wanted)

Just two simple controllers, with two setpoints.

Think them as two separate controls.
Ramp is just meant to be the lower temperature, slower reacting one, and the ramp is "shit has gone to fuck, lets keep the HW from overheating" fast one.

[wanton orchid]

yep

[twin dew]

Slow normal one, fast emergency one.

But yes, both work by the simple rule, that if temp is above the setpoint, increase the fan PWM in step specified, then wait the time before next check.

And if below, same, just slow down a step.

[wanton orchid]

lets say I put 80C ramp
at 80C it will be slow down, and at 81 it will ramp up
right ?

so it is single point control

[twin dew]

Setpoint 80C, fan speed 50%.
Actual temperature rises to 81C, PWM adjusted to 51%, then wait x seconds for response.
Then adjust again if needed, repeat.

If 80C setpoint, 50% PWM, and temperature drops to 79C, PWM adjusted to 49% (if 1% step).

And no limit on how high the PWM can go to keep the specific temperature.

[wanton orchid]

so if 80c -> 81C let say it end up +15% until it goes back to 80C
it will stay +15% until 79C ? or it will slow down already ?

[twin dew]

20% or so seems to be the minimum, 100% would be the maximum.

Depends on the exact algo used.

[wanton orchid]

because if it waits 79C its misleading

[twin dew]

In this case probably the stupid one that will not slow early.

[wanton orchid]

it's hysteresis

[twin dew]

Because with full PID control, you would also need to tune those extra parameters that control the "getting near to target" etc.

[wanton orchid]

if it waits 79C its 2 point algo even though you set single point
if it slow already at 80C then it's really single point

[twin dew]

Probably the ASRock Rack Closed loop is just P-controller, and missing the integration and derivation of full PID-controller.

Why would it do anything when the temperature is "correct"?

When not full PID.

P is correct, and I and D don't exist.

[wanton orchid]

because if temperature is correct, no need to blast fan

[twin dew]

No adjustment.

If the fan is too fast, it will overcool, and cause the temperature to go too low.

[wanton orchid]

single point is easier to implement than 2 point also

you remove the "do nothing" condition

[twin dew]

And if the values are wrong, you end up oscillating around the setpoint.

The response time and response speed ones.

[wanton orchid]

you will always oscilate around the set point

2 point is long oscilation

1 point is fast easy to stabilize oscilation

[dim sandal]

is there any mod to add fsr 3 (frame gen) to satisfactory?

[twin dew]

With properly tuned PID, with constant conditions, you don't.

[wanton orchid]

off topic

[twin dew]

And even wrong server for modding. Read the #welcome

[dim sandal]

my bad, thought this was the appropriate channel for asking about adding fsr / dll files

[wanton orchid]

this channel is not about satisfactory

anyway, I'll test this

see what it does

[twin dew]

But two-point you are talking about is just loosened temperature P-controller, which wanders between the low and high setpoints.

https://en.wikipedia.org/wiki/Proportional–integral–derivative_controller
And point of tuning a PID controller is to get result that smooths out and stops oscillating after suitable time.

But needs to be tuned to the response time of the system.

[wanton orchid]

no its random when in the sensor sensitivity blindness
single point is match

[twin dew]

And if you want overcorrection or undercorrection.

[wanton orchid]

if every second you do +1% then -1% then +1%
it's fully stable

if you have sensor sensitivity hysteresis problem

it will in fact be unstable

anyway I hope its single point logic then

[twin dew]

Ah, yeah, sorry, not even P-controller, as that would set larger change on larger difference from setpoint.

[wanton orchid]

ok now THAT is stupid

it never slow down, it just resets

[twin dew]

What were the settings?

[wanton orchid]

^

whatever the settings

the fact it resets it always stupids

[twin dew]

And there wasn't other controls for that fan?
Moment.

[wanton orchid]

why would there be

resets means : cooling feed drop, so heat burst, and also ofcourse fan sudden stop annoyance

it will ALWAYS be absolutly stupid

[twin dew]

Like I said, wait, I'm testing with lowered setpoint for the chipset

[wanton orchid]

speed reset can only be implementation bug

because it completely defeat the purpose of closed loop

this is no closed loop, but blast loop then

[twin dew]

Like I said, wait a moment for me to test.

[wanton orchid]

what dont you get

[twin dew]

Want to see that it isn't some other config thing causing it.

[wanton orchid]

this is reset

reset basically means that it WILL oscilate no matter what, and in a bad way
making the air flow ALWAYS stupid

who ever thought of that really

fuck you

[twin dew]

But if it doesn't work, just make open loop table with 1% steps.

Around the wanted temp for the sensor.

[wanton orchid]

that's what I did

[wanton orchid]

not 1% step, but step that make it stabilize

that's the opposite of 1% step

[twin dew]

With small temperature steps too I mean.

[wanton orchid]

i.e the most temp stable table would be 80 -> 0% 81C -> 100%

@brave plover we are talking $500 hardware btw

this is really bad

[wanton orchid]

btw it's even worse than reset on lower temp, it reset on the SLEW RAMP

so it is still in the ramp up field when it resets

this is double stupid

there is no excuses for that

there is a in-between between thoroughly tested, and NEVER tested

I will check if the ramp up still works though

if only slew is broken maybe that's ok

this still blows my mind

[twin dew]

Removed the other controls except the closed loop for the chipset and lowered the target temp.
Started from 400 RPM and went up to 1800RPM, then lost connection as some change I did is "unstable" (disabled lot of the IPMI LAN stuff) or my main computer side fucks up the management VLAN connection, waiting for the IPMI connection to return.

[wanton orchid]

does not ramp up on RAMP SETTINGS

[twin dew]

Fans still at 1800, as the sensor is above the setpoint, setting it closer to the actual values.

Chipset 71C, Ramp 70, 1, 1, Slew 75, 1, 1.

Changed to 5% rates.

Probably the direct connection with Intel 2.5G LAN controller which has bugs in even the latest revision and latest firmware.

Put switch in between.

[wanton orchid]

what would it have to do with fan control ?

[twin dew]

I kept losing monitoring.

[wanton orchid]

oh

[twin dew]

Because I lost the network connection to the IPMI.

[wanton orchid]

did you update it with the custom fimrware ?

the board does not work with intel native ones

[twin dew]

Not on the ASrock board, on my main computer.

[wanton orchid]

oh

[twin dew]

Intel i225 rev 4.

[wanton orchid]

anyway

[twin dew]

The i225 and i226 (i225 rev 5 in all but name) are clusterfuck.

[wanton orchid]

no ramp up on ramp, reset on slew

I will restore my open loop tune

never again asrock

no wonder I had trouble figuring out how it was supposed to work when I tested it

thank you baldur for pointing how it was not an understanding issue after all

[twin dew]

Yeah, or I'm misunderstanding it.
But I do recall some talk about it being fucked.

[wanton orchid]

I mean there is fucked and FUCKED

I was already pissed off by the ramp up/down delay not being taken account in Asus bios

this is another level

the last bits of trust I had left in motherboard cooling management .. is GONE

btw fuck 1C increments

1C is huge

you at least want 0.3 /1F increments

this sounds like graphic environnement rounding, then using that rounded value for low level management

[twin dew]

Set Ramp to whatever, set slew to 75, 1, 5.
Been rising the fan speed form 400, now to 1000 with chipset at 79C right now, 74 when started and the fan speed reset.
Will see what happens over time.

As the charts only update every 15 seconds.

chipset temp started to drop, 1400 RPM.

1500 RPM after longer pause.

76C, 1600 RPM.

Ok, dropped back to minimum once the temp went under the slew setpoint.
But might have been too affressive Ramp settings.
Might be that this is two-point with bad names.
In theory trying to keep between the two given values, but the settings I and you had for ramping down were overaggressive.

So another test with Ramp 70, 1, 5 and Slew 75, 1, 5.

Does it reset or ramp down once the temp drops below 75.

[wanton orchid]

I tested ramp, it does not work

[twin dew]

Testing if the "Ramp" is "Ramp Down" and "Slew" is "Ramp Up".

[wanton orchid]

my guess is the reverted ramp and slew value
interpreted it as 2 point set, but using it the reverse way
so ramp is down but never triggered, because reset triggers first after slew get below set point

anyway that's some baby level error

[twin dew]

The chipset temp is starting to come down on the second test. Will see in moment.

[wanton orchid]

note you can't set the same point

it cant ramp down, if it resets

[twin dew]

Trying to see if it was reset, or overaggressive ramp down.

But yeah, it reset, not ramp down. From max fan to min fan in 30 seconds.

So I need to set per sensor open tables.

Yeah, on the ASRock Rack support FAQ the terms are Slow Down and Speed Up, otherwise same fields.
They have fucked the AM4 firmware.

www.asrockrack.com/support/faq.asp?id=41

Sample Desc: Assuming above Close Loop Fan Table is for CPU temperature sensor, when the CPU temperature is higher than 60 degrees, the fan duty cycle will increase at a rate of 1% per second; when the CPU temperature is lower than 50 degrees, the fan duty cycle will decrease by 1% every 5 seconds.

The slow down part doesn't work on X570D4U at least.

And just ends up resetting the fan speed instead.

And yes, the disconnects were because the IPMI LAN chip on the X570 didn't like the i225 on my computer, because the i225 is fucked.

Before Intel engineer layoffs, their LAN chips were the best in everything.
After the layoffs, all have had major HW level bugs.

With that i225/i226 being the most fucked up AFAIK.
The server chips just had to have acceleration features disabled.

[wanton orchid]

it's always cheaper to lie on your product capabilities then claim the oopsies at court

naivety at every level

[twin dew]

And I need to try reinstalling my computer again at some point, because I have been getting weekly BSODs for KERNEL_SECURITY_CHECK_FAILURE.
Probably either the AMD RAID drivers, and need to try another driver version, or the security suite.
But kernel is detecting that some critical kernel data structure is corrupt.
Stack based buffer corrupted for subtype.

I did originally use AMD RAID driver version that isn't on AMD site yet, as the "stable" version was the same that had problems for me in past.
But seems the problem side was the AGESA/BIOS, not the driver.

Need to check with WinDBG at some point...

[twin dew]

Yeah, not enough information in the minidump, set to kernel dump for next.

[twin dew]

@wanton orchid The ASRock Rack AM4 boards are special products made on cheap for some specific buyer. The firmware has always been flaky on them as Rack threw consumer firmware from main ASRock together with snippets from actual Rack products on the cheap.

Sorry, the other way around, bolting consumer ASRock code onto ASRock Rack firmware for other products.

[wanton orchid]

they didnt advertise it as prototype open beta

also fan control logics should not be socket dependent

ngl this threw me off completely

I'm tired of this bullshit

[twin dew]

I was first thinking about the X470D4U, but that is way worse clusterfuck. The X570D4U is mostly working, the B550D4U is almost fully functional IIRC but almost impossible to get.
Don't know about the new AM5 versions.

[wanton orchid]

I didnt see any Am5 yet in their line

[twin dew]

B650D4U(-2T2L/BCM) is for sale in retail world wide at least.

In stock too.

[wanton orchid]

ahah

they ditched the intel nic

no more intel firmware fucking around

[twin dew]

Even on X570 during COVID.

And the fuckup was ASRock Rack side, that they wired one of the 10G ports to the IPMI.

And wired some reset pin wrong way, that only affected things during firmware update.

Just couldn't get the chips from Intel.

So switched to Broadcom.

[wanton orchid]

waiting for the ryzen 9000 updates anyway

[twin dew]

There is X570D4U-2L2T/BCM too, came during COVID times when ASRock couldn't produce the Intel version because of chip supply not being there from Intel.

And with the /BCM, they moved the IPMI slave to one of the 1G ports like sane design.

So X570D4U and X570D4U-2L2T/BCM have separate IPMI LAN Port, and one of the 1G ports also is connected to IPMI.
With X570D4U-2L2T they for some insane reason connected 10G port to the IPMI in addition to the separate IPMI port.

[wanton orchid]

I hope they learned with the x570d4u-2l2t

because it is really infuriating

the single fact that this board showed in almost any single exclusive specific workaround in the faq troubleshooting sections is telling

[twin dew]

But the firmware "killing" the LAN chip is AFAIK because of that wrongly wired pin.
The ASRock Rack specific firmware contains instruction to ignore it.

And that they made the same PCB design fuckup on Epyc board too...

But with the Epyc 4004 launch (Ryzen 7000 but Epyc branded), the MB selection for AM5 servers got much larger.

Supermicro etc. also making MBs for AM5 now with IPMI.

[wanton orchid]

intel nic firmware custom, hard to find
bmc/ipmi flash dos utility with spi overwrite
can't cmos the bios from bmc/ipmi
most board components not showing up (recovered from board components not showing up AT ALL)
bios setup/initializing problems
fan control issues
ipmi random reset
inconsistent ipmi network access
bios sometimes failing to initialize/train for some reasons
no agesa update published
chipset still undercooled even if it's apparently more stable than the one before who would so often burn down ?

this is a lot of problems for a $500 board

[twin dew]

The chipset and lan cooling is to server levels of airflow.

The chipset cooler cannot be higher without blocking long add-on cards.

RAM not training is probably because the board is limited to stock VRAM voltage, and most consumer kits expect 0.15V more for the "marketing spec" speed.

At least in early BIOS versions you could "change" the VRAM voltages, but that did nothing, as the VDIMM regulation is fixed.

And the Intel LAN firmwares that Intel publishes are just for Intel made add-on cards.
Not for the chips on anything else.

It is just that ALL the MB makers suck at providing updates LAN firmwares.

[wanton orchid]

I mean this is an atx board, so air flow is less dense than on a rack setup

[twin dew]

No, plenty of (m)ATX server rack cases.

And that is why the RAM and socket is 90 degrees turned.

And all the other heatsinks have fins in the direction they have.

[wanton orchid]

fair point for the ram

[twin dew]

The mistake is that the pages don't spec the minimum airflow visibly, as that is to be expected for actual server boards.

With that 200 LFM being very common spec.

1m/s airflow direct over all the heatsinks on the board.

The sinks are sized according to normal server board standards, and to level that is enough for the original buyer.
That hobbyist that then also buy the board don't expect that isn't that unexpected, but not really fault of the board.
The firmware stuff is.

[wanton orchid]

noctua 120mm shoudl still do it

[wanton orchid]

I mean as stated this is not as much an issue, I'm mostly blaming the firmware
but that (the firmware) is a big one

[wanton orchid]

I dont know what are their target temp, because server are usually rated for 50C
this is 25C air flow

or I'm wrong on the 50C server rating ?

[twin dew]

That SAS card was 200LFM @ 55C intake temp to keep it from overheating over silicon rating in full use.

So yes, with lower intake temp, the airflow requirement goes down.

[wanton orchid]

yep so if it does 75 at 25C intake, I dont know if it could survive well at 50C intake

[twin dew]

But the point is that the heatsinks are sized expecting significant amount of direct airflow, that doesn't happen in consumer cases.

Almost all the other X570 boards had small chipset fans instead.

Until much later variants with AMD side improvements to drop the heat output.

[wanton orchid]

do you know what that is ?

I mean why does it have the plastic thing ?

[twin dew]

Not really without context.
Some kind of MB connector stack with single USB and something below it.

As the pic is tiny.

[wanton orchid]

I ask because there is just plastic below right ?

[twin dew]

What MB?

[wanton orchid]

the decription says USB 3.2 gen 1 x4

B650D4U3-2Q/BCM

I'm just curious

in my book a single port flat would be easier to build

[twin dew]

Another version of the same MB has extra LAN ports above those single USB ports.

And probably couldn't get parts that would fit both this and that with same backplate without the raising the USB ports on this.

But yes, just upper USB port with nothing in the bottom position.

[wanton orchid]

the other version have nothing under it too

[twin dew]

But part for having single USB and LAN port at same position probably doesn't exist.

So they used that kind of weirdness.

That did exist already for some reason.

Or might be using same backplate as something else that has 2 USB ports per position.

[wanton orchid]

that makes sense

I'm still surprised that having elevated special usb port and ethernet ports parts like that
is cheaper than having adapted backplate

[twin dew]

New stamping form would be somewhere around 200k IIRC. And possibly extra setup costs from switching the factory between one more stamping instead of just producing more of the same.

Having to stop the line, change the part specific things, test, start the line for real.
Both production time lost, employee time used, and extra scrap made in the test.

[wanton orchid]

here the fan is completly handled by ipmi/bmc

so no agesa work around or alike

with updated ast chip

hopefully fan management fixed

[twin dew]

Same in the X570?
But the BMC firmware version is fucked for some reason.
Fucked up when they adapted it to talk to the fucked up UEFI.

[wanton orchid]

in x570 its ast2500

I think there have been big issues with ast2500 firmware

because after the load of issues, I didnt wait long before seeing ast2600 announced

apparently in this version it still uses big etherent ports for NCSI

but its broadcom so maybe be better adapted

[twin dew]

AST2500 is 1x ARM11 @ 800MHz, AST2600 is 2 Arm A7 and 1 Arm M3, so lot more raw performance.

[twin dew]

Because that "lower end" board version doesn't have any other normal LAN ports.
Just IPMI specific 1G and then 2x 25G SFP+.

The higher version then adds two extra 1G ports.

[wanton orchid]

yes, that version have 2x 1g i210(at)

[twin dew]

But might have just designed for the "lower end" without them, and just add the 1G ports on the other without changes to the others.

Yeah, and it is listed as having 2x 25G on the actual page.

Just mistake in that chart.

[wanton orchid]

nono I have done the mistake

didnt know there was 2 i210 + 10G

sfp28 is BCM57502

[twin dew]

Thought you were still at the B650D4U3:s which have the 25G (with or without 1G), not the D4U2:s which have either 2x 1G, 2x 1G + 2x 10G, or just 2x 10G.

Didn't read the model in the block diagram pic.

And they even have chart for Intel version of the 2L2T in the manual, but it isn't sold atm.

[wanton orchid]

anymore or not yet ?

if not yet didnt know they had plan for going with intel again

[twin dew]

Not on the pages as product.

And they do have B650D4U-2T/BCM on the pages, but that doesn't have manual etc. yet.

[wanton orchid]

sorry bad quality

[twin dew]

If it was sold, it is on the pages, just marked as not being sold anymore.
But all products on the pages aren't sold yet/ever.

But I really have no idea why they connect the slave LAN connection for the IPMI to 10G port...

[wanton orchid]

here we go

[twin dew]

And that one has slaves to both one 1G and one 25G.

And to separate port.

[wanton orchid]

maybe ast2600 is wired in safer way than x570d4u was and when disabled it really does not fuck with the ports at all (in which case it's your choice then)

[twin dew]

Like I said, the actual cause was wrongly wired reset pin.

Not that AST2500 was wired to the 10G port.
That just causes/caused performance issues.

That specific 10G port not getting full performance the other one got in some cases.

[wanton orchid]

btw I'm happy I found spec for rock 5b+
should be under way

[twin dew]

If you flashed the Intel normal firmware, the firmware noticed that the chip for that one port was told to be constantly in reset state.
So it didn't work.
The ASRock Rack specific firmware constains stuff to ignore the specific reset pin.

[wanton orchid]

still dont know how sim card is wired to optional 4g modules though

[willow pike]

@twin dew so remember when you told me 13900k/14900k were fundamentally fucked

https://www.youtube.com/watch?v=QzHcrbT5D_Y

it just doesn't end

[twin dew]

Not fundamentally, but in combination of then default settings.
The new defaults AFAIK work, but lower performance.

[willow pike]

in this video, server operators with workstation motherboards are also crashing

[twin dew]

Cellular modem compatible M.2 slots have their own key and own pinout.

[wanton orchid]

I really hope, ryzen 9000 learned from ryzen 7000 problems and matured am5 platform
because these will drive a lot of people to them

[willow pike]

not the kind that will crank the PL1 to a million watts

[twin dew]

That is connected to the MB sim slot.

[wanton orchid]

yep, rock5b+ have M.2 B instead of E, and integrated Wifi/bluetooth instead of optional

but dont know for the sim card connection

[willow pike]

there's a Lot that i couldn't fit into one message, you will find it an entertaining and disturbing watch

[twin dew]

Not WiFi, that is again another key compared to SSD M.2:s

[willow pike]

tl;dr my username

[wanton orchid]

I said B instead of E, E is for wifi
so no E key, wifi/bluetooth is integrated instead

[twin dew]

Ok.
But you can check the various keys and what they provide on Wikipedia for example.
https://en.wikipedia.org/wiki/M.2#Form_factors_and_keying

B has UIM for SIM card.

[wanton orchid]

That is for M.2 module, I mean physically

[twin dew]

Board has SIM card slot, that is wired to that B-type M.2 slot, that then connects the SIM card slot to the cellular modem board.

[wanton orchid]

I mean where the sim card is inserted etc, I still dont know, I'm waiting more review/pictures of rock5b+ but I'm hyped

[twin dew]

Ok, cannot know the exact placement on the board for the slot.

[wanton orchid]

yep

dw, I was just sharing hype

it's just that I wonder

[twin dew]

But the whole point of that mess is to have the card slot on edge of the normal case in laptops for swapping.
Instead of "sane" way of having the SIM card be inserted into the cellular modem module.

Which would put more constraints on the machine designers.

[wanton orchid]

and would make for sim card slot access fucked up

so practically unusable

[twin dew]

No, just harder to design for it to be on the edge.
As the cellular modem module placement would need to be on the edge.

[wanton orchid]

but physically accessing the sim card would also put more mechanical constraints on the M.2 and screw slot

and the modules

I like that it is separated

I just wonder how on the rock5b+

[twin dew]

Yes, easier to design, but makes it so that you can only add one into MB that is designed for it.

Plus for the laptop designers etc.
Minus in a way for the consumers.

[wanton orchid]

you mean instead would be better to have E key compatible 4g modules ?

[sharp oasis]

Mmmm. I love when my boss personally attacks me instead of being honest when he fires me.

[wanton orchid]

in a way 4g module manufacturer could still do these

[twin dew]

Not sure if the modules even use USB, or only PCIe.
The reason WLAN cards use E-key is that almost all the WLAN chips use PCIe for WLAN, but USB for bluetooth.

[sharp oasis]

almost kills self working for you in severely adverse conditions
"Not enough work ethic"
Gets fired and is told to basically idc figure out how your supposed to refund your work equipment.

I'm not actually kidding

[twin dew]

If the chips uses PCIe for the bluetooth too, they could be made to work in any PCIe slot.

[sharp oasis]

I was not sweating, this was the second bale load. My body was beginning to fail from what Kelsey said.

[wanton orchid]

wait B key have audio ??

[twin dew]

I2C can be used for audio chips.

Ah, sorry, it had even separate audio entry.

But might be just Wikipedia mistake.
The actual pinout etc. doesn't seem to support that.

But it has 4 PCIe WWAN confis and 4 SSIC WWAN configs, so maybe one of those configures some of the general purpose pins for audio?

4 config pins for 15 different configs for the pins (and not used right now state).

SATA SSD, PCIe SSD, then 13 different WWAN configs, 4x PCIe + USB3, 4x PCIe, 4x SSIC and one PCIe+USB vendor defined.

[obtuse geyser]

Question to excel gurus:
How can I copy and paste table with images from web, so that images are saved inside the excel document?
Using libre office calc, when I go offline it does this

[jagged snow]

Right until mobile chip naming gets brought up

Dont be a fanboy for anyone, it just makes you look like an idiot

[willow pike]

i am not renaming myself ryzen ai

(also i know the name is dumb but i don't fanboy amd)

[jagged snow]

really?

[willow pike]

perfectly honestly yeah

[jagged snow]

^?

[willow pike]

it's a pretty old name

that's called doing a bit

[jagged snow]

This is the internet, no inflection is carried through text unless you actually place emphasis on it.

[willow pike]

true that

[jagged snow]

Additionally, the majority of internet cretins would have been completely serious when saying something like that

[willow pike]

you do raise a good point

[jagged snow]

I'm glad to hear that

[willow pike]

been thinking of a new name but effort

[tough owl]

https://tenor.com/view/tom-and-jerry-intel-fanboy-computer-builds-pc-builds-expensive-gif-17297158

[stray badger]

7950x3d moment

[tough owl]

https://www.youtube.com/watch?v=oAE4NWoyMZk

Just a talking heads video

But IG GN has some sort of explanation that they've received

But aren't sharing due yet due to it not being fact checked

[past coral]

i have been able to get only pal world to run when i threw in my lower end graphics card

so im starting to be convinced i spent 1600ish on a brick

[jagged snow]

Can you clearly lay out what your symptoms are, and what you've tried?

[strong siren]

my only tip is to get the official nvidia driver tool and do the driver wizard thingy it shouldnt even need to reboot

[past coral]

black screen and close for palworld

laggy in end game for satisfactory

a few other games dont work all together but i dont expect a 2012 game to work well with newer stuff

[jagged snow]

How laggy? What are your settings?

[jagged snow]

Could be normal or not, depending what exactly is happening

[past coral]

40 fps low everything aside from render

[jagged snow]

Load up whatever save is giving you issues and type stat unit in the unreal console

[strong siren]

satisfactory lags when i have a ton of stuff built even with low settings btw

it might be an engine thing

[slender blade]

specs?

[past coral]

does this write shadows for moving stuff on conveyers?

[jagged snow]

Engine limits only come into effect in the largest of saves

[strong siren]

5600 3080 32ddr4

[slender blade]

5600 what

[strong siren]

ryzen cpu

[jagged snow]

Same goes for you

stat unit in the satisfactory console

[past coral]

i think im just going to buy prebuilds from now long lol

[strong siren]

i cant rn lmao im either just bottlenecked by cpu or theres something with my settings

[jagged snow]

@past coral

[past coral]

for real?

what did you do to solve this

[strong siren]

i didnt solve it lmao

i play at 30 fps when inlm in the middle of my fuel powerplant

the glorius 400 generator powerplant

[agile smelt]

what setting is ur graphics API on

[past coral]

api?

Directx12

i dont usually mess with that

[jagged snow]

What is it about this server that attracts people that are unable to help you help them

[strong siren]

happens

[past coral]

i will try it on vulkjan

[agile smelt]

ig direct x12 should be right for ur specs

but try vulkjan and direct x 11 anyway

who knows

also

cap your fps to ur monitors frequency

[past coral]

its running better with vulan

thank you guys

this game one more step towards being worth 30$

[slender blade]

its worth 45

imo

[past coral]

have you suffered a lobotomy/

you are insane

[wary olive]

@slender blade wym by “ew..palworld”

[past coral]

respectfully

[slender blade]

palword makes me go ew so i said ew palword

[wary olive]

Why

[slender blade]

a WHAT.. FIRE IN THE HOLE!!!

[wary olive]

Palworld is goated

[past coral]

i dont really like it but i do enjoy the fact i can use the balls on people

[wary olive]

Guns Pokémon and ark all mixed into one game? Beautiful

[slender blade]

lietenannt b-7 sucks epsilon 11 is so much better

[past coral]

thats what i thought

[cobalt ivy]

when i think of palword i imagine myself all people which were easyli turned on by any eevee-lution in Pokemon

[past coral]

i wish it would run for me 😭

[wary olive]

How does it not? Shitty performance?

[tough owl]

you tried

[jagged snow]

I did

[tough owl]

https://tenor.com/view/sad-cat-cat-in-front-of-sea-gif-16930437336998530733

[jagged snow]

Man 💀

I am in an unfortunate scenario

[tough owl]

Really are

[jagged snow]

No my gpu ran into the same issue as before with cooler pull away and I can't seem to fix it this time

I can't find my thermal paste, and the fastest I can get more here is 10 days

[tough owl]

Ah. That's really long. But I guess small town moment

[jagged snow]

Yep

And that's with prime

Newegg isn't any faster either

[languid gulch]

falcon 9 had a failure for the first time in 9 years

https://x.com/NASASpaceflight/status/1811608378520588583

Chris Bergin - NSF (@NASASpaceflight) on X

Here are two minutes of the SpaceX launch stream where the ice build-up began.

Twitter · Jul 12, 2024, 3:47 AM

looks like a lox leak

[edgy hazel]

People really act surprised at headlines like "Gay furry hacked heritage foundation" as if the Venn diagram of gay furries and hackers isn't almost a damn circle.

[twin dew]

Starlink launch, trying to see how many times used booster.

B1063.19, so 20th launch of the booster.

[languid gulch]

was the upper stage that failed, booster seems to have been fine

[twin dew]

Ah, yeah, sorry.

Just woke up...

[languid gulch]

there's talk of some sort of outgassing during upper stage's engine chill while it was still attached, just before MECO

and musk is claiming a RUD during engine relight

[twin dew]

Which doesn't rule out LOX leak as (partial) cause of the RUD.

So 334 consecutive successful missions, even when counting the ground test Amos 6 as the last failure.

Including Falcon Heavies.

[languid gulch]

yup

having a leak like that building up on the outside, then trying to relight, would definitely go boom

[twin dew]

Or just the engine not getting enough LOX and that causing a RUD.

[languid gulch]

or both

lox starved, it burps, then the buildup lights

[pulsar charm]

engine blew up during the second burn of stage 2, starlinks were deployed and elon said the starlink flight software was modified so their ion thrusters could go into "warp 9" (get as much thrust as possible)

spacex said they only have communications with 5 of the 20 satellites

[languid gulch]

yea i'm betting we're gonna see some interesting reentry videos

[pulsar charm]

new hardware failed instead of reused is really surprising to me

[languid gulch]

i'd like to know if the upper stage merlin engine (minus the bell) are ever reused by moving one from a booster to the upper

[pulsar charm]

probably not

[mental oriole]

Emmc test
Write data to sector - returns ok
Read data from sector - returns ok
Check read buffer equal write buffer - returns ok.
Reboots, and performs read -returns ok
Checks buffer , everything is read back as zeroes...

[twin dew]

The amount of mucking around in command line to install Debian 12...
I want BTRFS RAID1 with subvolumes and xxhash64, and Debian installer only supports BTRFS as single volume.

And xxhash isn't supported by the kernel package the installer uses, so needed to grab it from final kernel.

[soft bloom]

is raid1 that much better over simple LVM mirroring?

[twin dew]

Not using LVM in way, just pure BTRFS.

[soft bloom]

with lvm you can also add cache

i once wanted to be cool and use raid, but since discovered that LVM is jsut simpler

[twin dew]

Just another software thing, just BTRFS native and not LVM.

btrfs on partitions from two disks with native mirroring, just called RAID1 in there.
The Pi:s and firewall I have use btrfs on one disk with data duplication, everything is stored twice, so bit corruption can be noticed and corrected.

So not using MDRaid and/or LVM with some filesystem over them, but BTRFS for the whole stack.

BTRFS, like ZFS is meant to be used direct on disks/partitions, with no other layers between.

[soft bloom]

and what about resizing, caching, snapshots, encryption?..

[twin dew]

Things that BTRFS supports for all but caching, which I'm not completely sure.
And there is not going to be tiering beside RAM and SSD, and that BTRFS does at least.

https://en.wikipedia.org/wiki/Btrfs

[willow pike]

and zfs supports all of the above I believe

[twin dew]

Basically trying to create full stack GPL replacement for ZFS.
As the ZFS license is problematic.

Not there yet, but works fine for mirrored OS drive for me.

[willow pike]

it makes more intuitive sense when you take a step back and ask why volume management and file systems should even be separate

[willow pike]

a laudable goal

isn't oracle involved somehow

with zfs that is

[twin dew]

For ZFS, yes.
That was Sun Microsystems, which Oracle bought.

[willow pike]

a tale as old as time

[twin dew]

OpenZFS has been forked, but still limited by the original license from which the fork was made.

[willow pike]

and I bet oracle fight even that

[twin dew]

So Sun had open sourced ZFS under their own CDDL:
https://en.wikipedia.org/wiki/Common_Development_and_Distribution_License

[runic patio]

<@&387163995947270144> Are there specific channels for translators?

[white kraken]

Yes

[runic patio]

Can U put me in there? I did translations alread to Satisfactory

[white kraken]

I added the translation role to you, so you should be able to see a new channel for selecting the language.
Be informed that the translation project is currently on hold, as the 1.0 release is in preparation.

[runic patio]

@white kraken Thank U dude. Have a nice day!

[midnight osprey]

hey guys, is there a way, in Windows 11, to see how much power a USB port can give?

[edgy hazel]

Nah

[mental oriole]

5V

[edgy hazel]

and usb c?

[mental oriole]

From pc it's 5v

charger can be higher. But since they asking what a port from windows can give it's 5v.

[edgy hazel]

some mb have like 30W charging through their usb-c tho? is that still 5V?

[twin dew]

No, 3A max on each step, until 20V and up can also do 5A.

IIRC my MBs front panel USB-C header could do 60W when the extra 12V power is connected for it (20V at 3A).

[tough owl]

I think it's more than 5v if tb3 or better

[twin dew]

Nah, USB PD is separate from Thunderbolt.

[tough owl]

Ah

[twin dew]

USB-C port can support both, but one doesn't have to support either, or can support only one.

And then USB PD input vs output capabilities.

But the important question, WHAT IS TOLD HERE BEHIND THE PAY WALL?!:
access.redhat.com/solutions/6295491

Probably just some detection error, because there are just GSI 0-23 and then GSI 24-55, and -1 would seem to be just error.

[midnight osprey]

thanks, does that equate to 20 watts?

[twin dew]

3A @ 5V = 15W.

But normal USB 3 port can only give 900mA @ 5V, with active handshake (by spec, actual can be higher and might not need handshake).

Which is 4.5W.

[midnight osprey]

how about the 3.2?

[twin dew]

But usually those higher power ports are labeled and marketed by MB/laptop maker.

Not any higher for any of the USB3 versions.

[midnight osprey]

k

[twin dew]

And that IIRC included non-PD USB-C ports.

And USB PD cannot be supported in USB-A ports, only -C

But anyways, that coming server has: 2x 16GB Optane M10 PCIe 3.0x2 NVMe SSDs, 4x 480GB Samsung PM893 DC SATA SSD (X570 SATA ports), 2x 3TB WD Red HDD (Silverstone ECS04? / LSI SAS2308 @ IT-Mode)
Two of the SATA SSD:s are for main OS and various VMs, used 24GB per drive for the base hypervisor OS just to be safe that there is enough space in future.
Rest are for NAS with ZFS.

Currently with Ryzen Pro 4650G CPU, but want to switch it to Ryzen 5600 in future.

[night girder]

wtf is wrong with Microsoft lately?

www.ghacks.net/2024/07/11/microsoft-is-not-done-yet-more-ads-spotted-in-latest-windows-11-build/

Are they going bankrupt or something? They need more money from us for their AI shit? I don't get it. Why are they so hellbent on adding advertisement in everything they can?

[twin dew]

IIRC it was Silverstone ECS04, but anyways some 1:1 clone of the LSI SAS9217-8i, using the SAS2308 controller.
Didn't even need cross-flashing to use stock LSI IT firmware, as the maker hadn't even changed the manufacturers name etc. in the included firmware.

[twin dew]

Moar Money Better? Just extremely stupid Exec level.

[night girder]

How to run a product into the ground 101.

[twin dew]

Thinking that no matter what they do, people cannot switch away.

[night girder]

And people say Apple is guilty of a closed eco system.

Which they are, but Windows has them people as hooked as people are hooked into Apple stuff.

[twin dew]

Those PM893:s actually have full data loss protection cap set (actual data center drives), and were pretty cheap.

1 DWPD for 5 years endurance rating, so rated for 480GB of writes per day, for 5 years, for the 480GB, for total of 876TB written.

[charred pewter]

wee AT&T breach

[twin dew]

Why actually try when the fines etc. will cost way less than actually doing something?

[night girder]

Phase 1: out of control data hoarding by all companies. Phase 2: criminals realize all data is hoarded in one sweet spot for them to hack. Phase 3: profit.

Somewhere in there should have been a secuirty phase but they forgot about that 😂

[charred pewter]

bigger the corp and people in it, the exponential chance of getting engineered

[night girder]

Yeah, and it's dumb to hoard so much data. You are just a target at that point. Get greedy. Become a target.

[sharp oasis]

He hired me as a tech. And uhh. Oh god I'm gonna have a great time now

[night girder]

Too bad for the people who's data is stolen and have 0 control over it.

[charred pewter]

small businesses of like 3 people all sittin tight on the best secure shit in the world LOL

[night girder]

small companies sadly also put their data in the hands of big corpo (cloud etc). Not always, but it happens.

[twin dew]

And why try to secure, when that would cost lot of money over time.
When on a breach the affected people cannot move away, the fines are slap on the wrist and the "credit monitoring" in bulk is basically free.

[charred pewter]

aye

[twin dew]

Until US changes the laws to actually allow for fines that matter, nothing will change.

[night girder]

US change? 😂 What a joke.

[charred pewter]

just waiting for the news that OneDrive was hacked and millions of windows users files were breached

followed by GoogleDrive being hacked

topped with Apple iCloud

[twin dew]

Well, US is changing, but change into direction that would cost big companies money isn't that likely.

[night girder]

"What can go wrong if we put all that juicy information into one place. Nothing? Right? ... Right? 😅 "

[night girder]

Eh, they don't really change afaik. Maybe little chances. But they are pretty stagnant lately.

[charred pewter]

its why i still cant get behind any of those 'cloud' security storage solutions like icloud keychain, or 1password

[edgy hazel]

I just want AWS to be breached ;-;

[night girder]

Cannot blame you.

[twin dew]

Huge power grab by right-wing federalist judges has just happened for example.

From the two other branches.

[charred pewter]

my poor country 😦

[night girder]

AOC is on it 😉

[twin dew]

And then the possibility of Cult of Trump taking over the country permanently if they win.

[night girder]

He already won. Democrats fucked up big time 😄

[charred pewter]

Dictatrump

[tough owl]

If 2010 was a picture

[charred pewter]

prolly not good to get into it 😉

[night girder]

Frigidman, you are always welcome in the EU 😄

[charred pewter]

ive been eyeballing New Zealand

[sharp oasis]

I bet it hasn't been redone since then

[tough owl]

Politics in tech talk. Uh oh. I'll have to stay out 😁

[night girder]

https://tenor.com/view/abort-mission-press-button-press-it-gif-13456843

[sharp oasis]

I have to fix to tmr cus stores about to open and we need that network up even if it's temp

[night girder]

I just had to giggle when Baldur said that US had to change.

Maybe he is right, but a change backwards isn't a change to me.

Only change that matters to me, are the ones forward in the right direction.

[charred pewter]

https://tenor.com/view/jon-stewart-biden-trump-debate-2024-cannot-not-real-life-fuck-gif-11551874508584365387

[tough owl]

change can be good or bad and its definitely changing

[charred pewter]

😄

[night girder]

Is it changing when you goin backwards?

For example: you make a law that benefits people. Then you get rid of that law?

[charred pewter]

going in reverse is a change of direction yeah

[tough owl]

A left or right turn is still a turn is it not?

[night girder]

For example, is it a change if you forbid hospitals and doctors and medical care?

You going back to mediaval times, that's my point.

Technical, it's a change. But is it really a change?

You going back to a path you already walked on before.

It's not really changing. It's a dog chasing it's own tail.

The dog can turn left or right, it's not going forward 😂

[tough owl]

its still a change. Whether change for good or backwards ¯_(ツ)_/¯

[charred pewter]

Change, change never changes.

[night girder]

Yeah I guess so.

It feels stagnant.

the EU with GDPR, now that's a change.

Or AI generated laws, those are changes.

[charred pewter]

fucken backwards change

[night girder]

Electric cars. Or solar panels.

[charred pewter]

GDPR can suck muh

[night girder]

Why because it makes your job harder? 😛

[tough owl]

suck muh FREEDOM 🇺🇸🇺🇸🦅🦅🦅🦅🦅🦅🦅🦅

[charred pewter]

how do you love every website slamming you with popups about cookies lol

[night girder]

"freedumb"

[twin dew]

Freedom for the rich (including politicians) and the corporations, not the people.

[charred pewter]

omg cookies 🙀 ... screaming cookies are bad, is like ... well ... everything the right-wing screams about is bad

[night girder]

I hate it. The good thing is. EU knows this is done deliberately by companies to make users tired of the pop-ups and make em rise up against EU. So EU is making a law to forbid these practices.

[charred pewter]

lol

[night girder]

All the EU wants is a notifcation to users that cookies are used. It's the companies who made this whole pop-up mess to annoy EU and the users.

[charred pewter]

the EU law REQUIRED those to show up

[night girder]

nop.

Not how they are implemented now.

[charred pewter]

how else is a website gonna disclose to the user that some cookies are required, and others are not, and for a user to continue they have to make a choice

[night girder]

www.techspot.com/news/101396-european-union-set-revise-cookie-law-admits-cookie.html

[charred pewter]

i mean, the EU law had to foresee thats the only way, is with annoying prompts

[twin dew]

Notification for the mandatory for functionality, and rest as opt-ins.
But companies didn't like the opt-in change part to be not shown automatically.

[night girder]

These banners are displayed to adhere to the GDPR privacy regulations implemented by the EU in 2018. Reynders mentioned that the commission aims to alleviate "cookie fatigue" while ensuring users comprehend online advertising mechanics and make informed decisions about their data. One proposed solution is to require websites to remember visitors' preferences, thereby presenting the consent form only once per year.

The commission aspires for large platforms like Meta, X, and others to voluntarily commit to a "cookies pledge." This pledge would involve providing transparent information about cookie usage and avoiding user annoyance. It is hoped that smaller websites will emulate this approach.

[twin dew]

Because they want to force them to be opt-out, not opt-in.

[night girder]

It's not a perfect system, but I have to say, EU had the balls to do something drastic about it.

[charred pewter]

i still dont see how you can inform people about cookies, without informing them about cookies

[night girder]

and now they are revising the system. Which I can only applaud.

[charred pewter]

if a cookie is required ... tough shit, proceed or gtfo

[night girder]

It doesn't always work. I disabled cookies on my firefox browser. Some websites get in a loop asking for my consent. Others do it better.

[charred pewter]

and most session cookies are required unless the website is just a basic article site with no user pressence

[tough owl]

cookie popups are the most annoying thing.

Especially on mobile

[twin dew]

At least most now actually have the mandatory "Deny all" button (hidden) near the "Allow All" one.

[tough owl]

Good greif viewing the modern web on mobile is impossible without adblock and reader mode

[night girder]

And you have addons that click em away automatically.

[charred pewter]

yeah, we punt anyone who wont allow session cookies. plain and simple, the security holes in NON-cookie sessions are not worth even adding those methods. putting a session ID into every single url to maintain session state, is just bad news for that user

[tough owl]

Which should not be the solution but it's the best we got

[twin dew]

Doesn't affect the "mandatory for the site to function" cookies.
Only all the extras for tracking etc.

[night girder]

Agree. But better than to let big tech do whatever the fuck they want.

If you want to blame someone, blame the big Tech companies like google.

[charred pewter]

chrome isn't widespread adopted enough to use its internal data storage object (yet? doubt it will happen)

[night girder]

While they do play innocent now and come with "solutions" and a "cookie free browser" it's because of them we are in this shit.

[tough owl]

I hate Google with every fiber of my being. (I own a pixel)

[night girder]

www.businessinsider.com/googles-plan-to-replace-tracking-cookies-faces-big-new-hurdles-2024-7

[charred pewter]

i laugh when googles own cookies are flagged by my chrome as being blocked 😄

"no cross-site cookies allowed" hahaha

[twin dew]

Because that Googles "solution" is even worse.

[charred pewter]

that data storage thing is a mess

[night girder]

To be sure, the Privacy Sandbox isn't the only alternative to the advertising ecosystem once cookies are discontinued. Targeting using third-party cookies on the Safari and Firefox browsers hasn't been available for years, and several companies have come up with alternatives like adopting new identifiers or increasing their use of first-party data.

This is the real issue and worry for them.

Their precious advertising ecosystem. It must survive.

[charred pewter]

right

[night girder]

So that's the sad part, EU can make laws all they want. Big tech will do whatever is needed to keep this ecosystem alive. And that's with our data.

[charred pewter]

all that tracking shit

like when i say something in here, and then suddenly my reddit is showing me articles about the same damned thing 😉

[night girder]

But then google and co would argue the data is needed to keep the internet free.

[charred pewter]

HAHA

"Google Search Subscription" when

[tough owl]

I'm getting to the point where all Google Domains will be blocked on my network besides the ones needed for invidious.io/

next year

I'll place that bet

[charred pewter]

i already dropped their google-extended shit

[night girder]

but what if the google domains aren't that obvious?

[charred pewter]

wasnt following the rules of robots.txt to not grab every damned image on the site

[tough owl]

I'm sure there's some gist on github with them all

[night girder]

Then google just makes more that are not on that list.

[charred pewter]

then apple came in with applebot, then msn came in with gptbot ... and fuck all these ai bots

[night girder]

This solution doesn't seems viable.

[tough owl]

its more of a suggestion at this point

[charred pewter]

yup 😦

[night girder]

It's so original to make AI chatbots. Innovation at it's best.

[tough owl]

And it'll be updated maybe possibly

[twin dew]

Well, services USED to be run as a interconnected web where the runner paid for it, and usually the runner was the users ISP, so they got paid.
Then the "free for user" megaservices that get paid by selling the users data and having advertising took over (like Discord etc.).

[charred pewter]

this morning i just punted ImagesiftBot cause it was scrapping every single image it could find, multiple times even ...

[tough owl]

discord my favorite data broker

[night girder]

I sell all my data on discord.

[charred pewter]

I 'let' all my data be sold by discord

[night girder]

www.msn.com/en-us/news/other/your-discord-data-is-being-sold-to-law-enforcement-and-ai-companies/ar-AA1nkgqn

You talking about that?

[tough owl]

yessir

[charred pewter]

hehe

[tough owl]

Its lovely

[charred pewter]

hay discord ... 🖕

[night girder]

Oh, yeah I posted it before. Spy Pet.

"scraping over 10,000 servers throughout Discord" 😂

[tough owl]

I want all my friends to switch to Element or something similar

[charred pewter]

law enforcement trying to find the cult discords