#off-topic-tech
1 messages ยท Page 117 of 1
"The trees will actually grow in Fable 2!"
My task manager is finnish but you can see 2 threads idling while all others are ~100%, still HWinfo says its 151% cpu utility ๐
Probably counting each 100% used P-core as 200%.
Because of something getting triggered in the code because of the "missing" HT support.
Ah, sorry, older, still with HT.
My bad
I see 2 of the E-cores are listed as LP = Low Power? Guess they are the ones idling
Yeah, that CPU line has 2 extra low power cores in the IO-die.
So that most basic stuff can be done without ever waking the core die at all.
Sorry, SoC tile in Intel speak.
Compute tile with most of the cores.
Graphics tile with most of the IGP.
SoC tile with those two LP cores, small part of most critical IGP stuff, memory controller, most critical IO.
I/O extender tile, with rest of the IO.
Made so that in idle, only that SoC tile needs power to handle basic stuff, including refreshing the screen.
With Graphics tile only getting wakened every so often, and Compute tile if something needs more computing power than those two LP cores.
hehe, Sunita Williams en Barry Wilmore are still stuck on ISS ๐คฃ
8 day mission turning into a few months. Don't panick. Don't panick.
Boeings fuckups just keep on giving.
As SpaceX was supposed to only fly one Crew Dragon mission to ISS each year, since many years ago.
So the older capsules are starting to hit their original flight number limits etc.
Which is a limitation imposed partially because when the contracts were made, nasa gave spacex money but the assumption internally was that boeing would be the reliable one and spacex would have issues, and that's what the actual plan was built around
Lot of pressure back then to only select Boeing.
I mean, it was the "safe" choice so I don't fault NASA for it, since probably nobody outside of Boeing itself could predict the nuclear amount of bedshitting that company has done lately
Somebody spent 6 months on that icon, be nice
nah what sucks are the icons within task manager
the fuck do the icons on the left mean
Basically nothing until you expand the names too and learn them.
First icon is a broken tetris block
Second icon is a graph view
Third icon is a clock view, dunno why task manager needs a clock
Fourth is a fuel/pressure gauge, view, dunno why task manager now needs fuel
Fifth is a people page
Sixth is a process list page, by the looks of it
Seventh is a puzzle ...
And at the bottom you have a gear.
And what would you say just seeing these would say?
Me?
good thing there is text next to it 
So has Windows when you expand the pane:
yeah I don't get y'alls issue anyways
No, usually icons don't make sense until you learn them.
Same is true for those Task Manager ones.
what do you mean "No"? I said I don't get the issue and you just say "No" like I do get it

Sorry, wasn't for you, was for whyzenman
Iconography is language too.
With very limited "common" vocabulary.
And even that has lot of regional differences.
And when you go into the unknown, there just aren't established icons for the concepts.
Fuel gauge
Tetris view
Graph view
Media player view.
Add view button
YT now tries to play ads even when it's embed into Telegram
That doesn't seem legal
Why wouldnt it be?
Embed is just fancier linking, the stuff is still serviced direct by YouTube.
And youtube is still paying for hosting and delivery on those videos
Yup
revanced / ublock, the only way
When it's Telegram's browser?..
you weep
First of all because it's broken: video is loading while the audio from ad plays without any means to stop, skip etc
As far as places to be stuck, that one's not too bad

Boeing secretly keeping people in orbit to spot more UAP
when someone coded seconds to go up to 60

Someone forgot about leap seconds
They just accounted for this falsehood.
https://gist.github.com/timvisee/fcda9bbdff88d45cc9061606b4b923ca
- There are 60 seconds in every minute.
I am just wondering whats going throught their heads.
They are trained for unforseen circumstances. But if you flight back to earth keeps getting delayed...
they will be fine. They send up extra food and stuff to keep em alive.
The food isn't fast enough to get there in time ๐
Valeri Polyakov spend 437 days consecutive on Mir space station.
How they didn't go crazy. It's so small. Up is down. Down is up. Left is right and right is left.
Are they?
Only logical explanation I can think of is that, if it takes 1 million years to go from A to B by normal means of travel, and you do it with a wormhole in a matter of seconds. You skipped time (1 million years).
Also, a hypothetical wormhole, you go faster than light, which is time travel?
Wormholes might allow effective superluminal (faster-than-light) travel by ensuring that the speed of light is not exceeded locally at any time.
Yes, because space bends.
Wormhole connects two points in space. It's like a "shortcut" of what I understand of it.
Also, it's all hypotheticals and theories.
that's still faster than light travel, if it takes light minutes to get to Mars and you take a wormhole to get there instantly, you have travelled through time; the speed of light is the cosmic speed limit of casualty itself
it just happens to be how fast light goes
"travel by ensuring that the speed of light is** not exceeded locally at any time.**"
it still fucks with time
you're not doing the tachyon thing by literally travelling faster than light but you are fucking with casuality
now you talk how gravity effects time.
But this picture explains it good. To get from point 1 to 2. You either have to follow pink line, normal space travel. Or wormhole (yellow line) which is shorter.
yes i know what a wormhole is and how they do
Sorry, was meant for maxeek.
ah gotcha
one thing i liked about Interstellar is, when they realised they were looking at a wormhole, everyone knew unambiguously it was put there by aliens
a naturally occurring wormhole is like a naturally occurring iPad
@inner shale If you want to become 1 million years old, go find a strong gravitational field
oh no...
oh no
wait, you will not become 1 million years old. You will become 1 million years old for earth time.
The crazy part is how it also effects decay of atoms etc.
I don't believe. It's either real or not. And the proof is not there. Only theories.
prehistoric man was travelling in boats; we have absolutely huge boats now. birds show us that flight is possible; now we fly package holidays. we've always seen things fly about in space; now we too fly in space. but we've never seen anything go faster than light
the Alcubierre drive (warp drive) is what you get when you rearrange Einstein's equations to ask "what configuration of mass can go faster than light?" and the answer involves negative mass, which nobody has any proof for, and is more likely just the equations telling us to go fuck ourselves
yeah warp drives have the same casualty fuckery
Or Einstein made a mistake ๐
besides, any planet we travel to ain't gonna be hospitable, it's a much better idea to start building O'Neill cylinder space habitats
it's not impossible but there's a shitload of science we would need to re-invent if we were to come up with new ideas
Depends.
However, he did make some mistakes and had doubts about certain implications of his theories, indicating that while many aspects are accurate, not all of his ideas are without error.
alas, seems impossible for billionaires
Even Einstein was like; "euh ... take this with a grain of salt" ๐คฃ
yeah, yet a lotta people have been attacking and reinforcing and fucking with general relativity and his shit still holds
he's a dragon who wants a bigger hoard
True true, I am just saying, it's not because he is wrong, that his theorie of relativity is wrong. It can be both. He could be wrong about different theories. Or made a miscalculation. Or wrong conclusions. Misinterpretation of the data.
yeah, the same went for Newton
But his theories of general relativity can still be solid.
his theories held true and were bulletproof, everyone thought we had solved science and just needed better equipment, right up until we couldn't explain the orbit of mercury
then whole of science exploded
it would take a discovery that big to blow up Einstein
Why do you think scientists are still probbing at it ๐
gotta keep working
We still don't know what dark matter really is.
we know it exists! we've seen galaxies that don't got it
How it effects space etc. We have mathematical calculations for it.
but that's about it
Yup... that's not much. So who knows.
weakly interacting particles, micro black holes, mass effect element zero
fuck knows
Wait how is it possible for galaxies to not have dark matter? Dark matter is assumed the driving force of the spreading off the universe.
we saw a galaxy without it that probably lost it in a collision
Pushing galaxies away from eachother. One day, there is nothing out there anymore that we can perceive.
it's much more diffuse than other galaxies
that's dark energy
Got them mixed up hehe.
Dark matter and dark energy are two mysterious components of the universe that have different roles. Dark matter, which makes up about 27% of the universe, acts like an attractive force that helps hold galaxies together, while dark energy, accounting for roughly 68%, is a repulsive force that drives the accelerated expansion of the universe.
Shit, that's dope.
we at least know that dark matter is a physical object, not a force
dark energy is weirder
one that we just don't have the tools to measure it/detect it.
There is an interesting video on YT about what will happen if it keeps expanding. Some theories even believe the expansion will stop. Some believe it will even reverse back into one big mass. Ready for another big bang.
NGC 1277 is a lenticular galaxy in the constellation of Perseus. It is a member of the Perseus Cluster of galaxies and is located approximately 73 Mpc (megaparsecs) or 220 million light-years from the Milky Way. It has an apparent magnitude of about 14.7. It was discovered on December 4, 1875 by Lawrence Parsons, 4th Earl of Rosse.
NGC 1277 ha...
big crunch, heat death, big rip
you made the video?
Oh, it contains little dark matter. ok
no i am a nerd
but yeah if a galaxy can get rid of dark matter, that implies it's something physical, not another force or form of energy or a misunderstanding of gravity
Well the wiki says: that it *suggests * it has very little dark matter. So that's sound unsure.
for a while there people had modified gravity formulas to explain it
a lot of this is unsure, nothing is ever proven, only gets more evidence
Even the article source states:
However, scientists are not satisfied with any of these assumptions. In their opinion, a galaxy with almost no dark matter could not have formed. Therefore, they want to test their observations using the WEAVE instrument on the William Herschel Telescope located in the Canary Islands.
Now I am curious if they did more tests ๐
But scientists have been wrong a few times. Just recently even.
I watched this video a while back https://www.youtube.com/watch?v=UyzBoUvN3PM&themeRefresh=1
A Supercut Astrum Episode Covering the Voyager Mission. A huge thanks to our Patreons who help make these videos possible. Sign-up here: https://bit.ly/4aiJZNF
Astrum Podcast: https://www.buzzsprout.com/2250635/share
Displate Posters: https://astrumspace.info/Displates
Astrum Merch! https://astrum-shop.fourthwall.com/
Join us on the Astrum disc...
And he explains how scientist were wrong about the model of the Heliospheric current sheet.
ye
when did this become wrong ?
Months have either 28, 29, 30, or 31 days.
this is funny
The software will never run on a space ship that is orbiting a black hole.
It's a good laugh from time to time ๐
I don't think it's "nerdy" to get into science ๐
these days its pretty cool ๐
maybe, but they still seem to be scared
true, but it sounds derogatory
Wow, wikipedia is brutal:
A nerd is a person seen as overly intellectual, obsessive, introverted, or lacking social skills
well it's almost christmas
I have some path of thought which lead in the "misunderstanding of gravity" field of things as conclusions
first of all it's sad but academic != scientific
scientific level is very disparate
second, these analysis are often wet finger guesses trying to join multiple data set in a context they never have been produced for / validated for
often joining data with different scopes and unmatched filtering
which lead to absurd interpretation of these dataset even were they are themselves perfectly valid
and a number of miss conception about the scales of things talked about
and the huge impact these scales have
what we observe in james web and similar system isn't shaped by matter
but is shaped by space itself
try to apply glass ball crystal behavior shit theorem and concepts at these scales and you are COMPLETELY off your shoes
for instance
stellar system space isn't comparable to interstellar space data
and interstellar isn't comparable to galaxy space
and galaxy space isn't comparable to intergalactic space
here we are talking about galactic and intergalactic
you can let your shoes in your drawer
here we are all barefoot in the jungle
September 1752 has 19 days in the UK
If you want to step off the Julian/Gregorian calendar you could think of the Sansculottides of the French revolutionary calendar
Which is so much better than whatever the crap we use in current time
How so?
I hate 24 hour time
Point of calendar is to represent natural cycles
Division by 12 is clever
Machines will always count time differently though
Yesterday i heard that without sun people's sleep cycle resets to 40 hours.
It's frome mad vid about descending 1700m below the surface into the cave
I mean
That's pretty deep
I always wondered why each day i want to get to sleep 1 hour later. Is it all 5g towers from Elon preparing me for mission to Mars?..
Btw, Sabine mentioned it in recent vid
On one hand, I want to say "go ahead, we should push tech and expand, unite scientific potential and communities*.
On the other hand, I want to see any living thing to dare to go outside of earth magnetosphere. The claims are just... Unjustified.
Moreover, NASA has planned Mars mission, yet didn't do shit.
Sending humans after robots? In the past we were smart enough to try on smaller animals first. And were right about that. We don't know what it would be like for humans, or any living thing, to live outside of magnetic field - something what isn't to protect against but to maintain.
I'm all in for universal time, but at least keep the 12th
i like the idea of a 13 month calendar
Leap seconds are a thing, that 23:59:60 is valid time, the parser is wrong in this one.
And that was already talked about..
And that 1972-06-30 23:59:60 was the first leap second ever.
Plan is to phase those out.
All the G.E.T parts now 3D scanned and added to bucket model. Bad timing for asking prices for the steel parts, all suppliers allready heading for christmas holidays.. Anyway for a $500 chinese equipment the 3D Makerpro Lynx scanner has worked well for me, good results with little time put in to first learning step. Reduces manual fitting/adjusting work for 1st one to ne made.
how big is it? im guessing fairly big
Nominal volume is just little below 3000 liter or 3 cubic meter. Weights 3,5 tons. Inside width 1650mm. For a Doosan 530 excavator which is 52t
ok, so I guess i found a bug in Pandas. dam
Or just in one of the the libraries you are using to talk to it.
A leap second is a one-second adjustment that is occasionally applied to Coordinated Universal Time (UTC), to accommodate the difference between precise time (International Atomic Time (TAI), as measured by atomic clocks) and imprecise observed solar time (UT1), which varies due to irregularities and long-term slowdown in the Earth's rotation. T...
that's an interesting choice, since clocks can't display 60s second (even majority of digital ones),
and things like summer time savings don't addup 25th hour - they just roll back 1 hour, staying withing same range
I knew that we have leap seconds. i never thought they go over the usual range
That intended replacement (leap minute, in form of making one minute actually take two minutes) is even weirder.
Basically punting the problem about 100 years into future.
what's funny their formatter is aware of leaps
Seems the bug is in DateUtil library.
Or just datetime module of Python.
Unlike the time module, the datetime module does not support leap seconds.
it's teleporting from inside but it takes much more time from outside
relativity applies inside of it
it's just distorted
for you it's seconds
for outside it's months
from outside it's millions years
but you "teleported" few months in the futur
Sounds like you have funโข๏ธ with time KEKw
you probably have to cap leaps seconds to register a date
it's not for no reasons computer use cumulative seconds
if you want to go from cumulative seconds to another cumulative unit, do not convert to date in-between
But basically problem is that most software or people doesn't care about the leap seconds.
But that kind of seismic data does.
hehe
And whatever library that DEFAULTPARSER points to is the original source.
I want to bet it's datetime
Yup: from dateutil.parser import DEFAULTPARSER
Woohoo, I love being correct!
Honestly, I'm more wondering if the leap second is actually important for what you're doing here
well, honestly, i don't care about it in particular. but I need equal time intervals
Are the entries in your dataset time-variant?
that's what I was pointing
Because if they're time-invariant and just need chronology I'd just use t
^
Or use unix-timestamps
Yeah, I probably should have stored time in simpler format
I know enough about time fuckery to know you want to deal with it the least amount humanly possible
Which is probably why datetime just pretends leap seconds don't exist, because unless your application explicitly requires it it's better to pretend it doesn't exist at all because it greatly generalizes things
Doesn't work either:
In Unix time, every day contains exactly 86400 seconds. Each leap second uses the timestamp of a second that immediately precedes or follows it.
Unix time numbers are repeated in the second immediately following a positive leap second.
time strings were generated by different lib altogether
then Kaggle turned out to be naughty service that doesn't let me download big files from kernel...
so i decided to cut it by time, so i wanted to do it nicely (well, yeah, i could have found year and select by it)
and now it will look like pd.DatetimeIndex is the superior method of parsing array of date+time strings, outpathing pd.Timestamp and pd.to_datetime in both speed, functionality and correctness.
who would have known? (well, not yet confirmed...)
But just using TAI instead of UTC would fix the problems.
Leap seconds are way to keep UTC close to both TAI and UT1
I'd just pretend leap seconds don't exist if I don't have to
Unfortunately when it is historical data that was originally handled by including leap seconds, stuff gets hard.
If you don't include them, any data around the leap seconds if corrupted by the conversion.
well no it wasn't and uses the same parse routine
can't you have dynamic chunks ?
like having files storing multiple seconds ?
why cutting by seconds in the first place ?
then leap seconds data would naturally be included
let say you ask for :59 data
you'll get the next 3 seconds before 1:0
positions_df["year"] = positions_df["time"].map(lambda ts: int(ts.split("-")[0]))
greatest fix of all
I feel like Kaggle is slowly degrading as I use. functionality just breaks liek if it wasn't run by software but on some sort of physical maachine that gets wrose the mor you use it, slowly destroying itself...
now even Ctrl+/ doesn't comment lines, yesterday environment broke and can't find conda, some time before cell buttons disappeared...
wdym
well i don't mind time slowly sliding from HH:00:00
And there aren't even many places when i need to interact with it. for visualization and analysis I could just index it and forget about exact date until that point in time is interesting
make sure you're not loosing the original pace and reference time then
So one can still seek for that then
So I would say avoid sliding as it create aliases
ok so they replied no more conda... a bit sad, but ok
bad there wasn't any warning
Dont use steamos
Its designed for steam deck, not desktop
If arch is too hard to setup, use EndeavourOS. Its just arch but with a nice installer
lmao, and now Plotly plots are dispalying in notebook editor.
new surprise every day...
steamos is getting a general release soon
steamos uses an immutable file system so the only software you can install is flatpaks, or steam games
if you disable this and install regular linux programs these get nuked on update
Exactly
but you do have to disable this manually
Use endeavouros
does it come with gamescope?
if it's literally just arch with a better installer i wouldn't go recommending it to anyone who isn't already a linux head
the steam deck has helped with getting stuff available as flatpaks anyway
Community has gotten around it thankfully
that's the idea
instead of windows 24H2 having to tiptoe around every little changed file you've accumulated over the years, giving huge accommodations to antiviruses, you just swap out the whole-ass partition; you also get to easily checksum the entire thing
Because those "retail" OEM keys are for single computer.
And to allow it to be usable in another MB, you need to call MS.
And lie your MB died and had to be replaced.
https://blogs.igalia.com/berto/tag/steamos/ many ways to install software on steamos while keeping it immutable
huh. that link wasn't removed
Only the full Retail keys are transferable, and even those have limits on how many activations can be done (maybe only in specific time frame).
Bot might be broken again.
as people said I would recommend endeavourOs
because to switch to Linux you need a working system
and arch Linux alone does not gives you that but provides you with a way to build how you want it to be
the catch is :
you can't know how you want it to be if you can't experience things in the first place
so you need a feature full barebone at least
endevouros gives you that
while still keeping the archlinux flexibility afterward
so imo it's the best to start your archlinux journey
this feels less like "I want a windows alternative" and more like "I want a new hobby"
btw I have my own projects of distribution
and the actual scheme of "archlinux and it's open derivatives" is by far the best scheme imo
I want to create a similar scheme and enhance it
there should be no choice to make about "working yet screwed vs open and lost"
having "template and working built on open and flexible" will always be the stable way
expert will build structure
artist will build user end templates and maintain it
and everyone will be happy
like how I have an espresso machine and have spent ten years getting the most out of it and can now extract the best out of my beans, but some people just want a caffeinated drink
i can't tell everyone to go spend ยฃ1000 on espresso gear and go learn it if they don't wanna
it used to be the way though
with multi generation working gears
and archlinux is not ยฃ1000
if you stay on base configs and correctly backup them
arch Linux is even sometimes considered "boring"
it become a hobby if after having a working system you still keep tinkering things
my point is it can stay a hobby of a month
and then you just run the boat only to go where you want
imo the main reason not much people switch to that kind of stuff
is not everyone have room and/or money to run a spare system
and you basically need a spare system or two as starting point to safely get your hand on it
slowly shifting which machine is your "hobby" machine , which is your backup, and which is your daily
and eventually the windows one will become your backup
then you will clean it up for Linux
and your old Linux machine will be your backup
that's when you would have made the final move
Fedora seems to be good as just windows alternative
I just like arch's repos
omg 6.12.6 blinked back in
lets compare the two
ok no change, probably a cdn update issue then
except when you must follow stupid scheme that doesnt match real endpoint locations and you dont have keys or permissions to maintenance these
The issue with leap seconds is even funnier when i realize that it is just 1 single string in millions, because every next one, except for minutes with leap second, will be no more than 59, and second leap second with be 60, and after that at most 58
Higher resolution version:
Original source:
https://community.intel.com/t5/Blogs/Tech-Innovation/Client/Field-Update-1-of-2-Intel-Core-Ultra-200S-Series-Performance/post/1650490
Intelยฎ Coreโข Ultra 200S Processors (codename Arrow Lake-S) launched at the end of October with significant advancements for multi-threading performance, performance-per-watt, platform I/O, and overclocking. Editorial conclusions on gaming performance were more polarized, with noteworthy statistical ...
Yup read that too. Makes sense.
wtf is this crap, and why did my default wallpaper change?
wtf did they do to teams ๐คฃ
Looks at the old tp-link routers in the back collecting dust
Although from what I understand, the problem is that many people keep default password on tp-link and not because there is some hidden backdoor implemented by the CCP
Current TP-Link consumer stuff now also has mandatory cloud access connection.
Active even if you don't create an account.
And you cannot disable it without creating an account.
Yeah, a bunch of propaganda.
Guess I'll grab the hammer then 
Oh wait, nvm, I have old routers which don't have cloud access
I don't use tp-link anymore btw
Yeah, just very common (because cheap), with bad security hygiene in the firmware programming, either by design or just by doing it on minimal money.
someone gave a good tip; buy the hardware, wipe the software.
But that mandatory cloud remote access is too much for me to be able to recommend them for anyone anymore.
Yeah, fair. But that's not why the US is investigating it. It's just a witch hunt / hunch. With no proof (so far).
I use asus router + Ubiquiti U6+ atm (plan on replacing asus with Cloud Gateway Max)
All of which is from this.
Question is if that bad security hygiene is by design, or just from the cost cutting.
Another unsubstantiated propoganda story. The microsoft blog is about a botnet of a few thousand hand-hacked routers of different brands, most of them from TP-Link (because market leader), and all old models that have been EOL for years. They have also had a firmware update to close the leak. The wall street journal article suggests that these are new routers that are now the best-selling model on amazon's site, that is not based on anything. This has nothing to do with TP-Link, and is simply American propoganda for their anti-China policy. It is special that the American government now also uses tweakers as a puppet.
ZyXEL routers are even worse.
ZyXEL has constant security fault notifications about their routers.
And EOLs them very fast after releasing on market, stopping updates.
And often still keeps selling the EOLd models for years after firmware updates have stopped...
Getting the notifications for all ZyXEL products as I have subscribed to get notifications for the switches I have.
China opposes the US's overstretching the concept of national security and its discriminatory practices targeting particular countries and companies, said Lin Jian, spokesperson from China's Foreign Ministry, on Thursday, noting that China will take resolute measures to firmly safeguard the legitimate and lawful rights and interests of Chinese companies.
Look, in the end, we costumer are entangled in a economic war.
The remarks came in response to a question asking for comments on a report by The Wall Street Journal stating that the US weighs a ban on China's TP-Link Technology Co over national security concerns.
Apparantly is Chinese. If I read it correctly, "China's TP-Link Technology Co".
Thing is, TP link states their products are conform with security requirements.
Hikvision, Dahua, Huawei, TP-Link, I wonder who is next in the line....
They have a billion CVEs compared to others ๐
And to be fair, the response wasn't so bad:
"We welcome any opportunity to work with the U.S. government to demonstrate that our security practices are fully consistent with industry security standards and to demonstrate our continued commitment to the U.S. market, consumers, and addressing national security risks"
Really?
The mission of the CVEยฎ Program is to identify, define, and catalog publicly disclosed cybersecurity vulnerabilities.
CVE: TP-LINK There are **403 ** CVE Records that match your search.
CVE: NETGEAR There are **1256 **CVE Records that match your search.
And netgear is American ๐คฃ

ZyXEL only had 270 results
I don't think the cve website is the right way ๐ Because Netgear has other products too. And TP-link maybe too.
Latest TP-Link CVE I found:
A vulnerability was found in TP-LINK TL-WR841ND up to 20240920. It has been rated as critical. Affected by this issue is some unknown functionality of the file /userRpm/popupSiteSurveyRpm.htm. The manipulation of the argument ssid leads to stack-based buffer overflow. The attack may be launched remotely. The exploit has been disclosed to the public and may be used. **The vendor was contacted early about this disclosure but did not respond in any way. **
I was... wow that sounds bad. But then I found latest netgear CVE:
A vulnerability classified as problematic has been found in Netgear WN604 up to 20240719. Affected is an unknown function of the file siteSurvey.php. The manipulation leads to direct request. It is possible to launch the attack remotely. The exploit has been disclosed to the public and may be used. The identifier of this vulnerability is VDB-272556.__ **NOTE: The vendor was contacted early about this disclosure but did not respond in any way. **__
So basically... we're fucked 
Yup. Pot calls the kettle black.
So what is the best way to go about routers?
To reiterate; buy hardware, wipe softwre.
honestly, it's the best thing I read ๐
OpenWrt can be installed on routers as firmware and is open source.
Aruba, Barracuda, Belkin, Cisco, Huawei, and Linksys brands. And it's based on Linux.
But do we trust opensource as router firmware is the question.
And I can imagine, if they really want, they can still build in security holes even if you wipe firmware.
Would have to security audit the firmware first.
Also true.
And yes, hardware security holes too..
Well, time to make my own router ๐
How hard can it be 
And most of the open source firmwares need to use the binary blobs for the "extra" chips etc. from the original release.
Just modify a rasberry pi ๐
Which are often firmwares for subparts or sub-chips in their own right.
Now I am curious if anyone made a raspberry pi into a router.
Tutorial on how to turn pi into router ๐คฃ
Assuming it's a well-set up OS project, yes, more than a closed-source application
maybe I need more sleep
tehcnically, how it's called?
This is just a burrito
But if it were deep fried it would be an enchilada
And it has enchilada sauce
i can't imagine how to it eat.
lots of sauce calls for something to hold it with
but i mean, burito with a fork and knife?..
Yeah, fork
that looks like something that'd give me heartburn for 3 days, & 100% worth it
i could drink that sauce all day
one of my work friends' mom would make tamales a couple times a year for the place, about 50 at a time
Heh, that's awesome
cve.mitre.org/cgi-bin/cvekey.cgi?keyword=ubiquiti 47 for unifi
amd is where it's at now for cpu's anyway. they've been killing it since ryzen first launched more or less.. now intel up to 12th was good... if you build a system with a 13th or 14th gen or their newer one.. sucks to suck i guess. xD
I'm sure they'll come back.. we need them to because the market needs at least two players. otherwise you end up with nivida..
Because AMD doesn't seem to have ever taken on the design role in mobile that Intel has, I think laptops are still a strong Intel bastion. As long as Intel can afford to keep doing a huge chunk of the laptop design for OEMs, and AMD doesn't (or Nvidia with the rumored APU), they can hopefully stay afloat.
Nvidia is probably a pretty big threat to laptop market share. Nvidia stickers sell.
I'm not sure what's really going on with that on mobile.. the amd apu's are really good, but you don't see them in many laptops. I'm not sure how amd's apu graphics compare against intel's now.
For most consumer purposes, laptops are so powerful now that it just doesn't matter.
But I can't tell you how many people have bought new laptops and told me "I got the i7 with Nvidia!" with excitement, when they have no use case for high performance.
for most normal workloads yeah.. ive got one with a amd 3550u/1050 and an intel 10th gen with intergrated graphics... my amd laptop is showing it's age as the last time I played satisfactory on it it wasen't that great
if i ever need a laptop with any kind of power, i'll let work pay for it
there's so much deceptive bullshit around laptop power specifically, & throttling or outright lying about specs
i still don't know why there has never been a massive lawsuit about calling a laptop's GPU the same as a desktop model, when there's 0 way to get the full power of it out of the laptop
Definitely not gonna happen in the next 4 years
and half the time its like a 3050 or 1650ti or other low end junk that is almost matched by a 980m
"8gb is plenty" out of nvidia now officially sounds like "gamers don't need more than 4 cores" out of intel ~2019
2017 was 8 years ago btw
not yet...
any good steam sales?
Nothing caught my eye
On subject of weird computer buys:
https://fi.pcpartpicker.com/list/MHNVwY
As long term office type computer for nonprofit, as the specific one computer cannot run with Windows 11 on officially unsupported hardware, with chance that any automatic update could make OS stop working.
And I know many of the components are "overpriced". but that PSU for example was selected because of the 12 year warranty, when the "Good & cheapish" options at half the price would have had only 5 years, and so on.
Actually only 500GB SSD, but site didn't support that variant (cheapest non-HBM usable M.2 NVMe drive I found from the specific retailer the buy was from).
And the RAM is 5200MT/s subvariant, again not supported by the site.
And possibly need to buy USB controller card later for the cases USB-C front panel header.
Currently those are too expensive and not needed yet.
Now for everyone of you to flame for the idiocy of my recommendations!
Part List - AMD Ryzen 5 8500G, Fractal Design Define 7 Mini MicroATX Mini Tower
Total price was IIRC about 625e without Windows, legit Windows 11 Pro separately from official source for cheapish subsidized Windows lisences for non-profits.
And yes, those are the slower in reality 8Gb x16 chips, but going to 32GB of RAM wasn't warranted in this case.
Yes.
I need to be flamed for my choices!
wait, what do you need it for ?
Even Satisfactory on Ultra textures now seems to need more than 8GB of VRAM.
And several new games basically need 8GB at lowest.
Time when you cannot run some games with "just" 8GB of VRAM is rapidly approaching.
Even at lowest normal settings.
laptop owners gonna be sad
i dont think nvidia made any with more than 8gb until ada
nah ram is still the choke on laptops but yea i am running into issues with 12GB now but obviously i could lower settings so its not the end of the world
so i could defedently see 8GB being a problem on newer games
did the visual fidelity increase even a bit?
Using same texture for multiple locations is basically forgotten as tech, for the biggest reason for texture VRAM need increases.
And then stuff like RT need significant amounts of VRAM for themselves etc.
Claiming 4GB is enough for gaming in 2024 is stupid.
oh, perhaps very simialr thing is what CroTeam meantioned in one of their talks:
they did also use marketplace for assets, but then it became technical debt to unify materials used in them.
Well, they did it anyway (fix the materials). So yeah, just got to get the hands out of the ass and then it works without bagilion bytes of VRAM...
well, I won't be denying reality if majority of games need more.
question is - do they need it because they increase visual fidelity at leat 30% (yeah, i know, that's stupid percentage) (not even talking about 1:1 proprtional increase), or devs just use more because they can (won't be talking about that conspiracy that devs cooperate with GPU manufactures to keep rising VRAM demands...)
Either you get players nitpicking when they notice same texture being used in multiple places.
Or you get players complaining about excessive VRAM usage when "common" textures aren't shared and are each individual...
we have only like 100 something atoms, of which about a dozen is common. nobody complained yet.
so i really don't see why anyone should care
When you do a village, do you texture all the house walls with several shared textures (for example 10 sets?), or make each one have their own.
If you share, significant amount of people will complain about that today.
Even if only one building with that specific texture set is visible at any time.
I think those people didn't touch grass
Or "repeating" concrete textures or like.
Or textures that get pixely when you stick your camera right into them... etc.
heh, TTP2 is at the level when you stick your camera into the wall and you see that it doesn't even seem like texture. nanite doing it's work (or that's like really deep texture... not sure)
honestly their demo was pretty good
It's both probably. Pushing the engine, less optimization.
i wonder how easy it is for companies to launch discreditation campaigns against games that have visual flaws, even they are nitpicky
you know, simply to raise demand or damage morale of competitors
Playing enshrouded, yeah about that 4GB ...
.
Doesn't tell the actual usage.
Would need to have the system lie to the game about how much VRAM is available until the FPS starts to drop.
As well coded game will use extra VRAM as cache, unloading stuff only once everything is full and it needs something new, only then dropping oldest unused stuff out.
Which is why those HUBs "This setting needs this much VRAM" charts are completely useless.
As they use same very high VRAM card for all the detail levels.
Black screen simulator
At night the game should only take up like 50kb of VRAM tbh
but what about 4k UI elements?
I assume they are vector???
Well, you want to be absolutely certain that everything is pure 0,0,0, and not for example just 0,1,0!
Trust me. The game makes everything 0,0,0 at night. I can't see shit
Still would need to be converted into bitmap texture for the final compositing.
Bro, 11gb +- doesn't fit in 4GB
But point was, we don't know how much of that 11GB VRAM usage is "real" usage, and what is just old data being left in VRAM "just in case".
Real as in needed right now to render the current frames.
I am pretty sure, whatever of my main straem games I open, it will always be over 4GB. And that was my point. 4GB is dated.
the way a lot of games work now is that if you have the space, it will use it
MHW: Freshly loaded into the game. Old game too.
its faster to load from vram than from disk
In the hub, which is small. With no monsters and no other people. 5GB 
on menu it sat around 2GB.
this video i found to be interesting on optimising graphics: www.youtube.com/watch?v=jusWW2pPnA0&themeRefresh=1
again, it is probably loading data that it knows it will need for when you start a game
Let see if I can sneak into a server.
again, I said; games require more than 4GB.
oh absolutely
Baldur was like "oh, but how much old data is loaded into VRAM". I just boot up the game.
they have since about 2017-2018
It's a fresh game. So 3 games, all one shotted the 4GB limit.
2 of which are very old.
Squad loaded into a game:
i think the r9 Fury is the only 4gb gaming card that had any reason to exist after around 2014/2015
And my point was just that it probably isn't that full almost 12GB, but it is impossible to know how much is actually needed at that specific detail level, until you try with card that doesn't have enough VRAM and it spills into RAM.
I mean, it's better than just stating facts without anything to back it up.
Just "If you have enough, you cannot know how much is needed".
anyways i recommend watching this
and no, its not that unhinged threat interactive guy
And just believing nvidia saying that 4GB is enough is weird to me too.
And that is absolute crap.
While all minimums I see are 4GB and recommends are 6GB+ for games.
Satis doesn't even try ๐ญ
Figure it out yourselves.
Hey, you can play Fallout 4 with 4GB of vram (according to system requirements) ๐
And those minimum cards have at least 4GB of VRAM, which isn't always enough for Low textures in 1.0.
And then the recommended has 8GB cards, which isn't always enough for Ultra textures in 1.0.
this game uses raymarching, basic RTGI, simple RT+SS reflections, and more cool graphics tech and it only needs:
What game is it? 3GB available space ... is sus.
So looks like a indie game? Spahez 2?
Tiny Glade ๐ ๐ค
I am having too much fun in enshrouded.
I still want Stalker 2. But waiting for patches.
"Over 1,800 fixes and adjustments, including the first iteration of A-Life fixes." 1.1 patch. Damn, they going hard.
it was pretty nice before patches, i did notice A-Life being weird tho. Hopefully that is good now
cant wait for BG3 patch 8 tho
i want my swashbuckler
What is A-life 
its the offline npc sim
Aah ok ๐
it was from the original games, its how you encountered random stalkers fighting dogs, or bandits, or just patroling
@stray badger Tiny glade has demo btw.
Too late
How do I convince my girlfriend to let me buy a rack mounted server
But where is my flak for buying overpriced parts?! ๐ญ
I think i found cleverer fix: start with datetime that isn't HH:00, but HH:59
Just buy it? My two cents. If it's your money, and you don't have shared responsibilities. But ofc, if you bought a house together and need to pay of loan, that's different.
Do you have some place to actually put it in so it doesn't make everyone in the house/apartment deaf in few days?
Oh, is it because of the noise?
You normally don't want a rack mounted server for home use, because of that noise problem.
basement.
ok so I think this explanation is wrong
https://youtube.com/shorts/BA1te3mmY5k?si=gZkJRysxFmyEnWbH
My first thought: wait but i want realtime data on body positions
my second thought: oh, but gravitational influence travels at speed of light, so it perfectly makes sense
Imagine if EliteDangerous wasn't just computing skybox according to your position in galaxy, but also with respect to to how objects traveled in those hundreds of years the light took to reach you...
(well, pretty much the same, yet... )
Ok so now my perception of time is shattered
and that is not all!
Finally, UT1 is the least uniform time scale of all because its clock cannot be housed in a laboratory, nor is its rate established by any human convention. It is, rather, the clock whose โhandโ is the rotation of the Earth itself! The direction that the Earth is facing determines not only the coordinates of every city and observatory in the world, but also the local directions that each site will designate as their local โupโ, โnorthโ, and โeastโ.
It is hard to predict future values for UT1. The Earth is a young world with a still-molten iron core, a viscous mantle, and ice ages that move water weight into glaciers at the poles then release it back into the ocean. While we think we can predict, for example, Jupiterโs position thousands of years from now, predicting the fluid dynamics of the elastic rotating ellipsoid we call home is โ at the moment โ beyond us. We can only watch with sensitive instruments to see what the Earth does next.
this must have been the best documentation reading time I ever had
no..... at least not yet
I wonder if it will just run Minecraft
I just want a rack mount on becuase ot the drive cages
I love it soo much
You can get tower cases with at least 8 hotswap bays too.
tell her that server rack can be used as laundry storage
and a shelf for photo albums
(see what i did there?)
Isn't that close to stuff like what synology offers?
but old server/workstation is cheap. I dont want new
Micro-ATX example:
https://www.silverstonetek.com/en/product/info/server-nas/cs382/
For just a case.
But yes, various NAS manufacturers have more quiet options too.
Yeah, just that the problem with most rack servers is that they make hell of a racket when on.
there is also freenas, but those seem regular servers
To level of "annoying in next room with closed door"
most some dont which is nice
Up to 216 TB ๐คค
and for those quiet ones your probably pay extra too.
If you get a case larger than 2U you can use larger, slower and quiter fans.
it's only 480โฌ to make it quiet
or you could build it yourself for like 300
it's rated for 800W
Well, that isn't a server MB in the pic anyways...
And CPU is the easiest part to cool quietly in rack case...
Its all the other chips on the MBs that are also with "undersized" heatsinks as they are sized for the high airflow.
(RAM slots are wrong way around for real server MB)
And to cool 800W with 1x120mm radiator, that is that deep, that is expecting some insane fast Delta fan or like.
Not meant for making the rack server quieter, but allow more power at the same insane noise level.
3000+ RPM 120mm fans aren't quiet.
ehh vacume cleaner kinda loud isent that crazy
But all the time?
depends on the vacuum cleaner.
look at this and say what (aside from a FPV gun) told you that's a game
https://www.youtube.com/watch?v=j68UW21Nx6g
color is dam important
How some game devs are REDEFINING color theory.
Patreon โบ https://www.patreon.com/c/AiAngel
Gumroad โบ https://aiangel.gumroad.com/l/URTM
the weird blur, the cyberkenetic lines on the hand of the second screenshot, the flying drone
but mostly the weird blur/depth of field
weird indeed
clever
https://www.youtube.com/watch?v=IaXdSGkh8Ww
awesome
tech YouTube is like golden mine
Visit https://brilliant.org/alphaphoenix/ to start your free 30-day trial, and get 20% off a premium annual subscription!
So apparently buying a high speed camera wasn't enough, because after two videos with it I decided to build my own, but 5 orders of magnitude fasterโฆ
In this video I'm filming the motion of light as it flies across my garag...
oh way more than that if you have the 3D performance to go with it
the giveaway: everything's too clean
THE BLUR
every realistic bla bla bla is just blured to all hell
anyway i was going to say that colors in games often feel kinda... i forgot the word starting on "v", so will quote saturated"
I think instagram filters made their way into masses, but I alwyas find unedited desaturated photo being much warmer
feels similar to movies the colors always feel a smidge off
Car paint is too nice
Vivid?
I though it was it... and then i got doubts
Was Honey a legitimate money saving tool? Or just an affiliate marketing scam promoted by some of YouTube's biggest influencers?
Support my channel on Patreon: https://patreon.com/MegaLag
Instagram: https://www.instagram.com/mega.lag/
X: https://x.com/MegaLagOfficial
That's a quality thumbnail
i don't see thumbnal...
remember the times when YT first was showing you thumbnail instead of ads?..
(also, btw, is this nerd, science, or tech stuff? I think this can be placed in general OT with broader audience)
THROW BARBARIAN TIMEEEEEEEE :D
class action when ?
When someone sues them
so it's a poaching extension that doesn't help anyone but themselves, shocker
real
im going to buy a bomb suit
use one of their coupon codes
then sue them
a what now?
I think i cranked the code of how to get a data of what is going on our system:
First choose correct time scale, so that at least you are aware of its flaws (or take tdb because it doesn't have any, lmao)
Then understand what ephemeris word means, choose one that is balance between accuracy and weight (or just take the latest largest one available)
Then choose correct coordinate system to apply to everything
become the code
Don't forget about keeping the same units when you save them as floats without unit
Ask chatgpt if it's worth your while to use general relativity instead of simple newtonian equation -
Get scared
As i ma writing this in near total darkness i noticed my right field of view slowly pulsating getting even darker...
Feels scary AF
Write the code, get happy that it works
Get sad that it takes ages to complete
Forget that another lib that you couldn't make friends with because of how it doesn't convert time easily.... Had a nice way of factoring in apparent etc (i forgot the words) position of bodies relative to Earth
Get sad that your code is useless and perhaps it was easier to just go with what gpt was saying...
But it was using a bunch of magic words that seemed much more fitting for atronomy, telescopes and stuff. And in the library that doesn't work as easy, it seemed.
Huh
13900k swapped to 9800X3D and waterloop up and running, all seems to work except my old Samsung 970 EVO NVME drive decided to partly die and went to "drive failure" "read only mode" after swapping it to the new motherboard. Luckily can access it withhout draining the loop. And its not the OS drive
Did basic overclocks, enable XMP/DOCP, UCLK=MCLK and PBO +200Mhz, curve optimizer -15 all core.. or think so atleast. Lost with the AMD terms and setting naming ๐
Does this look correct
Well, if you get crashes in changing load scenarios, at least you will know where the problems came from...
I would really suggest leaving the CO and PBO off for start, until you are sure the RAM is stable.
Because first questions are if that 6400MT/s RAM is stable with the specific IMC.
And if that 2133MHz IF is stable with that 1.25V VSOC.
I did run Karhu tester for 2 hours with those settings without errors. Will try gaming today and more memory testing for the next night
Cinebench R23 is 23,3k points. CPU cores 75-80ยฐC and 155W power
Did you think to check stat unit Game difference?
HA. 5080 will be 16gb. what a joke
Its a console command for frame time etc in unreal engine? I'd like to compare these 2 CPU but I dont have any data with the 13900k and most likely wont assemble it with GPU anymore
Yes, would have had to collect the old computer data before disassembling.
Both stat unit and stat fps on:
Where that Game is the time to run game logic on CPU per frame.
So on some level that can be compared even with shit GPU in one of the computers.
But makes that computers CPU heat less, and so boost more, giving lower Game time than otherwise to some extent.
As the CPU will burst for the logic, then idle while waiting for GPU to finish.

Do you get the same speed when downloading games on steam/epic or the servers are overloaded?
I was thinking of going for 500 mbit or 1 gbit after getting fiber but not sure if I am actually going to get those speeds when downloading games from steam/epic
seems to cap at around 800Mbit
Hmm, guess I will go for 500 Mbit/s then
An example of downloading 150 GB:
15 minute difference doesn't bother me much tbh
it depends where you are and the isp you choose
here I get at least 950 consistently unless something else bottlenecks (cpu / ssd)
1.7gbps peak
what's the command?
dis is my daily, going 9950x3d when they're out
"stat fps" and "stat unit", like the message said?
wasn't clear
yall something really weird is happening to my pc
it has pretty much come to a stall
it takes 42 SECONDS to open firefox
turn it off and on again
25 seconds to start Xorg
3 times as much time to boot up as before
game performance is also horrible
I have tried restarting obviously.
maybe storage drive issue, check crystaldiskinfo
most people don't ๐
what is weird is that it doesnt even use all of my resources.
Been playing with the 9800X3D. No crashes yet with IF 2133Mhz, so its now 2200MHz. Left Karhu memtest run for my dinner, 1h 20min no errors
cant say ive seen somebody just toss a fan in to cool MB/RAM like that. How does it not just fall over?
2200 won't be faster than 2133 with your 3200uclk
double sided tape under the hub
could be. What I have noticed is that after the first time i open firefox (which takes 40 seconds), the next one is almost instantaneous. It could be that in the first run it has to load from storage which for some reason has become slow, and in the 2nd run it has it caches into ram which is why it is so quick
I will check that app you mentioned
crystaldiskinfo is not on linux from quick search
i use both OSes, both got the same rough problems
very low performance without even utilizing any resources
for linux i can use smartctl
ah i see
check disk stats with something at least to makes sure it's not perma queued and running awful
looks good to me
since its both OSes, clearly a problem with hardware or bios
maybe something is wrong with a bios setting
different problems, but I feel they are related. In linux everything is ungodly slow, while in windows when I am playing games i have probably 1/4th of the performance
but outside of games windows has no noticeable difference
havent tweaked those in a while
XMP is already off I have had 2 fresh windows installations corrupted one by one by memory fails so i had to turn it off
PBO is on, but i doubt pbo could possibly affect such stuff
very strange
yeah, i dont know man
it feels like its Disk operations that take long somehow, because when i try to open something after i already opened it, its instant
but the first time can take half a minute
caching happens in the RAM so ram loads are instant
maybe the first time where it has to read the disk it takes so long?=
I was under impression that FLCK does not need to synced for best performance, go high as you can. And I have little gains in cinebench R23
There is minor sweet spot at the 2:3, where next one will be slower.
But 2 slots faster is again actually faster.
But there is "problem", that IF has now very good resending on errors or like, which can make faster MHz be slower in actuality, if that causes there to be much more ECC errors and retransmissions.
Computer can be perfectly stable, but have lost 1/3rd of the memory performance from IF ECC errors.
Yeah I read about that, its hard to detect? Just benchmark and see what happens?
With the kind of profile that you are using it's unsynced, but there is still a latency benefit of a couple of cycles when uclk to fclk is exactly 3:2
in practice 2200 fclk for 3200uclk will be at best about the same as 2133 fclk, but a lot harder to test and more likely to be unstable (especially with high vsoc)
And like I said, that only matters for next speedbin of IF up, and two up is always same or better.
3200 uclk 2133 fclk has around 1ns lower latency than 3200 2200, i pretty much wrote the book on this so there is a good chance that you are citing me to disagree with me without realising
It is very weird and confusing because sometimes things don't act like you expect, but it's at least consistently weird so that with enough exploration and writing down you can see what happens. E.g. the 2167 and 2200 multis have worse internal fabric timings that you can't actually read out or control, short of flashing an ancient AGESA where they were tighter but probably unstable. Going from 2033 to 2067 to 2100 to 2133 has a large latency reduction on each step, but 2167 is actually slower than 2133 and 2200 is only on par but lacking the 1ns 3:2 sync benefit
since fclk is involved for multi-ccx communication i might humor the 2200 for some tests on those, but with 1ccx there's less benefit for going with the highest fclk number
- on this
needs thorough performance testing
Personally i haven't seen lost performance since a very old agesa, it either works or it doesn't on my sample and 2200 works. It used to error correct like crazy on 2167 and 2200 before AGESA update that made them slower but stable
had that happen once check your hardware cause you prob just blew something
@sterile plinth what type of storage are you using? (nvme/sata)?
nvme
its a modern pc
Could be something wonky with pcie controller or mobo depending on what exactly the lane config is and what slot you have it in
Try a different one ๐คทโโ๏ธ
I think it was "vibrant"
My monitor had this setting that made my eyes melt
Have you ran crystaldiskmark yet?
sent above
oh, not the benchmark yet
I use arch btw
Babies first words
lmao
that really needs Torvalds' head on it
This pricing makes no sense.
I am part of the hivemind now
Yea I use arch Linux
Welcome to the club
i use arch steamos
๐ I use Ubuntu๐คช
Wonkiness like that has been pretty common with external HDDs for at least 15 years.
ngl i thought you wrote wokiness
i bet that's prestige pricing just like everything else
extra 15% capability? +50% cost
But why have 22TB cheaper than 20TB?
Probably
and iirc some drives in certain capacities have a better MTBF, and when the manufacturer knows that, the price is just magically higher for those
I knew for a long time that my notebook does tiny noises whenever it's CPU does any load
you know, classic
I once heard how it affects audio channel (some weird windows drivers)
there was a couple time when on Ubuntu playing Journey also messed a bit with audio channel (as i understand, it's sand simulation is quite CPU intensive)
but today, i heard some noises consistent with when I move around a 3D model from 50MB HTML file
the only drive i've ever had die was in a Dell
cheapo 1TB, and it started screeching just past warranty. fuck them
they've screwed me over 3x over the years, and not in small ways:
-laptop dGPU fried itself from shit solders. class action suit settled w/o admission of wrongdoing, they sent me a $10 check for a $750 repair that wasn't covered
-tower's gpu had half the vram as advertised & specced, kept it hidden until i put it in another system and found out
-same tower had that shit HDD
Are you sure it kept that VRAM amount hidden?
Dell has had special half-VRAM to normal Retail Nvidia GPUs for long time.
But AFAIK those are specced normally, just possibly "forgetting" to specify the VRAM amount unless you look into full specs.
Misleading: Yes.
Lying: No.
Those yellow wires are dangerously close to that extra holes top edge.
yeah thats the hole we cut for the 8 pin gpu connector
itll get cleaned up a bit
Why get 5070 when you can get 9070 XT!
Yeah, but they went with "lower" number, on that 9070 instead of 9700? (when you ignore skipping 8000)
But someone in AMDs marketing side had "bright idea"
8000 was taken but yeah, very odd decision
How was 8000 taken?
wasn't on the box, site, or the stupid dell software suite stuff. didn't find it until i downloaded a third party app
What computer model?
So I can check the site(s)?
Like I said, very intentional misleading by not showing that amount anywhere but the almost-hidden full specsheet AFAIK.
But also not listing wrong number anywhere.
failing to disclose that they cut out half the vram from the original gpu spec definitely feels like lying to me
that's worse than selling a car with only 2 tires included in the price, at least you could install 2 tires afterwards
i think the only think halfway decent on that system was the front I/O
by their mobile chips or smth iirc
On CPU side, yes.
But no 8000 series mobile dGPUs either.
how they pick model numbers
true
But still going on with "8000 is 2024 products, 9000 is 2025 products" thing.
Where the year for those changes already in fall...
Even when laptop CPUs already did go to different naming.
idk if it's better or worse than samsung's "2020 galaxy is called the S10, 2021 is now called the S21"
This won't even work if they're trying to confuse people into thinking it's an Nvidia equivalent. The valuable part of that branding is "Nvidia" in my experience. Way too many nontechnical people have told me 'I paid extra for the Nvidia!'
that's a fantastic way to tell people you're ripe for extortion
"But AMD has bad (unstable) drivers!" (5-15 years ago?)
lol no more than Nvidia
Yup.
just let people know about the 3090s melting with New World when it came out
But that is one of the large noises that Nvidiots make to dis AMD.
From very bad power management Nvidia has on hardware level.
Only looking at 1 second or so average power.
And not immediate power.
can tell ya right now that 95% of all the issues with my 6800XT was caused by me fucking around & not the drivers
Leading to that very spiky power usage too, where the momentary power was often double the average.
which is especially hilarious considering how many things they have locked down on their cards
AMD had similar problem with RX200 series, but fixed the issue after that.
But Nvidia intentionally disables all the fancy power monitoring and limiting features in the VRM chips used, and instead has that firmware level slow monitoring...
i know every company will have at least one of those "OH FUCK" moments like that, but man swearing off a brand entirely for it.......
That unlimited FPS in the New World main menu just allowed that bad current limiting to make insane levels of current swings, burning the VRMs with too much current.
More headroom, less swings as the chips didn't have so much "spare" processing.
which makes me suspicious that lower end cards might have more room for overclocking than they're allowing
Basically the game was so well coded in that main menu to use all of the chip, that it would render frames hitting all of the chip, causing insane current pull, way above the normal limit, then get limited by the averager to low power, then lose the limit and repeat.
Keeping the average power at the TDP, but pulling insane momentary current in cycles.
oh i remember watching the streams of people's cards catching on fire
who ended up eating that cost, was it amazon or nvidia?
No code should be able to damage hardware.
(except some very specific cases, like cycling HDD between power save and full power etc.)
And it was just weird to see it blamed on "bad code", when the problem was too good and optimized code XD
eh, if you're designing it to be potentially damaging, i'll allow it. but for a regular product that's supposed to be stable for years? hell no
Just missing FPS limiting.
was it the 3090Ti Kingpin that had that 3rd "for liquid nitrogen only" bios that if you ran it without LN2 it'd just melt
Those Nvidias GPU power swings were so bad that ATX3.0 includes parts that are just for that.
Now partially obsolete already as Nvidia implemented some better power controls in 4000s.
ATX3.0 PSU needs to be able to provide up to 200% nominal power on 12V rail in short bursts.
Don't remember the time limit steps any more, that 200% was for low ms IIRC.
yea think i'm staying away from them until they stop that shit. scares the hell outta me
0.1ms for that 200%
plus my 6800XT is more than adequate in like 95% of cases
"Oh, your 600W GPU pulls 1200W in short bursts, and that is to the spec"
And the amount of EMI those Nvidia GPU power cables emit inside the case...
Causing issues for all the other components.
your own nuclear meltdown plus an EMP
Stable current: no magnetic field.
Fast large changes: insane magnetic field.
Satisfactory causes some computers 5GHz WLAN stop working from that.
Motorcycle side KTM is on trouble too, will see if chinese will buy it for the brand
https://fi.pcpartpicker.com/b/nVhtt6 part list and 2 pics of the 9800X3D build
Thanks, its dusty and all the panels still off as I want to observe for leaks for a while and have some other peoples SSD drives to attach and transfer data, hence the lonely power cable at the bottom.
I had to replace the 45ยฐ angle from CPU block outlet, it developed a leak from where it rotates after attaching both ends to block/hose. Guess the sealing was hardened from time. Just few drops when aplied force to it. Its really tight fit with GPU terminal. But that what it is with custom parts, not every combination is tested.
Still depating about the CPU block orientation and how it then positions the coldplate to relative to CPU dies... Manual shows it rotated 180 degree to socket... But every build I found pics of, has it same way as mine.
Manual would suggest that outlet port is top left relative to socket
So input from center, flow to each side, then in bigger channels to bottom on each side, and then to the outlet.
Yeah I have the flow corretc. I'm guessing the rotation does not really matter. Maybe its symmetric, did not measure
i'd be terrified to do custom water cooling. for me that'd mean that i have nothing but top end components ๐คฃ
Basically symmetric.
With in theory little more on the that left side edge when outlet is in that down and left orientation.
But very minimal difference.
So outlet on lower right on standard AM5 would be that "optimal" where the left edge would be over the CCD(s).
So that first pic in the thread, with the logo right way.
my previous cooler was failing me because it had a whopping half a heat pipe over the CCD
My thoughts too. Maybe I will rotate it when the time comes for GPU change, to have more room. And then its coolant replace too. Now I just blocked the lines and swapped the CPU block with rest of the loop full.
I meant how it is in that pic of yours of the build.
Yeah got that, I meant that as the difference is minimal, rotating for making room is way to go.
But that depens how the waterblocks for 5000 series are
Looks like the frame of the top fan has a shape there that makes it look funny
Actually I could now run those cables thru the rubber grommet to right as there is no thick top radiator in the case
0.1ms 200% is still conservative
it is beneficial for psu cabling margin
it remove the ability for manufacturers to push too tight cabling
and 0.1 200% is too short to all what Nvidia did
either Nvidia was pulling over much longer than .1ms or was spiking even higher
Wrong place, but you have incompatible mods active.
how would i remove the mods if i cant open the manager?
Hello there fellow Pioneer!
You're currently chatting in off-topic channels.
Please note, the entire off-topic category is reserved for not Satisfactory related chattery. To comply with your FICSIT Inc. contract, please make use of the appropriate channels in the Satisfactory category. Thank you!
I really would like at least some authority able to review products when problems are assessed
like some people or company could fill complaint and this authority could rule out noncompliant products
Wonder if we'll see smoothing capacitors on card inputs now
what do you all think the highest end RDNA 4 card will perform like?
On TWiG, Leo, Jeff, and Paris compare the output of the different available generative video AIs.
For the full episode, go to: https://twit.tv/twig/799
You can find more about TWiT and subscribe to our full shows at https://podcasts.twit.tv/
Subscribe: https://twit.tv/subscribe
Products we recommend: https://www.amazon.com/shop/twitnetcastnetw...
Google Veo 2.0 looks the best compared to others imo
Pika 2.0 steak multiplication though 
https://www.youtube.com/watch?v=UGO_Ehywuxc
another banger from Welch labs: mustwatch for critics of LLMs
Go to http://incogni.com/welchlabs to take your personal data back with Incogni! Use code WELCHLABS at the link below and get 60% off an annual plan.
Welch Labs book will be back in stock in January!
https://www.welchlabs.com/resources/imaginary-numbers-are-real-high-quality-printed-book-pre-order-ships-dec-9-2024
Welch Labs Posters:https://w...
Only gonna cost you both arms and legs!
https://www.youtube.com/watch?v=-CPhgNHVhO0
Ok so now i want it.
I bet this could fly very well with some kind of winter/Christmass update
The highly anticipated December update for Tiny Glade is here, and itโs bringing incredible new features! This update introduces unlimited lighting, giving you the freedom to create beautifully illuminated builds without restrictions, and movable clutter, allowing for even more creative control over your designs.
This update is a game-changer fo...
well it kinda does but the link leads to the game... so I gues it's no more
That's confusing
"Meet the FixHub Soldering Toolkitโthe compact, portable solution for repairs on the go! With a fast-heating iron, a 55Wh power station, and a travel-friendly tool roll, you can solder anywhere your creativity takes you. Customize your repair experience with advanced safety features and the most important soldering tools designed for precision.
...

I nearly survived this. at the cost of browsers...
The dam thing doesn't clean memory ๐ฆ
and i made stupid typo so after two runs I don't have what i need, and it probably won't survive the next attempt
it's just a 2.7G table with 68mil entries
wish me luck
question: how long would it take to solve 68mil inversed geodesic problems?
Not even knowing what it is... Probably a while.
you take lat and lon of first point, lat and lon of second point on Earth's surface, having ellipsoid params of it (model, geod) you compute a length of shortest arc to connect these 2 points
huh, 90 minutes
Not too bad.
i can't decide if i love or hate this little hack
So, python has variable scopes for functions
def wrapper():
counter = 0
def func():
counter += 1
print(counter)
return func
f = wrapper()
for _ in range(5):
f()
^ won't work because UnboundLocalError: cannot access local variable 'counter' where it is not associated with a value
But, if we make a little fun...
def wrapper():
counter = {0: 0}
def func():
counter[0] += 1
print(counter[0])
return func
f = wrapper()
for _ in range(5):
f()
It just works ๐
you can use nonlocal for inner functions if I remember correctly
Yes, but this little example just makes the error message seem untrue
especially since
def wrapper():
counter = 0
def func():
print("this far is fine")
print(counter)
print("this wont print because error")
counter = counter + 1
print(counter)
return func
f = wrapper()
for _ in range(5):
f()
and the error is the same UnboundLocalError
I wonder how many 10 year Python programmers will flap on this one
oh, ChatGPT suggest even more disgusting, yet pythonic solution:
at some point I stopped liking ability to just add atributes to whatever.
probably since those ocuple times when Python slaps your hand saying "no, this time you cannot"
laughs histerically
when you are half-way done through 90-minute job, and realise that you should have increased it's scopy by like 5% but then you wouldn't have to re-do it 5 more times ๐
i read that as "dicks are superior"
yea i had 0 interest in the soldering iron when they first showed it off except for it being portable. i do have the $75 kit tho, which has 100% paid for itself with what i've been able to access & repair
which kit did you buy?
can't remember the name of it, it has the screwdriver set, the spudgers, tweezers
Pro Tech
the guitar picks are great for opening up controllers & remotes without snapping off those stupid little plastic clamp things that love to break
the bent tweezers are sharp
like sewing needle sharp
me thinking why PNG render of a plot takes so long
checks the plot dir
See HTML already done
checks size
3.1G
...
I wonder what Chrome would say if I have asked to open it ๐
Oh I read it as you had a portable soldering iron for $75
If Gordon meant something to you in his years as a reporter, please consider joining us in supporting Gordon's family: https://www.gofundme.com/f/join-us-in-supporting-dara-glenn-and-marianne
Watch Adam of PC World / The Full Nerd's video. Adam was a long-time friend and colleague of Gordon: https://www.youtube.com/watch?v=id65yZst6ds
Watch our...
RIP Gordon
wdym by "flap"
doing dict is inefficient for this though
better use list
shorter processing and more straight forward
I don't understand what is even the point of the wrapper and function?
print(counter)```
Or if the weird stuff is actually needed, actually pass the variable?
counter = 0
def func(counter):
counter += 1
print(counter)
return counter
counter = func(counter)
return counter
f = wrapper()
for _ in range(5):
f()```
Ah, the outer functions variables are available for USE inside inner function, but cannot be modified?
Because they should have been available, or there has been a change...
And like said earlier, just adding that single nonlocal line would have been enough:
counter = 0
def func():
nonlocal counter
counter += 1
print(counter)
return func
f = wrapper()
for _ in range(5):
f()```
Current Python requires that, when in earlier ones it seems to have been implicit.
Which is why there are lot of examples out there where it is missing.
Meaning want give accurate prediction of behavior or explain what happens
Yeah, probably right
Well, enough, but I'm not sure which gives more smell: hack with accessing member of variable from outer scope or use special word "nonlocal"
which just means "use outer scopes variable"
I thought so at first, but
The snippet with "this will print fine" proves it wrong: the int var cannot be read either
That the x of wrapper and x of func are the same variable, in this case.
Yeah, like I said later, that the old implicit functionality was removed and now needs the explicit triggering with that nonlocal <variable> declaration.
And lot of internet guides etc. just haven't been updated to include it when needed.
How that explains that simply reading the variable has different behaviour?
It didn't have?
You couldn't read the outside functions variable in the previous tries either?
Until you started to use object instead of basic variable.
Just means "this variable is from parent functions scope"
For nested things.
Which seems to have been implicit in Python 2, but needs that explicit declaration in Python 3.
Made a mistake. Put my second NVME drive into a wrong m.2 slot and it dropped GPU PCIE to 8x
Fixed now to 16x but its funny how they share the lines on this motherboard (ROG X870E Crosshair)
Seems Asus used 4x from CPU for the USB4 controller?
Then M2_1 is 4x from CPU.
And then the 16x from CPU left over is shared between the two PCIe slots and two M.2 slots...
At least that USB4 controller is only thing I can think they could have used those 4x normally used for second M.2 slot for...
Or might be on that SLIMSAS connector?
No, that is just PCIe 4.0 x4, so probably from chipset.
Not spent much time to dig into that. If I had something in the second PCIE slot. Would have to leave GPU 8x
Would be so nice if Asus actually included the block diagrams in their manuals...
But I'm fine now that theres one m.2 slot left which wont drop GPU to 8x.
compare of Timespy results in 8x and 16x
So basically in run-to-run variance.
yes
But that was from the MB manual, showing what the speeds are for those four connectors depending which ones are in use.
So either 16x PCIe, or 8x+8x PCIe, or 8x PCIe and 4x+4x M.2, or 8x+4x PCIe and 4x+0x M.2
That would be the correct way to do it, however it's a bit bloated if you only need it for one thing. The nice thing with decorators... you can chain them and they're easy to disable, just comment the decorator line above the function.
But assembling basic machine was so nice, when you only had two power cables, two fan cables and case front panel cables to connect...
I can explain if you want
well let's at least tldr :
def any():
var = ambiguous
whatever()
ambiguous = var
any()
```will already give you the error
I'm curious about why Satisfactory, which uses Unreal Engine, hasn't been ported to platforms like macOS or consoles. Unreal Engine offers tools for porting games to various platforms, such as Windows and macOS. Does anyone know what challenges or reasons might be preventing Snutt (the game developer) from doing this?
tools != automatically
automatically != working well
working well != easily maintainable
btw now how many different hardware are Mac using ?

don't make me say what I didn't say though these tools probably make it easier
but it does not mean it's always obviously worth it
We already see with the ingame Vulkan supports status that just enabling something doesn't mean it will work properly.
And for MacOS, you need to use Metal for that.
And then all the issues with file system differences, third party library availability etc.
even then I was thinking performance will be a problem
sure some professional use mac for productive softwares
but now biggest stuff is done offloaded somewhere else
and pc (especially custom built) will still be more performant otherwise
Thank you for your opinion
also if you want to support the latest Macs you need to translate into arm as well
I shell not forget to put that to test and see
then we'll need to check your understanding about 2 aspect of things :
- variable's scope (any programming languages)
- variable's definition in python (specifically)
then it'll be intuitive
In C++ and C to some extent, there is what's called "translation unit" and it's mixed with namespaces to define scopes from ground
I will not go into details on this but if you already came along this it may help you understand what's being talked about
In any language even scripts like python there are 2 stages for variable's access :
the interpreting stage, and the runtime processing stage
the most basic interpreter will chain these 2 stages together
(process as it interpret)
(btw yes runtime processing of variable access still happen in high performance compiled languages even though it may not work how you think it is, it's not for no reasons we find that 6Ghz CPUs are nice, remember 6 000 000 000 processing cycles per second)
I guess people are partying now
What is that?
It is something edible?
Merry Christmas tech goblins
I think it's a boss fight.
I was working.
Ok so what's the outer scope supposed to be for this to show what was intended?
Well, it's somewhat common to find something that takes time to get computed
So such wrapper can get part of utils.py
on the basic variable against object I am like "what?"
I mean, everything is an object
and any variable is just a label you can on anything to use later
having immutable objects (like ints and strings) doesn't mean that you can' += them, because you can - it's an operation with a label
(although there's a thing that some integers have constant object IDs... shouldn't matter here)
and there are scopes, that should keep track of labels. You are free to declare a label with same name as something from outer scope (colliding argument name with something from outside of function, for example)
wdym outer scope ?
simply executing this function will show the error plain simple
well, yes. but then what's the demonstration?
demonstration of what ?
there why this happens and how you would go without error
(although i guess you meant to add any() call in the end)
what I show is not a demonstration, its an example
ok
there's an error when func is called because ambiguous is not defined anywhere (i believe define is a more matching word since error states about that variable not having value
But I don't get what was the purpose of that example. I don;t see how it advances argument
it's not because it's not defined anywhere
take that code (with a call after yes, I will edit if you want)
but commenting out the second ambiguous line
then you'll get the "no value / undefined" error
if you want I can get to the fundations, so it'll become obvious, but first you must not misleading yourself about what's the error reported
even if you write ambiguous = 5 in outer scope, it will change nothing in the first example
I asked about outer scope and pointed out the call to be sure we talk about same code.
Because is there wasn't a call - then it's all fine, function object gets created.
If there was defined ambiguous ouside of func, it would have been fine until the call of whatever()
Huh, I didn't write in time
wdym not write in time, this is no question, if you take the example code as is (I edited the call btw) and simply add ambiguous = 5 in outer scope
it will change nothing
clear
I meant that you mentioned that before i wrote that it's not as you implied
except it is, and I have tested it, and there is no otherway
my bet is you didnt uncomment the second line
I said first example
tldr : UnboundLocalError does NOT care about outer context
undefined's error is NameError
fun is in that i tested too before sending lmao.
except I think i did
bar = 1
def func():
foo = bar
print(foo)
func()
this is not it
instead of
bar = 1
def func():
foo = bar
print(foo)
bar = foo
func()
^
yes
yes "what"? ๐
probably instead of that, and yes it is what is happening in the example code
and itll complain UnboundLocalError at foo = bar
so adding assignment of the outer-defined variable later makes the first line of func became erronous
we will come to it
and the fact that it complains about that is hella missleading
we are talking about assignement later IN specific the function
otherwise it would be hella ambiguous but you are not noticing yet
it's only fair protection
it can be otherwise however good, but I am complaining about the fact that it is missleading as it is.
(yep I understand that simply raising that same error on the other line would be probably no more udnerstandable}
dict check
bar = {0: 0}
def func():
foo = bar[0]
print(foo)
bar[0] = foo
func()
So it is about assignment of variable.
when it is assigned locally (inside) the func, that variable is no longer considered as being of outer scope, which makes the line where it is seemingly accessed from outer scope to attempt to read var from inner scope that wasn;t assigned yet
If that error message where to say "... It is defined/assigned at X" that would be just good. understandable
so just before I dive deeper into variable scopes
now that you get what's the scope of the error
: (I edited var, because we are not interesting in it, so now every var is not specific object, so we ignore it's specificities)
var1 = 0; var2 = 0
def any():
var1 = ambiguous
whatever()
ambiguous = var2
any()
```is what I already shown, but what about
```python
var1 = 0; var2 = 0
def any():
ambiguous = var1
whatever()
var2 = ambiguous
any()
```can you tell what's the difference in behavior (you would expect, as the first one Errors out and we know about that already)
(I guess you probably already written about things along these lines, but I'm trying to keep a format in which it will not get ambiguous because of other things)
it should be like var1 is not defined NameError whatever checks โ
same as what ? we talked about 2 Error, the problem is not raising same error or not, as they are anyway very different
but lets say it is
we dont care about outerscope here
that's it, but poorly worded
I wouldnt mind if it wasnt ambiguous, but how poorly it is worded is probably rooted in some confusion you have about how scopes behaves, which led to this error not be intuitive
ok so if they are defined.
than var1 will be fine
var2 might be too
(honestly, I try to not even try to modify variable from outer scope. not talking about object's attrs though)
checks...
ofcourse they will be fine, it's not the focus
the focus is about ambiguous
you said it earlier, with misleading words
but I want you to think about what's really different in each
me when learns inner scope assignment is not referencing outer scope at all times
var1 = 0
var2 = 1
def func():
ambiguous = var1
print("whatever")
var2 = ambiguous
func()
print(var2)
well, ambiguous is just a normal variable there. inner scope and all
var2 is a tricky one a bit
now you can add ambiguous = 5 in outerscope and check the result
var2 is inner, the outer var2 Exist and work but is not the inner var2
same would happen with ambiguous
var1 can't be anything else than global/outer scope
var2 is not ambiguous because we are directly assigning it in the function as any other local variable
it's how the python code work
thats the second part I was talking about
how you are actually Defining variable in python (not simply "assigning" to them)