#programming
1 messages · Page 509 of 1
firefox uses rust?
yes

firefox depends on onnx which needs to be recompiled when cudaSupport is on
i've actually been wondering if it's worth me moving back to linux...
i'll have to see how far proton/wine has gotten in running my usual daily apps
100%
linux, ubuntu, gnome. right?
What is you daily apps?
Do you play snt games with anticheat or use adobe?
rust is a virus, cancel fast and delete it
win win scenario
no virus + no scam
see you on windows in 3 days
fingers misaligned while typing 
currently using jetbrains software, discord, steam (mainly umamusume, screeps, but also have ets2 in vr mode and whatnot), trying to think what else
Not me. I did not break his finger 
ngl unlikely i'll move, i don't wanna fuck with my backups again
Hmmm, maybe not then
Wait until valve finished polishing vr
all those work on linux
Works yes, but still require effort for the vr thing
i'd personally recommend having a separate input for https://github.com/flox/nixpkgs and grabbing all cuda-enabled packages from there alongside their binary cache 
yeah, once valve's gotten everything ready for frame
kind of iffy outside unstable though, their "stable" branch doesn't follow 25.11 iirc
in windows too
vr itself takes effort to set up
tho once again my backups are a pain and i've not even finished setting them up on windows yet
Ye but you can bother customer support for every hardware ever on windows
and so on linux?
that's like, the point of linux
Not always
You go to community if the manufacturer doesn't officially support linux
Which is most of them
why bother customer support they'll just give up the second you say you've already done all the basic troubleshooting steps
nope
depends entirely on the headset
just hope and pray random user on random internet fourm has already solved your issue 
"how do you fix X?"
<comment deleted>
Thanks it helped!
Or when the fix involve downloading dodgy patch/driver
sudo claude -p “Fix $ISSUE”
the oculus CV1 has really shitty driver support on linux. i got it to work by manually editing some code. but i vibrated the whole time
Can you run Claude under sudo actually
Free head massager 
I thought the headset vibrated lol
vibrate at this user!!!!!!
"how do you fix X?"
stinky sock a clock the rock mock a lock minecraft block big flock
This post was mass deleted and anonymized with Redact
My most likely guess is the driver do some filtering
So the hardware sends out raw data
That's why it is jittery
the issue is the CV1 uses 2 cameras
and the use both cameras to generate a depth
but for the x and y they switch between the cameras to double the polling rate or something
why is computed justin all of a sudden fill me in on the lore
ts not funny 💔
What is ts?
ttypescrip
this or this shit
Is this one of the guard always lying situation? 
ts "ts" means "this shit"
fucking llm
literally anything that not claude opus is useless
every word is missinformation
the LLM way

happens
it happens when you hibernate, charge, and resume
It would be even funnier if it is desktop pc
100% reproductivity

worst case scenario i have a machine i can dedicate to vr
so not a huge issue if linux can't run it, but can't think of any reason to move to linux rn
would require more work than is worth it on my end rn
it really depends on the headset. oculus CV1 is worst-case scenario
besides no-name brands
eh, tbh the vr machine is more capable at vr than my personal machine
Sometimes I wonder how you are able to function cloud
oh?
vrchat is cpu bound im pretty sure. i never got more than 40% usage on my 3090 at the time. and now that im on a 7900xtx, or if i use the server with a 4090, that percentage will be even lower
what do you even mean by that 
how about you write the prompt in complete sentences instead of fragments... i'm a human and i wouldn't be able to understand what you want from me
I feel like every mods here always encounter the most weirdest of issue 
i'm not encountering a weird issue rn tho
I mean, windows is pretty shit
unless you're reliant on a bunch of windows-exclusive apps
i also have a CV1
i'm too scared going down the rabbit hole of trying to get it work on linux... so i'm just waiting for the steam frame
true, but what i have works rn, and moving os' would require a bunch of work that i just don't see being worth it
... tho having cron would be so fucking useful for backups, windows task scheduler sucks
Did it perform any web searches
its not unplayable, especially since its not visible to other players when you're close to them cuz of motion smoothing. only your own view vibrates a bit and people in the distance will see you vibrate from afar
yeah that's fair
personally I'm waiting for new pc before I switch
because I'm gonna have to setup everything from 0 anyway
i would add 1. 2. 3. in list at the end, and remove \n from first sentence; but it doesnt matter, 99% sure ai are fine with it
it just fucking awful
in fact it answered following structure, but it just said everything wrong (just not true)
no
You own PoV vibrating even if a bit is not exactly "small" issue 
Motion sickness is very much real
What model is this and are you on the free plan?
im not easily motion sick, i casn handle it. does give a headache after 2 hours
When I ask Gippity anything even moderately technical it starts web searching everything
is it what happens to gamers when they move from their chairs for the first time in weeks?
mby when it comes to my somewhat often os reinstall i'll install linux instead for a while, see how i feel about it compared to windows now that the linux gaming scene has matured
eh. i feel the main issue is the way you wrote the whole thing. the 1.2.3 wouldn't have helped
you need to ask a complete question and not just. A? b?. if you talk to it with natural language it will understand it better
knew there'd be at least one person suggesting nixos

those linux tribalists
i've been curious about it (or hell even just the nix package manager itself), but idk if i'd use it
i might play around with it in a vm or something for a bit before deciding if i want to stick to it
its handy
Is it leggy tho?
it seems very cool
who is nick sauce and why is everyone talking about him
been using it for a day and no problems yet
but this is all future stuff, i don't really have any intention of moving from current setup rn
i am not paying for ai - they all are useless
and opus (assuming you have to run it 3+ times for same input) doesnt jastify anything
id better search myself
all my exp with ai is outputs like this, it just fucking useless
except free opus, free opus is peak, it actually helps, if it is free
- i have other things to worry about rn than reinstalling my os
OK so you’re running crippled models
the thing i noticed abt nixos is that downloads are VERY slow compared to other distros
specially cachyOS
mby if i won the lottery and got myself one (1) 1tb nvme ssd, i could try out nixos in dualboot
What thinking level did LMArena offer for Opus?
yes only 1 1tb nvme ssd, that's all i could afford with £22m
ask shiro, shiro is rich fr fr
lol none now
they removed it
no
if you mean in past, that sometimes took longer then their timeout for quite complex input
so it just was aborting thinking
Did it give you options to set thinking level at all
no
Probably the lowest one possible then
i have a spare P310 1TB if you want it, 80 bucks
cheap
most nvmes go for at least half the world's gdp these days
i bought it for 76 euro a year ago lol
before the prices rose
it doesnt matter.
paying for opus make no sense.
using lower lvl ai is useless.
its all.
it useless even for just searches.
it was literally search.
ye
-# omg is it dns issues again
classic dns
always dns (unless it's dpi)
No, it's armey
sometimes dnssec

but that's just dns with a trenchcoat
i have no idea why dns sometimes just cant make it
vps provider provide no dns config btw
i asked via email
Try nslookup instead of ping
Nslookup can bypass the cache if you use a DNS IP directly
i kinda want whole system to work without manual reset
i basically just doing this tun0: inet 10.123.0.15/16 scope global tun0
and everything go to server|| (everything, no exceptions by domain names, it just not reasonable with full vpn via tun, or i just not know how to send via real device bypassing ip route)||
i have no idea why dns would break, it just regular ip packet going through tun
oh yeah this reminds me of a funky issue i'm having
this machine is on and is on the same network
i use aes so it is dns
if it would be dpi, i would have connection reset, or just nothing work
but current connections are fine, just dns cant resolve names
And you said you don't have weird issue 
Also, patch your system fellas. Another privilege escalation bug just dropped
Dirty Frag

so it's mac address is being broadcast
however the packets never reach it

What are the odds it was LLM found
you mean one that fireship made video of?
Low Level
does dns-over-https help though?
Highly likely since it is of the same categories as Copyfail
arp-scan -lg my beloved

Several people are typing...
meow
looks like it's at least LLM-assisted
got openreach modem -> tplink router with wifi6 capabilities, which is also a switch since most routers here are
one cable here goes to the server, another cable goes to a tplink SG105E switch with vlan support (tho the upstream router itself doesnt support vlan, so i use a port isolation-based vlan, which is currently disabled)
here it splits off: one cable goes to the machine in question (18:c0:4d:9c:a2:5c), another cable goes to a larger switch (don't know the model number) which then holds pretty much anything that needs to be able to talk to anything else
the whole point of the vlan switch is i sometimes need to isolate that machine if i ever need to give anyone else access, so they can't touch stuff i don't want them touching
↩ (@v4bel)
Haha, it's been a while everyone.
︀︀
︀︀Unfortunately, I didn't have the help of AI agents when I was working on this research. I started playing with AI agents after this project. If I had worked with an agent, the project timeline would probably have been cut in half.
︀︀
︀︀When I tested Linux kernel auditing with Claude Code, the agent couldn't find vulnerabilities of this bug class on its own. It was only after I showed it this article that it was able to reproduce some of the findings.
︀︀
︀︀In this era, "discovering new attack surfaces" and "identifying clear bug classes" seem to matter more than ever.
Actual pro at works
This is some other bug
ye I know
Or maybe I am of the blind
point is that this tweet was from before they found dirtyfrag
Meanwhile how many vuln windows actually have without us knowing 
It’s not like LLM can’t use IDA/Ghidra
And while Windows has obfuscated stuff, the important parts are not I think
I mean, theoretically LLM could just feed the raw machine code
it just can....
error probability is high and unpredictable, but it can
Sure, but I think for big things having actual decompiler is better
Honestly, I wish LLM could have the whole pipeline tho from source code to ast to ir to asm to machine code
All of those rich contexts that we're missing out
isnt ais already have some sort of ast?
or they literally just get code as input?
i really not into ai so idk
opus did a lot of compile time wrong syntax catches, i had to explicitly say that it is design/intermediate never compiled, ignore compilation errors
why would it need ir?
how would it work?
i mean you can just input output compiler results
and make llm gen code to placehold everything that make it fail to compile
fck ai
Richard Stallman dancing to Soulja Boy at MIT...
The GPL didn’t prepare us for this.
F
would be better if it would be never discovered
oyyyyyy doctos
Richard Stallman dancing to Soulja Boy at MIT...
︀︀
︀︀The GPL didn’t prepare us for this.
fixvx is easier to type in
yesn't
oh it embedded on desktop
https://nitter.net/i/status/2052463477852176765 this one i have seen
it has the wrong blue color and no comment RT like and views indication
do you have a favourite ntp cerber by any chance
fixupx
same thing as fxtwitter
i'm unsure what you mean
A personal favorite
there's also that one but unsuitable because it tends to get blocked or logged by automodding and also 💀
discord gif finder down?
discord attachement is down
am i alive??
holy shit finally
i've been trapped in discord api issues purgatory for like
15 mins
yes
Hi cloud
alooo
Now profiles are dead
Nooooo
Microservices dead
Wait i can just send images encoded in base64
Surely mods will like this

I am revived 
ive been fine this whole time
i only get gif finder not working
Hundred million dollar corporation by the way chat
John discord strikes again
the new dr stone episode is goated
discord outage wont let me send messages? nah id win
this is but a mere inconvenience to me
it has worked for a while
sometimes it jsut doesnt go through but resending it works after a coupel tries
I think i boosted this server 3 times around a month ago 
Eugh discord is fully down now
Oh now it works again
Discord has never heard about HA apparently
They have low availability instead lmao
Seems like it
Anyway I wanted to post a lab beaker as a plant pot xD
D:
language!
i can't even respond to you
So it seems like I can send messages, but I don't receive messages as they come in, instead I get a block of 20 of them all at once
I can see Discord for now, but I can't see profiles yet.
All your names are white for me as well
Perms are also not real
So no private channels unless you were already in one before it started
discord
you cant post images, but image links will embed
name colours work for me
i cans ee profiles and everything
Thankfully, I really like the colour schemes of a lot of the names!
so im doing
at your message, you jsut dont know it
The Nitro ad isn't even loading for me lol
I think when your message is read it isn't send at least no immediately

I extend the ONI mod but I will soon stop for today. New features Neuro get RP answers from the Party phone(not tested yet), can browse the wiki , can read notifications , can control time a little , get extended colony info , can query the state of other dubes
@cinder trellis Hey — I’ve been working on a small system called Halgorithem focused on detecting confident-but-wrong outputs in LLMs.
Instead of using another model as a filter, it tries deterministic checks (like math execution and source comparison) to validate correctness.
I noticed Neuro-style systems can struggle with confident hallucinations, so I thought this might be relevant. If you ever experiment with something like this, I’d love to get your thoughts.
i cant use reacts its over

Weeeeeee
Accend my boy
chatter not found 
That's on your end
They aren't dead yet

it is on your end
god, I wouldn't have known otherwise
a non malicious vedal ping is usually a zep/mute, I think
i dint get muted, but did get zepelinned when i did it by accident
I think embeds are client side so that explains why those still work

Time to bring out the at vedal gif

Guys can we program VedalAI to sing this please
https://youtu.be/3o8Kf9XTjUE
A song about how Hatsune Miku v6 kinda sounds like a mosquito.
No AI used in making this song
Lyrics:
You there, do you think my new voice sounds weird?
Like something buzzing in the air?
My friends told me I sound like a mosquito.
I think they're jealous of my new voicebank.
They're bullying me but I know
One thing for sure: I'm not a mosquit...
Where do I get this emoji?
I need to verbally harass encourage my friends to be better
we want another screeps server?
i had one running for some time and no-one used it
we arent using yours' rn
SCHIZO
yeah ofc not, i have it turned off cos no-one touched it after i set it up
shiro stopped playing and i didnt have acces to it
Cause you took so long to set it up we all lost interest in screeps 

i didnt lose interest, i jsut never got given acces
who's hosting the current one or am i banned from it for crimes against screeps
there is no current one, its jsut public server
i can hsot one tho
i have my own server now

or i could redeploy the one i have set up already
true
i'd have to move it to a more capable server tho, but that'd take me idk 5 mins
Yes, gib ip 
pretty sure there's a button for it on portainer
Still dont know what screeps is
sam
It it a game or some tool
get them into screeps

game
you can spawn litle dudes at the cost of some energy
and you program what they do
so they can collect energy to make more, or build shit, or invade shiro, or kill shiro, or steal shiro's room
its like robotwars but with programming and rts elements 
which is an interesting fact to say when my neighbour is here and does have fighting code afaik
not really

yet
i'm wondering... should i rather work on a game or work on my screeps...
partially cuz i cant posses another room rn, partially cuz there's another free room next to me, and partially cuz politically you have leverage as a neurocord mod

the last part 
but doesn't shiro have leverage as a neurocord admin
what is that?
ye but shiro wont ban me 
he likes me too much
screeps but i tortured it and put it in a docker container

okay so
Interesting
i don't have a "move this container elsewhere please" button
but
i have a copy and paste button
i could just... redeploy the container but on the better server
is that a custom server?
(i'd just restart this container but i'm decommissioning this server in a few months so i'd rather not redeploy it just to move it
yeah
are you paying for this server?
cuz i own mine, its next to me

i'm paying for both, but i do have a homelab
an screeps is on the 3rd one?
Relatable
1 is a dedicated box i rent from hetzner
currently screeps is on the ovh box
i'm moving it to the dedicated hardware box
because the ovh box has no purpose for me and i plan to decommision it

shiro was also mentioning stuff about a server for fumoquest
yeah shiro wants a server for fumoquest
i did offer to run the software if shiro can dockerize it
i offered mine but he's afraid of getting doxxed
Lmao
oh ye
Fumoquest data mines?
@rough bloom the tailscale only shows ipv6 now
just waiting for him to get back to me on whether he can dockerize the fumoquest server (or give me access for me to attempt to dockerize for him) @real sierra
ye same thing
does that work or is it still bad?
no difference at all
esp since the ovh box is
- a vps
- only has 8 cores and 24gb of ram, the hetzner dedicated hardware has 16 cores, 64gb ram

Sounds like overkill
im offering a server with 40 cores and 256gb of ram, and a 4090 that im gonna sell eventually.
i run a ton of things, so i like to have the resource capacity
not the core issue I think anyway
is it not?
having a shared access server provided by someone who doesn't really know what they're doing is just
the homelab is tiny tbh
4 cores, 16gb ram
give you access... to my hard drive?
does bro not have this thing in version control 
i know parts of what im doing, but everything combined is a bit annoying
networking is overcomplicated and sucks
you neither have VMs nor do you know how Tailscale works 
or any sort of access control presumably
i know partially how it works
i assumed you'd have this in git or something
Proxmox
it's a unity project
idk how to even put that in git
Bruh
do you mean Tailscale ACLs
okay that's a bit more complicated, but you can put it in git
you just put the assets folder in git
All of my unity projects are in git
Just ignore the libraries directory
Because its all temp bloat
I mean its better to have the project files and related too
U just dont need cached stuff
ye
wdym?
Never used tailscale
that's basically just a firewall
i should have access control set up for https://power.astonmachan.com/ i think
there's no point in it if you're going to intentionally share access
if it isn't then pls lmk
i want to give acces to my compute, but not every storage drive
so thats why ACL
ye ACL not do that
at all
this is what I meant by you should know what you're doing if providing access like this 
but I wanna keep it closed source
ACL lets me block incoming traffic no?
then i can set up who gets to acces what hardware myself

private the repo then
only give access to whoever needs it
tailscale has "tags" for that tho
isn't that a paid GitHub feature
i fucking hope not considering i have 100+ private repos
Lmao
I am debating on moving after the recent bs
So im probably gonna migrate to a self hosted
i wanna upgrade my homelab eventually
Tailscale is for networking, ACLs act as a firewall
firewalls don't do anything if you already have SSH access
just need a new motherboard i think
I need to upgrade the whole thing
Dont got time lately
This is my rare leasure conversation
but then i might need new ram, which then i'll need about 5 winning lottery tickets...
they made it free around the time when the Microsoft acquisition happened 
fuck em acls, iptables 4ever
i thought that was orgs
Lmao
Cant remember the last year i used iptables directly
not sure, could be
definitely wasn't unlimited like it is now
Its been like this for atleast 6 years
Imagine GitOS
Every file and write needs to be commited
Swap from ur gaming branch to work branch
what if you git committed your nix config and had dedicated branches for work and stuff
do you need to commit objects that are being created when commiting?
(i am making nixos users cringe)
no, someone somewhere is probably doing that
there is no ssh acces tho
thats disabled rn
must take super long to switch branches
how tf do you want to deploy anything then
usually shouldn't
because cached
-# still a stupid setup
the ssh would go through the ACL, meaning only certain people can get acces
so there is SSH access 
so
its disabled rn
but if i were to enable it
id use
{
"acls":
[
{ "action": "accept", "src": ["group:dev"], "dst": ["tag:nas:445"] },
{ "action": "accept", "src": ["group:admin"], "dst": ["tag:nas:22"] }
],
"ssh":
[
{
"action": "accept",
"src": ["group:admin"],
"dst": ["tag:nas"],
"users": ["root", "autogroup:nonroot"]
}
]
}
something like this for example
copy past incompelte 
man what is the point of blocking ssh if you still can access the server the other way

???
literally unusable library 
I am 99.8743% sure you couldn't have a private repo at all as a free user previously
such accuracy
I know because I wanted to be able to do that, for I can't remember what
In fact don't orgs still not allow private repos without Premium/Pro/whatever tf they call it?
they do
as in, they allow private repos for orgs
wdym "the other way"?
sounds obsessive
the other tag doesnt have acces to ssh
well you are able to change settings of the acl somehow
i can just put it in a drive they dont have acces to

i think i see the issue
there is no point in having the other group get acces if they dont have ssh
so then
you have to put the tailscale in a sandbox
and put shiro sever in a different sandbox
NixOS Containers shoudl be fine right?
nice
if only it didn't have a gazillion firmware issues
putting white around that is really dumb >.>
my 27gp850 is too not-an-oled to see this

you don't see the ring in the center?
thats the ring light for the camera no?
so it is the ringlight
there are almost no diffuse reflections, and the specular reflection is more than fine
G6 is def in another league, yeah
obviously matte will be less, this is the QN900F LCD
but the fact the G6 is doing that with glossy is pretty crazy
its less specualt, more diffuse
less noticable overall, but makes it so nothing is truely black
I read put shiro in a box and got concerned till I realized the context
Lmao
I actually have time I should probably write the firmware for my next project ngl
109% cpu usage 

Stummy hurt
Also what doing
compiling vllm
take long
many rams, many cores, but cores slow
2GHz base clock, 3.7GHz boost, seems to be running at 2.7GHz the whole time tho
compiling cachyos kernel on nixos 😨
might be bios settings cuz i didnt want it to be too loud
the kernel takes like 5 minutes to compile lol
Yeah
i think the base stripped down linux kernel is really smoll ye
50MB it seems
Tiny Core Linux offer a bootable base (kernel + basic tools) in the 11–17 MB range 
copying path '/nix/store/6sd5ji3bzbksfqf8krc6d4sgydfigrb8-python3.13-vllm-0.16.0' from 'https://cache.nixos-cuda.org'
if only there was way around that 
been way over 5 minutes now
like 7 minutes
actually 12 minutes
as long as it isnt docker
depends on how many cores you have available i guess
40
500 cores
i5 3470 im kidding
nix.settings = {
extra-substituters = [
"https://cache.flox.dev"
"https://cache.nixos-cuda.org"
];
extra-trusted-public-keys = [
"flox-cache-public-1:7F4OyH7ZCnFhcze3fJdfyXYLQw/aV7GEed86nQ7IsOs="
"cache.nixos-cuda.org:74DUi4Ye579gUqzH4ziL9IyiJBlDpMRn9MBN8oNan9M="
];
};
i don't think they cache 25.11 builds in the nixos-cuda cache tho, would need unstable (which you would want for vllm anyway) 
the compiling is gonna finish in maybe 5 min anyways if it doesn't slow down too much.
so imma just let it finish
that poor server, suffering for no reason 
gg
they're only like 20 bucks each
I am passively reading this whilst working on smthin, this is peak reading material of a channel
oh yeah you might be here for an hour
i feel like the cachy kernel should be cached but i don't feel like checking
ok so from what i see from tty3 it is still compiling i just got kicked out of the session
thats good
might be the cheapest FCLGA3647 cpu on the used market
ig

value of the ram: 1.2K, value of the gpu: 2.5K, value of the cpus: 30 bucks 
now what to write the driver in, I could do it also in C but like thats effort. I also could do it in rust but thats also effort
rust purely because i've been looking at rust all day
rust works, but C is the most commonly used for drivers im pretty sure
cas I really cba with anytmore memory shit
fair
I have been writing the firmware for this shit in C because its emebed shit, and thats pain
What the heck is happening in the room below mine
Working on a stream deck stlye thing, but works like natively on linux
anything besides construction it seems
These creeps are just screaming some random looping message
They're also wondering around everywhere
Including surrounding rooms
Seem to avoid rooms with other players though
checks out
Well I have something going on now
i should get a i7 3770 next
the annoying bit is that I actually have to build the prototype after writing the code
and then fix the code
and the prototype
at the same time
mb tho, and ram
i just got a peak cooler and a rx550(2gb, and free)
it is faster than the 5 3470, but in 2026 it really isnt worth investing in
thats actually preety good for getting the parts for 4 prototypes
Cheaper then my other supplier was gonna be
there are some i7 3770 on ebay for 11 bucks i guess
ehh only buy it when u fry that one
thats like 70brl in brazil, oh ye but with import taxes that goes to 90+
idk the value of brazilion money
and in brazil certainly its the same price or more expensive
0.18-0.2usd
but like, relativley, how much brl is a wage or whatever?
kernel is at linking stage atleast
im actually gonna be using my 3d printers soon at last
we get 1.6k minimum so thats like 270$?
https://youtube.com/shorts/M1-FgfatAeU?si=FsHhQmGx5qDpa8sM
lol Gemini flipped him off
A fun little experiment to test out how well ai chatbots can manipulate a robotic hand given limited information. The prompt was to generate a csv file which was then directly read and executed by custom software.
Learn more at: https://gestureplatforms.com/
Support on Kickstarter: https://www.kickstarter.com/projects/gesture-platforms/gesture...
i feel liek for 70 brl it still isnt worth it to invest in a dead platform from 2012
ye
anything more is 400+ like
u have mb, ram, import taxes
pre 2022 pc building was very cheap, importing psrts from china
now it is kinda of a hit or miss
and i5 3470 is still good nowadays
not like i will play AAA games and run full simulations
fe
ryzen 5 4500 is 45 usd, so 220 brl.
then you'd need a mobo and ddr4 ram
30 usd for a cheap am4 mobo so 147brl, and ddr4 ram, assuming we go really cheap for 16gb 3200 at 10usd, so 50brl
so totals 417 + shipping
not great tbh
whwere tf are u getting ram from 😭
ye that isnt real
bruhda
i have my ways. there's a reason my server has 256gb
I triedf to buy a 16 gig kit recently of 3200. It was gonna cost me 56 gbp lowest
I got my current kit just before prices spiked
So
gg
ryzen 4 4500 is 600brl minimum in brazil from ebay
hmmm
the 4500 isnt even t at great, i mainly picked it cuz it was cheap and then you'd have am4
300brl+200 shipping +100(very lowball for import taxes)
in brazil you can find it for 549 in here with no shipping
thats fucked
cheapest am4 mb 320brl
bruh
2x8gb ddr4 800 to 1000
so a 1.7k upgrade
which is kinda the price of my current pc
ddr3 is still expensive, and the psu, and rx 550
the 4x16gb 3600 sticks that i have in my desktop rn i bought for 100 euro, and they're currently worth 1330 euro
bruh
the avg in brazil prob is around 2-3.5k a month dunno, still very expensive
someone make a go-fund-me for this poor guy stuck in 2012
ong
it aint that bad
it is
do you only have pacman and the pinbal game for win 95?
tbf till august of last year I was on a intel i5 6400, with very low speed ram, so it can be made work if u know wat ur doing
that is wrose then what I had tho
so
gg
i was only i5 3470 with no gpu before this
gg
and i made it run zomboid as a host with mods
Linux is a must
ye
truly
goat
Its the reason I know linux so well
lmfao
cas
tho it did kill my last psu doing that
not smort or meta
but it was a cheap one anyways
1 hour and still compiling the kernel
why would you write in either of those
when c++ exists
everything
get this guy out of here
personally would just not
C++ has its uses
but for what im writing it does not
makes sense
Maximum core temperature 67 °C
huh, im used to seeing like 100C on cpus. seeing 67 on that i5 3470 feels weird
how would C ever be better what
bleh
I am currently writing the driver, which I am doing in rust
there are reasons to each of my choices for good reasons, and come from a reaosnble amount of experience
300 euro is 1.739 brl 
also in all right why are u questioning someone elses language choice
my max temp on my i5 3470 is 55C
100% at max frequence
my pc is already at 70C rn
gg
i think the cooler is a bit op for it
ye
but it makes snse
which cpu?
i have my fan curve set to only put actual effort in at 80C
5950x
damn
my cpu chilling at a nice temp of 58*C at abt 30% usage
I need to clean the cooler
thats a tmrw job tho
cas its normally at abt 54
i5 3470 tdp is 77w my cooler is 100w
my room is jsut generally warm rn, and the server exhaust is in the direction of my desktop
seems kinda like a bad idea
its somewhat temporary
random code snippet
but nothing is as permanent as a temporary solution
so real
no more playing with playstation1 graphics now that i have rx550
not really
you get the emotion engine
i can run warthunder at max graphics at 60fps
im pretty sure the rx550 is between ps3 and ps4 somewhere
which doesnt seem too much, its warthunder, but considering i was running on extremely low at 14fps
dunno
gpus are great
it is ye, just checked
you dont need anything past a 1080ti for good gaming
i guess ye

I am probably gonna have to fix over half of this later
any game that requires more is bad afterall lol
spoiler: you wont
ehh
It will either work
or it wont
or it will be broken in the most obscene way possible
or it will be working barely
eh
in the big 26, with games like res evil 9
mid game
what do you expect?
this is a triple AAA game, it wont run too well on a 2060
you said a 1080ti
1080ti is 30% better
this game aint good if even the recomended needs heavy upscaling
bro
be realistic here
its res evil 9
it does 1080p low setting on a 1080ti at 100fps
its not gonna do high settings
but the 1080ti is 9 years older than the game
i get that lack of optimization is an issue in the current gaming industry, i would know im a gamedev.
but just because an old gpu cant run a game doesnt mean its bad
God I wanna be rich and have 16x B200 GPU cluster as my personal ai playground, like shadow
pretty much it does, a game is not only visuals, as you say yourself, if a studio focus on visuals with no optimization so it cuts the playerbase, it is a bad game
I played re9 on 1650
there's the difference between visuals, optimization, and having optimized visuals
you cant expect 9th gen visuals while keeping the game playable on 7th gen hardware
$100/hr to rent 
Btw in this market what's the affordable yet powerful gpu?
7900xtx
In Nvidia?
Lol
This is equivalent to which rtx cards?
I don't have knowledge on RX cards
Ignore them
Ohh
3060 12gb or 5060ti 16gb
no
Or 3090
nooo
Those are your three options
kaine
30 series are bit old
Is there new gen coming? In rtx?
not anytime soon
Suspiciously non-cuda shaped amd
I heard they're making just AI chips
fuck your cuda
all my homies hate cuda
Okay well some will try and get you to buy T100 which are way way way older
How much vram does this have?
So this is objectively a new option
24gb
3060 is alright, but imo not really worth it compared to 7900xtx
im multitaskign and getting confused 
Is there 4070 or 4070ti equivalent but in half price? Like RX?
thats the point
RE9 is optimized
Yeah
If ur okay with it not working
*for ai
rx 7800xt probably
Ow my toe
That’s a large toe
when the minimal uses heavy upscaling to be playable and so does the recommended, doesnt seem so much
the minimum graphics thing they show on steam is exegerated, you dont need upscalign on a 1080ti
you can find people playing it online on all kinds of hardware
I wonder if dual GPU will make a comeback for upscaling and frame gen
except when they put 640p to 1080p upscaling as recommended for minimal
Run 120fps 1440p then display at 240fps 4k
no
dual gpu is horrible
2x the price for at best 50% the improvement
Someone didn’t read the last 5 words of that sentence 
Playing RE9 with the Nvidia GeForce GTX 1080 Ti in the first episode of my play-through series testing different GPUs in each episode! Enjoy!
⏱ Timestamps ⏱
Intro - 0:00
1080p Custom Ultra / TAA - 0:38
1080p Lowest / TAA - 8:16
1080p Custom High / TAA - 12:54
1440p Custom Low / TAA - 26:07
1440p Custom Low / TAA / 35 FPS Lock - 27:59
1440p...
brother
you can see upscaling is off
120fps
on a 1080ti on low settings
re9 is genuinly well optimized
you're dying on a hill that doesnt exist
no, it wont do high settings, or more than 1080p, on a 1080ti
but the gpu is 9 years old
you cant expect a ps4 to play ps5 games without compromises
Every pixel is one 1/1080th of the GPU 

it does high settings tho kinda its in the video, 90fps 1% lows 41fps
and still, that proves my point further
1% lows at 41fps is unplayable for a fastpaced game liek res evil 9
no good game needs more than a 1080ti
eh
really? i played fastpaced at 30fps for years and saw no problem

well ofc i did but wqsnt unplayable
it was decent, 40fps is way better, 50 amazing, 60 perfect
maybe for 1080p
but a 1080ti is around a 5050 in performance.. it's not powerful by today's standards
plus its 1% lows, not the avg, and 41 doesnt seem bad of a drop
it is
1080ti in todays standarts is kinda like "if it runs at all its good enough"
3060ti is the actual baseline
except 3060ti only has 8gb of vram
it's about 25% more powerful, yeah
triple A studios are going for good graphics on modern cards. they don't really care whether a really old pc can run it since its a modern game designed for modern pcs.
the "low" settings are only there to make sure laptops can play it
if you tell a studio that their game doesn't run well, they tell you to get a better pc
the low seems more like a placebo than really low
in fact many did lol
like EA
they change the textures to lower res, use lower res models, and do more agressive LOD
also cheaper post-processing
i think low setting feels like placebo cuz at that point you're cpu bound so the only real difference is vram usage and if you look really close
idk re9 low the textures looked kinda the same high quality
cuz you're viewing it on a 1080p screen, on youtube with compression
trust me, you can see it
Let the great signon begin
Android










its using 100GB

