#programming
1 messages · Page 283 of 1

Hmm what about ai os ?
She will help in asm part or EXE TO APP FILES PART ?
BYE IM BACK SOO SOME ENGLISH CLASS
( I FAILED AT LANGUAGE SUBJECT)
||breakcode i need fix ascll || well today was tired day fixing some suss things and troll function
Can You Tell Me The Semantic Meaning Of This Message Please
ifeel like i'm losing my mind
break bye


huh
the game awards are happening
no wait
its "nominee announcement special" whatever that means
Exactly what it sounds like, nominees being announced 
you good? 
this 100s quote
i have no idea what the fuck a bigram ram is
but it sounds funny
maybe he means big ram
🐏 (large)
horror movies suck these days, if they want to scare me they should just show the Vulkan API
because working on this project while watching is still scarier
no

your pc requires one hp worth of power to function
a horse has a maximum of nearly 15 horsepower but an average sustained output of about 1 horsepower
i think the sustained horesepower number is fair
so my pc is one horse
this implies your pc generates power
you say that as if horses do generate power
they do
then nothing generates it
im confusing myself with my bad arguments
go tell your diesel generator it’s now a converter

more like dsl converter
:badone:
uh huh
me stuffing as many horses as my PC case can fit
🎠
a horse has to use more than a horsepower worth of energy
who said cars are good?

there are only bad, tolerable, and okay cars. no good cars exist. each car has a flaw somewhere
careful some of my favorite streamers are cars
clearly someone missed neuro car
Ellie is working on fixing that
i cant wait for Vedal to start his own car dealership and sell cars hooked onto Neuro's API
By the same logic apples cannot run people over and they are therefore-
Neuro convincing drivers to take a sharp right turn into a cross walk

passat b5
These Artificial Intelligence Virtual YouTubers are so fucking funny it makes me want to merge without looking
we need an edit of that
I could tbh
However it’d be funny to swap the color of his car to look like neurocar but I have no idea how to do that so I’d waste all of my time on that
3D printing gears doesnt work well apparently
the tolerances are too bad
too much friction
maybe wd40 would work
wait, wd40 is corrosive to plastic
eh, cant hurt to try
i cant im too busy suffering with vulkan

It happens sometimes
is that bad that my own anti bot protection started blocking me out of my own shit
😭
No.
I think you may have misconfigured something...
it started happening when i updated docker tho 😭
i didnt touch config for while
it worked
Do you still have the old version?
I assume that the config files got overwritten.
i updated everything
docker related
and anubis config isnt even touched at all
Oh, that's weird.
Is Anubis your anti-bot library?
@_@
well that was weird
I'm glad its working now
facebook scrapers are annoying
like i did put in robots.txt it to stop scraping my site
and it still does it
7k reqs per day are coming from facebook scrapers
Oh wow
I didn't know they did that. I thought that they would at least be a little courteous.
They are scrapping everything they want nowadays, hence why Elon has gone completely insane and made the Twitter API paid. Not even Google Search doesn't have any good Twitter links.
Imo, the worst scrapper is Yandex. They weren't caring about your stances even before the AI hypewave, which is when most started scrapping en masse
yandex stopped scraping me for some reason
Lucky you
its prob that i did put controversial shit on it
How controversial does one have to be for Yandex, of all people, to stop scrapping contents?
idk
my Twitter scraper still works after my account was suspended 
Well, Elon did say that he wants to build a completely new codebase for Twitter exactly because it's not built for being private and stuff
Granted, he said it soon after purchase. I assume that professionals have delivered him the whiplash of the expected costs for such a project
how did you get manage to get banned by twitter
i saw that much fucked up shit on twitter
and nobody banned it
You can literally get banned for saying the word ||cisgender|| because Elon Musk doesn't know what it means, lol
i didn't get banned yet for being trans on there lol
bsky lowkey better
Hello fellow
enjoyers
Hello

Anyone here use eclipse?

else if
elsif
elif
eli
ei
e
cl elon musk might be after my head now
i sent grok 4.1 a decoding task as a test and its been thinking for the past 20 minutes or so
imagine the amount of compute wasted if it fails aswell 
bro has already singlehandedly wasted kilowatts of energy
oh btw @sick owl
ai studio has 'high' media resolution setting now 
need some vision benchmarking for this 1
im cooked.
Apparently Musk claims Grok 5 will be 6 trillion parameters
Not a fan of the guy but I appreciate that kind of investment in the field, will be interesting to see whether that yields any meaningful step change over current models
Will pretty much be the thing to either kill or save the scaling paradigm
I'm 1000% fine with scraping, as long as the ones who are scraping me are Neuro and Evil.
Take my data, queens. 

I just needed to get some files from my windows drive, then it forced me to update, and now it says "nevermind"
one of these days im gonna rescue the 200gb of data i want to keep, and then im gonna nuke the windows drive
You cant just
Like
Plug it in while on your Linux install
And extract them
We love encrypted NTFS
if you don't have bitlocker enabled there should be nothing stopping you (assuming it's not in hibernation/fast startup mode)

It is probably fast startup
its probably fast startup ye
Just disable it then you can mount 
it takes maybe 2 minutes at most, stop being lazy 
you underestimate my laziness 
im here now because im refreshing colab, waiting for it to give me an A100
vanorstinkma -> 

thanks for reminding me i was supposed to send a support ticket today to get access to A100s again
new problem for tomorrow's me 
if someone gets rich enough can u hand deliver me an A100 thank you
I was doing some research on VR haptic gloves as one does. I'm not questioning how a HaptX pair just exists on ebay. Entire air condenser included.
5 digits iirc
oh no
it's the last "affordable" data center gpu from nvidia
5000 or 10000 depending on 40gb or 80gb
affordable being a big stretch
is the 40gb model fine?
if 5K is all it takes to lure and trap a vanorsigma...

true

give me 5 months o gather the money
5k, good considering buying new it is contact only 
wait what
reminds me of that meme
air condenser?
if you know what im talking about 
"that meme" is very descriptive 

i know what meme now
so why does this thing need an air condenser?
is the a100 the best gpu they have on colab
bye cuz you said you were gonna dissapear 5 hours
liarsigma 
For what I understand, it is used for force feedback. Supposed to be used to grip virtual objects.
enjoy your a100 
thats actually pretty cool
This partcular company targets workstation applications, like slapping on a headset and gloves to control a robot.
oooh
ai 
Experience the power of touch in VR and robotics HaptX equips your workforce with the industry’s most realistic touch simulation. Improve results while enjoying the sensory richness of real life. WATCH THE VIDEO HaptX Gloves G1 HaptX Gloves G1 turbocharges workforce training simulations, allowing team members to build vital muscle memory throu...
dk2 = devkit 2?
no they're just big fans of donkey kong

understandable
when they dont have a price on the website and jsut say "contact us", i already know im too broke for that shit
oh so the idea is to use compressed air to simulate the force huh
Which is why I thought the ebay listing was insane
in line with similar
The HaptX Gloves G1 are priced at
$5,495 per pair, with a discounted $4,500 per pair price for a bundle of all four sizes. An additional monthly subscription plan of around $495 is required for ongoing service, maintenance, and the Airpack system.
nah what
sounds about right
monthly subscription at half a grand 😭

you pay for the support mostly i'd imagine
hopefully the ebay one isnt on subscription
the airpack hardware is included, so you should be able to mod it if its a software lock
but the sdk being locked is probably the worst part, cuz then you dont have the option to do integration
not worth the effort having to RE the whole thing
login locked sdk
bruh
ye not worth the hastle i think
unless you already know how half this stuff works and you can easily make your own software for it
sadge
Looking at more options, time to ignore whatever product uses this
I've never used an A100.
How good are they compared with something like an RTX 3060?
I know they have very high memory bandwidth.
basicly, you're commparing a go-cart with a F1 car
It just gives an error code after getting to 50% with the update. No reboot just never updates. If you want a screenshot I can update the laptop to get the error code for you. Btw apologies for my delayed responses
they're made for different things
Alright, down to either of these two due to https://github.com/Adjuvo/SenseGlove-API/
These use the well known string method for force feedback
i would guess the V2 model is not worth double the price
I'm under the same opinion as long as the orginal continues to exist
This is already a reseller due to SenseGlove not doing direct to end user
My god, I am too poor to get into that hobby.
while the A100 has double the fp32 performance as the 3060, the main benefit is the massive amount of memory and the bandwidth. there's also some differences with tensor cores and stuff.
from my understanding at least
And time for the non force ones
aka buyable
https://www.bhaptics.com/tactsuit/tactglove-dk2/ Also makes vest and other stuff
if i did more vr i might consider it, but rn its really not worth it for me
Not worth for me either, just doing research
What do you do with GPUs?
i have a 3D printer, time to make one myself 
im a graphics programmer, so uh.. graphics
Neat.
Shaders are cool. And annoying to debug.
most of the time you get some type of usefull error log at least
For what the A100 is used for, training AI as it is a lot more vram usage. They are pretty fat programs.
yes 
1 ryot for when you have your shit together, another when your life is a total mess, based on these logos
Memory bandwidth is also a huge factor in some workloads, although I am not quite if (or how) AI utilizes it.
Speeeeeeeed
It isn't always faster. Some workloads are bound by other factors, like compute.
Sam be like https://github.com/LucidVR/lucidgloves/wiki
But then again, I wouldn't actually know, since I'm not an AI expert
hell ye
right but i mean the general time you'll ever notice it with llms is where the model needs to move data around, a 70gb model will be a lot faster to load from mem and generate on 10000GB/s vs 100GB/s
usually the model itself is what's moving
the compute is less of an issue for the end user, it's just giving it enough memory, and beyond that, enough FAST memory for it to not take forever
if you have to put any layer of the model in system RAM vs vram it's like
10x slower
instantly
$60 per pair price point in parts 
not too bad
Are you talking about memory size, or bandwidth?
well, size and bandwidfth
system ram is way slower
so that's the slowedown aspect
the issue of it being there in the first place is a size issue
So buying a 3d printer, all of the equipment, parts, and time is still cheaper than these
even vram is too slow
many of the inference time speedups/enhancements are there to "cleverly" move the important things around specifically to avoid them leaving VRAM - standard attention will do like VRAM -> compute result -> write to VRAM -> read again to do your softmax -> write back to vram again for result
flash attention tiles things such that they can stay in SRAM
vram is high latency but bigger amounts transferred at the same time
compared to normal ram
Who could've guessed, close cache go brrrr
The time that it takes to move a model from the system memory to VRAM, is limited by the maximum bandwidth of the PCIe bus.
What I'm referring to, it the time it takes to move the necessary weights from the VRAM, to shared memory/cache/registers. That is what I'm referring to when I say "memory bandwidth."
Another example is nvidia releasing a JIT compute library for array operations, that was in here 1-2 days ago IIRC
for random access ddr5 is faster than vram. its only for huge blocks of data vram is better
yes, but it's all affected the other portions so they have to be mentioned
The bottleneck is having to read cached values from hbm
but yeah the vram itself is still bottleneck here
the bandwidth between GPU and VRAM specifically
At first it's not an issue because nothing is cached yet so the tensor cores can go nuts and rip through
stoopid machine
Should just be faster
flash attention for example tiles things so that the part being used can (or tries to) stay in the cache
Yeeeeeeeeeeeeeeah, sadly the moment your working set doesn't fit into L2 your performance drops off a cliff
yep

What kinds of programs do you write?
anyways really from the perspective of running LLM inference (or training jobs) the maximum throughput of system ram is just the pci bus as the bottleneck always, because it doesn't matter how fast the dram itself is if it has to send it via carrier pigeon to be read and used
I'm currently working on a GPU cuckoo filter for my thesis 
toast is the original creator of the toast notification
The model stays on the GPU the whole time @_@
not if the model is larger than the vram
im back in this godforsaken blender file 
i need to make this more natural/wonky
which is usually the issue for a consoomer
Oh
A100 GPU has significantly more on-chip memory including a 40 MB Level 2 (L2) cache
the beeg gpus have beeg cache for this reason
The rtx pro 6000 Blackwell has a 128 MiB L2 which is nice but still not enough 
Never enough L2 
wheer rtx pro 6000X3D
ai devs on their way to ask for 16gb of 3d vcache on their gpus
alreayd got ya cover
Unironically please yes

no graphics, no proccesing, no unit, jsut cache and tensor
the KV cache becomes an issue even if the model initially fits in vram as well, the size of that grows as the conversation tdoes
so you start having to shuffle around eventually
Isn't the kv cache usually fixed size based on the context window or am I misremembering
I don't know what a KV cache is...
the context (prior conversation)
Yeah, I'm definitely not an AI guy
All you need to know about the kv cache is that it saves a lot of computation
KC cache is sorta interesting in that it can apparently store model intent https://www.anthropic.com/research/introspection
So this is the goal of all AI bros 
i mean the context is literally the local model's output which is also its intent, logically following that, as it's stateless
The algorithms are kinda neat
this is not the paper i expected
there was one kv cache communication paper
I hope I don't have to beg NVIDIA to keep FP32 cores on their GPUs...
There are some epyc CPUs with over a GB of L3 iirc
dw
?
I was looking for that exact one but the L3 in this only goes to 384MB 
They'll just cripple fp64 as they've been doing
i doubt those will be going anywhere on their consumer gpus, since we kinda need those for graphics
idk the only one im thinking of i was reading because of a multi-turn multi-agent thing i have been working on for way too long
basically was about how you can avoid re-doing the kv encode//constructing the convo history every time a new agent begins its turn
That's also bad
HPC guys are complaining about this
might not be what you were talking about at all
but that was just because they often have mostly the same kv cache when it comes down to it so you can reuse a lot
oh ye thats correct ig
make them different lms communicate via kv cache instead of prompt output
yeah im waiting for the firstr real production thing where they dont ever decode before arriving at an output
Good thing there's somewhat of a solution for this at least, aka buy an amd card
classic nvidia issue solution
I found (worst trade offer in history be like)
classic intel issue solutoin too
how epyc
epycally reduced cache
i was hype for b580, but with how intel is going rn i have very litle hope for the next gen of gpus
Hey Intel's GPU division has been doing pretty well
but noone uses it
It isn't quite that simple. Their architecture is different. If a program is specifically tuned for NVIDIA GPUs, then they might have to redesign it.
the arc b50 had a bit of fanfare around it
rocm is slowly getting there
Can we kill those two CPU skews
and vulkan sometimes is surprisingly good lol
like is arc gpus worth using than chunky nvidia gpus on the cloud
Yes then you go do that, it's not rocket science
but yeah it's not the same
because of support
but people are doing it because nvidia sux
amd has also been much friendlier to linux and such
Like where is your L3 Cache
I wanna replace my 5070 ti with a 9070 xt but I cba to write my thesis via ssh
Its not quite that simple for them, but okay...
i do like seeing big l3 even tho it's just desktop big
Speeeeeeeed
9070 xt only made sense at $600
I guess Nvidia's 5000 series does have some bugs
Still more L3 than those two server cpus
9700 would've been a buy for me at $999 probably
but like why we discussing L- caches
Why not, they're great
caching
are #programming doing low level kernel writing or someth
It's where all the performance comes from
my 9700 doesnt run llama.cpp well guys, wtf?

aint nothing gonna run with that
L cache is good cuz cache is great and i love cache and all the everything is cache
No idea.
Usually, I would try to opt to make my working data smaller, so that I don't need a huge cache in the first place.
you dont interact with cpu cache as a kernel writer
Smaller programs = more speed
you interact with it as a software writer
Triple typo
(man I'm dead)
it's just cache general atm
even for gpu i dont think you interact with cache
Technically you do
it was about gpu cache then it became talking about server CPU caches as examples of thicc cache
erm
Texture cache is somewhat programmable.
Shared memory is a sort of mapped cache.
Apperatly some cards from green can do so
yeah i just said to include it since it was mentioned before (in cpu context)
i wouldn't know since the lowest i go is vulkan
deepseek had a thing for it
Could you give me more details on that thesis program you're working on? I'm starting to get curious
that was one of their talking points when they were talking about how speedy they were using ptx
let me find
I think most of us are just stating knowledge
how far medium has fallen, from I don't want to pay for it to I shouldn't be using it
I've gotten paywalled in the past, that was a fun research day
the only thing that ever got me to pay from a paywall on an article was wsj because they offered like 6 months for $1
and i was like aight fine
i got scammed tho cuz i never use wsj even still
There's technically a repo but it's uhhhhhh tied to my irl identity
The goal is basically just to take a cuckoo filter and make it run efficiently on an nvidia gpu 
no more gems at medium 
Oh. Yeah, you don't need to dox yourself.
It sounds neat, I hope it turns out well
checked my wsj and found out we're still go for hype train despite the stock ticker and other headlines looking spooky
but it's ok because the moon will soon be a giant bezos face a la dr. gerald robotnik's ARK
im actually so nervous for the bubble to blow up, i feel like it's just gonna take my entire 401k and stuff and shred it into atoms
ripperoni
Thanks 
The implementation is basically done tbf, it gets closer to nvidia's blocked bloom filter than I expected in terms of performance so I'll take it 
number only go U P

yeah i dunno at least my retirement stuff is not all in nvidia or even tech since index funds but every member of those will still probably take a shit anyways
now my retail stocks are managed by a real expert, myself, and are up to the neck in nvda/pltr/slop/slop/slop/AT&T/GM/slop/slop/slop
so that'll be so joever i wont even bother to look at it
Great time to be a broke student
steve from GN would probably look at this and think it's a fishbowl for holding high temps and probably be right
but i wish nvidia wouldn't regularly use this image of this case but have 0 plans to ever manufacture or sell said case, and it's not a real case as far as i can tell
cuz i want one
so i can fill the front part with ugly ass hDD cages
Doesn't hyte make something very similar
yeah it's kinda similar to many of the fishbowl type cases
there was one at microcenter i saw but it was like 3 trillion dollars for a midsize case
It do be like that
I like my black phanteks box
Except that it doesn't have space for 2 GPUs
it was this i saw
i kinda like it except it's an asus product so i cannot buy it
i liek the intake vent line thingy
my corsair 4000D airflow has been pretty satisfactory overall
The epitome of form over function
i actually havent bothered to see how that intake vent actually works out
because i instantly disqualify any asus product from contention when i learn it's asus
if you clean a vacuum cleaner, you are a vacuum cleaner 
Valid
i liked my nzxt h440 but now would avoid nzxt
great for a nas box tho as it's now doing for me
not quite the epitome when these ballers exist
Okay but that's sick enough to make up for it
im not sold on the pyramid pcs ngl
https://www.youtube.com/watch?v=qYy7FRtUghw what about this epic shuttlecraft beamed back in time and shrunk to become a pc case
This weird mini-ITX PC case is shaped like a spaceship and extremely unique in its design. The case is the "Segotep EDI," and we're live building in it for the first time.
Sponsor: Buy ASUS' ROG Strix X570 ITX on Amazon (https://geni.us/SkMwj)
Patrick & Steve will be signing all Wireframe Mouse Mats sold during this stream at no extra cost! Mus...
what
what you just have to take the shuttle hatch off, kick the pilot and copilot out into the vacuum of space and then you've got basically a table to put stuff on
i odnt want my pc to be a modern art piece ngl
and then nobody will know where your pc is
except for the random wires coming off the shuttle everywhere
The only non standard PC case I'd accept is something you can mount on your back for VR.
at least the pyramid pc is able to be put next to a monitor withoutmuch trouble
the spacecraft pc woudl take up so much space
pyramid pc
coolermaster has been selling a case called the cosmos for years that is just massive and idk what is so special about it other than it being made of like solid platinum or something but it's always been infinite money for some reason
what the jelly
they keep making new ones
surrender now
these are cheap enough that i hadnt considered i can just buy these

cheap enough that the quality is worrying lol
at least with like open air "test bench" or the weird abstract stacked platform cases you really cant cheap out on materials that much because there aren't hardly any, anyways
it’s so bad i’ve been to an acquaintances place once and he had like a $500 build in it
i want that to be me
or a similar one to the one you linked
with my 4770k just sitting there like "wtf is my life"
poor old thing
whenever i grab a 5700 to put in my old x570 board to use that for the nas instead because god dAMN is the 4770k slow in 2025

AINTNEURWAY There's a button right here now
I used it at one point, then switched to Zen, then switched to Chrome because Zen was using all my fucking RAM
yeah vivaldi is basically chromium zen//floorp
in terms of what it'll do
floorp is more off to the side
but
zen and ff are somehow so goddamn inefficient now
Yeah, it seems though like Zen uses like 10x more RAM then FireFox
i liked zen a lot but had to stop using it due to firefox itself being poopoo
and then zen was another layer on that
i dont even have ublock installed on this vivaldi instance and it handles it all without issue
It regularly exceeded 4GB
on windows my browsers would regularly crest 16+gb ram
Let me guess, you don't close your fucking tabs evilSMH

@fast pagoda I don't remember, can you install websites as apps with vivaldi? I assume you can since it's chromium, but I just want to check
Okay I'll install it
i like tiled tabs (missed that from zen but vivaldi has it)
and it seems that whoever has the opinions on keybinds at vivaldi has the same ones as me because usually i'll guess some dumb keybinds for an action and it'll just be the default on it
I'll be uninstalling firefox too, because I noticed that it sneakily turns this back on after every update
kek
i have like 45 tabs open rn
had to count
and a few of them are VERY memory hungry
so i'm using 25gb of ram w/ vivaldi alone rn lol
which i dont understand
because the system is only reporting 19gb used
rn
Is it more normal if you aren't a lunatic with keeping tabs open?
i need to restart
so i can check
pretty sure yes though
chromium in general uses a lot of ram but it really just reserves most of it usually
it can relinquish it if needed
it just hangs onto it like a greedy lil buddy
I don't even understand how browsers work, like is all this shit just to load favicons for tabs?
that's a lotta vivaldi for not much
favicons are a mess
every browser uses its own standard for it and kinda supports the others
ok restarting to see what kinda damage it is
You can basically just think of browsers as operating systems, each with their own way of doing things

CEF 
last site i did these are the favicons i included (webmanifest is kinda part of that)
afaik this covers most of them
shouldnt need one per page unless you have custom ones for every page
but like why
Why can't there just be one standard smh, there has to be some way to make it all just use favicon.ico
I assume safari is part of that?
.ico is the issue
you can have several icons in a single .ico file
which icon is chosen is
up to the loader
from weekee
ok pacman finished brb
I'm still switching, but I just noticed that they hide this in the context menu
what is the benefit for a button to be right there anyways
is it an agentic browser now
or is it just a chat button
cuz like bruh
also cant restart yet because matrix and futurebound extended mix of magnetic eyes is going long but i can't say i mind (however i am required to listen to it in full if i hear it so i gotta wait)
.ico is old as dirt, it's like internet exploder 5 old
and now of course you have html5 which recommends multiple references per resolution of icon
and then apple being apple
who cares about favicon.ico
we live at a time when non hidpi devices are a minority
right yeah forgot about that
They make you accept something before using it
that said i do have a 512x512 in that thing i posted up there
i concur the entire situation is goofy asf but the humble .ico is less at fault than a victim imo
also the icon being defined in html makes more sense
yeah that part i agree specifying it is good
It seems to just be one of those browser agent things, but I will not be showing the popup because it leaks my name 
screenshots of my entire page and especially screen are personally where i can't handle it unless i have direct explicit control over when which this does give (according to it) but still gives me the heebie jeebies
i did test out project mariner earlier this year
but that doesn't use YOUR browser
it has its own
same as gpt's browser agent
idk how their "ai browser" works because they put it on mac only for the lulz but also havent tried the other "agentic" browsers since i dont want to get prompt injected -> financial devastation true combo'd instantly
browser use MCP and project mariner/gpt agent is the most i've gotten to there
i would've thought that thing would be mariner-like but i t seems like it's more local just screenshot sharing
not actual agentic stuff beyond the normal stuff it can do in the chat ui
unless im reading wrong
i should try it since this phone cgave me a year of google ai pro
but gemini 3 isnt on there yet so i dont wanna
NAHH I'm not reading all that, I just tabbed back in to find this wall of text 
Also, 
I just be talking
Giving an option to hide branding is something google would never do in a million years 
yeah i usually hide the workspace button and then either hide that button entirely (put the menu up top) or turn it into a hamburger menu
Another important setting
vivaldi was using like 2gb pre-restart after closing all but 1 tab
restart and it's using almost 4gb on vedal.ai
but i think this is more of it grabbing ram for pages in my sidebar and stuff
i think those may just kinda be loaded once you load them
or even before
they probably get suspended in some way but like ill leave music playing on a sidebar tab and it continues as usual
just loading a basic page, CRAZY enough they are about the same
the random zsh prompt that spawned in this window is 80 million lines long dont mind that
Works fine for me, I think this is just you
i just think it's the whole "if it's there, it'll use it"
finally deleted swap
i dont think 128gb of swap was also needed on top of the ram
still hadnt run swapoff for like the last month anyways
also, uh, i need y'alls opinion especially from experienced overclockers.
last time i've been offered a 14900HX modt for $300, supposedly is a ddr5 board but it turns out to be ddr4. but im offered to have it be switched out to a 13980HX modt instead for ddr5 board but because its like $10 cheaper supposedly, i get 2 extra tubes of thermal paste alongside with that
worth?
at some point i have a feel that 13980HX has a better binning, but 14900HX in most synthetic benchmarks has slightly higher score but its most likely on stock settings
it could potentially be an upgrade since im aiming for overclocking, but not too sure
I always google man pages
can't be assed to view them in the command line
What's fantastic is when I run <command> --help on an embedded device and it doesn't work because the command tried to delegate that to man which isn't installed 
really important note here. even slightly difference in binning would make the difference even if it means i can push just tiny but 100MHz more
its for overclocking, and im aiming to max it out, i know that extra 0.2GHz don't make any difference in real world but im practically just aiming for the numbers personally.
Check out the full version on our YouTube channel now! #GOTOcon #ProgrammingMistakes #Programming #SoftwareEngineering #WorstMistakes #MarkRendle #Linebreakers #ALGOL #Deloitte #AutoScrum #Accenture #ScaledAgile #Scrum #Viral #ViralShorts #GOTO
Full version available here:
https://youtu.be/Y9clBHENy4Q
Mark Rendle - Creator of Visual ReCode wit...
I love programmer naming
imagine it just returns an RNG of non negative integer
like straight up rng
or, for some reason i don't understand, it picks a random unallocated memory address or uninitialized stack and do absolute value on it
if only the things i install could freaking HAVE MAN PAGES
alien pages
nah you gotta go get htem online because frick you
get man pages for free with Amazon Prime!
all i got was grubhub+ for free with prime
wtf
i want man pages
i did use man2html (or if broke horribly and i ended up using something else, cant remember) to do a crappy browser for it instead which is kinda handy sometimes
LMFAO
I can't understand the paradox on this lmao
actually\
uhh
i mean that wouldnt be bad
i just leech off my parent's prime subscription
i get all the perks and im not paying
real email
but
i swear to god it looks like they used comic sans or something
just looks like lucida grande or something to me
This looks like every second sans-serif font in existence
The e looks a bit different I think, unless that's a rendering issue
probably weird rendering, i was just resizing a screenshot over a screenshot
That'll do it 
Interesting, OnePlus gave out a 50€ voucher for the 15 and is now giving out a 100€ voucher for the 15r
Kind of weird logic if you asked me but hey I won't complain
too long to stop the recording but
black friday moment
kinda like google giving me 500 dollars to get the pixel 10
sure thing gogol ill suffer this heavy burden
Not even that, it's a regular preorder bonus
oh is it new
Yeah
i saw the other oneplus phone on sale, ig that's the last one
i liked the seethru
oh wait
that was nothing phone
i think
cant type
Close enough
close enough
I just wonder why the cheaper phone gets a bigger discount, unless it's not cheaper 
people that purchase the cheaper one are probably also more likely to hop on a good deal for the cheaper one whereas otherwise they might not upgrade
or something
idk
Maybe
psychologists working in their strategy dept decided it was worth it
Numbers don't lie there has to be a reason
google has a bigger discount on the pixel 9s right now as well
even tho theyre last gen
until there's a 10a they're that
I'd get a pixel and put grapheneos on it if the hardware wasn't so awful
Not getting one plus, the last one we had no longer has a battery
I wish I could say the same, my 6 pro was awful
i had a 6 pro (i think XL)

I swear pixels are just built diffrently
they are either rock solid, or completely jank
6 pro -> 8 pro -> 10 pro no issues with any of them other than where they forgor to have drivers functional for the gpu for tensor 5
good ones anyways
all my googole devices have been pretty solid
i'd say the worst of the bunch was the samsung nexus 10 but that's just because it was an old android tablet w/ like android 6 or something on it max)
My 6 pro had a bunch of smaller problems that I could live with but the final nail in the coffin was a weird bug where the screen was locked to 60hz for a solid 10 seconds or so after unlocking the phone
and that's a nexus not a pixel
that's strange
nexus 5 was great too, as was nexus 4 and my brother had a nexus 6 that he liked
i think
Every month I'd hope for a fix but nothing happened 
There were also some issues will calls where I'd be super quiet half the time which yknow, not really ideal for a phone
I'm glad to hear your experience has been better though 
I think the most jank phone I had was the LG V60. Something about that install was cursed, from the display completely blacking out to the entire system slowly dissolving into dust.
Mobile reception was really bad for some people too, thankfully I didn't have that problem
lol the v60 was an epic phone (that i also have sitting here)
still works
oh you know that already
Yep
dual screen baybee
someone apparently has some sort of frankenstein build of android 15 working on sailfish (pixel 1)
i feel like that's basically impossible but it looked like that's what it said in the xda thread
That day was something, listening to music outside and noticing that "hey the screen is no longer functional after closing a malfunctioning app"
that's happened to me on just about every android phone i've had in history
I only had it with the v60, but yeah I'm not surprised. Only recently an android system update made it so the service restarts
You'd think the compositor would autorestart if killed, but nope. Only implemented 3ish years ago.
SMOL
Latest official build for the nexus 10 lmao
Phone sized tablet 
Though I actually have phone sized tablet
tab
Samsung
(typed on a Samsung phone)
i was in awe of my buddy's note 5 back in 2015, i thought that phone was so cool
memory says 3 but I doubt it at all
however i have still yet to own any samsung hardware besides this nexus 10 i think, most of the nexus devices were LG i think
it'd be a tab A 7 because they are named based on their screen size
not the like model succession
or just tab 7
Ah
I can't wait until this phone is being replaced and I can patch the kernel and do whatever I want
I love for all of these years it still has it's SKU tag on it
there was an old crappy 7 inch tablet verizon sold that they branded themselves to be like the verizon ellipse 7 or something, i still remember one(1) review i saw while browsing their website looking for a phone upgrade at one point, someone had bundled that tablet in and hated it so goddamn much
they called it the verizon epileptic 7 all over the review
Making games is also programming right.
i still chuckle about it randomly
write one(1) lua script and nobody can say there wasnt programming involved
So yes
this channel is for all technical things of that nature anyways
if that's hwat you're asking
I'd say mabe but I introduce to you https://discord.com/channels/574720535888396288/1336984748882399242
there IS that also
Alright in the future I might be a game dev so
threads are kinda their own, slower forum thing
i miss phpbb
well, not like it went anywhere
I should write a readme for my thesis project at some point 
I kinda have zero knowledge on programming as a whole but I am extremely interested and adaptable. (When I’m on coffee) so maybe planning is and preparing is the thing I am doing
Me and Sam both check this place often, we did meet each other from the first neuro game jam
we'd be on 3
i figured but i didnt want to skip one
all the subathons/birthday stuff blends together
i cant even tell ya what happened last year vs 23
other than the long wait was even longer i think
I'm getting to that age, closer to 22 now than USA drinking age
gamedev 
The past couple years are all a blur 
they really are
I blame covid still
especially literally all of the covid times
this message just means you're older than 21.5
I would say don’t drink but I can’t stop you
i also would say don't drink
it's about time to stop drinking by 21
True
I have a good idea of what it would to me, so I'm choosing not to
Just drink cofffee
so 21 being the limit in the us has always been weird to me
good
Coffee is the better choice
In Russia you drink at 7
it's not great no
ye nobody gives a fuck on this side of the ocean
Depends on state laws, some states have interesting loopholes
i dont know, never happened to me, closest was drunk gumby chase w/ the cops in school
Drinking minors I mean what can the police do but scold you
I'm looking at you "13 yo can drink for religious reasons"
makes sense
wisconsin next door to me parents could let their kids drink
at discretion
but they had to be monitoring them
technically

I drank at like 11 but I completely stopped because I don’t like it.
those types also exist
i had the classic "here if you want it so bad take a sip idiot" from my dad and hated any beer because of that for a long time
Only vodka taste good to me ngl
i cant touch mango or peach anything due to burnett's peach & mango
it permanently ruined those flavours for me and i doubt itll ever recover, i hate them now
Unfortunately for me, I only have soda really. Don't need much caffeine to live because I never tempted energy drinks.
i quit caffeine in college for a while
Soda? Eww
Soda taste horrible to me
actually the hardest thing ive ever done in my entire life
caffeine addiction is such a mystery to me
no exaggeration
i have never been more miserable
than when i quit all caffeine suddenly
i quit for a while and it got better but i was absolutely ruined for 3+ weeks
worst headaches ive ever had
sweating like crazy
basically totally useless
failed a class because of it
i can be drinking coffee thrice a day for a month, then just stop and never even remember i used to
smoking on the other hand is a completely different story
that's addiction + it has some natural anti headache properties
caffeine that is (and coffee itself)
like it can't be just me right
this is my query regarding my adhd meds
i can just stop
That's a common thing if your brain is wired differently
i feel like i should not be able to just completely stop and not even think about it for like days to weeks
without some form of headache or soemthing
which is weird since afaik nicotine and caffeine are very similar
I think every body reacts differently obviously
nicotine is like caffeine+++++
Good coffee does that
But having nice beans go stale is just depressing
that's why you don't brew your own and just go outside
in high school our school was rural so a bunch of students were hella all about that grizzly wintergreen and such
nicotine was ruined for me rapidly
For a while I actually had issues with caffeine in the soda doses. nasty heart pain. Though that eventually was figured out.
throw a giant hammer in ur lip, puke everywhere and swear off nicotine forever, ez
i kinda forgot not everyone lives in a city
I drink coffee and green tea as one
i kept thinking i was dyin in hs because of "heart pain" but it was probably just all the caffeine
are those pouches?
if so: you're not supposed to swallow
at all
no you are not
That was probably because of you having all of the caffeine without anything else in your food gutter
you will eventually swallow a bit just from the way saliva works, kinda not avoidable in entirety
but the main issue is putting 25% of a tin in your mouth at once
then don't i guess
well, the folks who are super used to it will without trouble
also it kinda hurts your gums so like might as well go berserk (bad logic)
programming btw
Okay, not to break this one up. But I don't feel like summoning mods today
We got off topic
beat you to it



wasn't me LMAO
Nah it was probably me
it was probably partially my fault




chat


she hates me
not disappearing

cool HDR stuffs
