#programming
1 messages · Page 148 of 1
I'll die faster when I see these not being packed into a variable
dont worry about it
also use := instead of =
should i keep old scores like my linux scores
why?
i think only keep them if its a difrent system?
more performance-wise 
https://stackoverflow.com/questions/4879592/whats-the-difference-between-and-in-makefile
I can't survive that because I'm still class 7 am to 2:30 pm
@olive sable can you run yours and tell me your peak
aight gimme a sec to go to windows

also what's your windows
thank god I don't have to build llvm myself and encounter more errors
We need to figure out a way to make Linux be faster again
pc-windows-mscv
There shall be no test that Linux loses to Windows
i remeber seein an 8 once, but now its doing 69841
# Leaderboard (Linux)
| # | Name | FPS | OS | CPU | GPU | RAM |
|---|----------|--------|---------------|------------------|--------------------------|-----------------------------------|
| 1.| Quack | 52958 | Linux NixOS | 13900 (ES Q0L4) | RTX 4070 Super | 4x16GB DDR4-3600C19 @3200MT/s C14 |
| 2.| superbox | 35600 | Linux Mint | 12700K | RTX 4070 TI | 2x32GB DDR4-3200C16 @3800MT/s C19 |
| 3.| Iggly | 30573 | Linux CachyOs | Z1 Extreme | AMD Phoenix |
| 4.| Shuni | 27158 | Linux NixOS | 7950X | AMD Raphael (integrated) |
| 5.| konii | 23141 | Linux NixOS | 7600X | AMD raphael (integrated) | 2x16GB DDR5-5200C30 @6000MT/s C30 |
| 6.| Prism | 22809 | Custom | HP Prime G2 RevD | (Graphing Calculator) |
| 7.| Iggly | 18677 | Linux CachyOS | i7-1260p | Xe Graphics (integrated) |
| 8.| Sam | 18153 | Linux CachyOs | 5950x | RTX 3090 |
| 9.| toast | 17000 | Linux NixOS | 5900x | RTX 5070 ti |
|10.| unkomp | 9458 | Linux Arch
# Leaderboard (Windows)
| # | Name | FPS | OS | CPU | GPU | RAM |
|---|----------|--------|---------------|------------------|--------------------------|-----------------------------------|
| 1.| Quack | 124130 | Windows 10 | 13900 (ES Q0L4) | RTX 4070 Super | 4x16GB DDR4-3600C19 @3200MT/s C16 |
| 2.| Vanyilla | 86370 | Windows 11 | 9950X | RTX 4090 |
| 3.| Adesi | 74199 | Windows ?? | 5950X | RTX 4080 |
| 4.| Sam | 72349 | Windows ?? | 5950X | RTX 3090 |
# Leaderboard (MacOS)
| # | Name | FPS | OS | CPU | GPU |
|---|----------|--------|---------------|------------------|--------------------------|
| 2.| Vanyilla | 6034 | MacOS | M4 Pro | - |
oh i got 124K on peak

Windows startup is annoyingly latent
On Linux once it's up it's all loaded up
124k 
fr
aight im going to bed now, gonna dream about the frames
i will figure out a way
time to try on my threadripper :)
To make Linux faster?
im gonna try ot use a profiler tomorrow to see why its faster on windows
we'lls ee
i downloaded intel vtune
What's that?
Interesting
ssd doesn't matter lol
I wonder how much more I would get if I lowered my timings then
I know for certain all timings reduced by 1 is stable
you can try
even just for general compute ram is a huge bottleneck on am5
I have cl18 so not too great
@olive sable i think we need to separate windows and linux scores ngl
Maybe for the best ye
i just forced my system to run CL16 despite the ram being rated at 19
....on 4 sticks
god i fucking love ddr4
they do use different frame presentation methods
You can do that on intel, but my amd would jump off a southbridge
Doing so would go from 3800 CL19 to 3800 CL18
3200 MT/s?
If so that's linear
Or not
I calculated with the wrong base speed
It would be about 1 tick under linear from 3600 rated speed
Either way, I wonder how low I could get my timings
Does anyone know good memory stress-testing software for Linux?
compile chromium
hmmmm
I mean one that's actually designed for testing memory and knows to check for errors
Also Chromium compile sounds like more of a compute benchmark
looks like 32 bit libs were pulled in
I need a stress-test that runs for a couple hours straight
at 120gb of source, it is very much a full machine workload
Downside of a non-K CPU I guess
That's actually crazy
But I need something more controlled
0.008056 ms per frame. 8.056 µs. 8056 ns 
When you have to start counting in ns you know shit is insane
Anyways goodnight
superbox what's your ram model
just few magnitudes away from GDDR6X latency lmao
Just a few 
@nocturne olive
i hage uh
This
gotcha
2x16gb ddr5 or something
I feel like this channel is the most isolated from the rest of the server ngl
Might as well be its own server even
But overclocked to 3800 CL19
cas latency is probably 30 or something since ddr5 
@glad path what's your graphing calculator lol
speed?
HP Prime G2
Hardware revision D
i think i was running at 5200mhz
not always though, it boosts because expo
Add a graphing calculator section lol
my motherboard supports up to 5600mhz but i dont want to push it
/srs
I dont own a ti84
Also my calculator can run c but the firmware doesnt give you that option 😭
Theres no custom firmware for it unfortunately
awa
# Leaderboard (Linux)
| # | Name | FPS | OS | CPU | GPU | RAM |
|---|----------|--------|---------------|------------------|--------------------------|------------------------------------|
| 1.| Quack | 55273 | Linux NixOS | 13900 (ES Q0L4) | RTX 4070 Super | 4x16GB DDR4-3600C19 @ 3200MT/s C14 |
| 2.| superbox | 35600 | Linux Mint | 12700K | RTX 4070 TI | 2x32GB DDR4-3200C16 @ 3800MT/s C19 |
| 3.| Iggly | 33191 | Linux NixOS | i7-1260p | Xe Graphics (integrated) |
| 4.| Semicolon| 32100 | Linux Arch | 5950x | RX 6800 XT | 4x16GB DDR4-3600C16 @ 3600MT/s C16 |
| 5.| Iggly | 30573 | Linux CachyOS | Z1 Extreme | AMD Phoenix |
| 6.| Shuni | 27158 | Linux NixOS | 7950X | AMD Raphael (integrated) |
| 7.| konii | 26364 | Linux NixOS | 7600X | AMD raphael (integrated) | 2x16GB DDR5-5200C30 @ 6000MT/s C30 |
| 8.| Prism | 22809 | Custom | HP Prime G2 RevD | (Graphing Calculator) |
| 9.| Sam | 18230 | Linux CachyOs | 5950x | RTX 3090 |
|10.| toast | 17000 | Linux NixOS | 5900x | RTX 5070 TI |
|11.| Assist | 13037 | Linux CachyOS | i3 8100U | UHD 620 |
|12.| unkomp | 9458 | Linux Arch
# Leaderboard (Windows)
| # | Name | FPS | OS | CPU | GPU | RAM |
|---|----------|--------|---------------|------------------|--------------------------|------------------------------------|
| 1.| Quack | 21589 | Windows 10 | 13900 (ES Q0L4) | RTX 4070 Super | 4x16GB DDR4-3600C19 @ 3200MT/s C16 |
| 2.| Vanyilla | 13531 | Windows 11 | 9950X | RTX 4090 |
| 3.| Sam | 11000 | Windows ?? | 5950X | RTX 3090 |
| 4.| Adesi | RETAKE | Windows ?? | 5950X | RTX 4080 |
| 5.| Caibi | RETAKE | Windows ?? | 14700F | RTX 4060 |
| 6.| Quack | RETAKE | Windows 10 | TR 3990X 64 core | 2x TITAN V + K80 24GB | 8x32GB DDR4-3200C16 @ 3200MT/s C16 |
# Leaderboard (MacOS)
| # | Name | FPS | OS | CPU | GPU |
|---|----------|--------|---------------|------------------|--------------------------|
| 1.| Vanyilla | 6034 | MacOS | M4 Pro | - |
quack what is up with your machine
guess we doin MacOS aswell

Prism's random graphing calculator in the list is never gonna be not funny
Ill try and get it running on my m3 tmrw
im tempted to get rev design/ES mac just for this ngl
youre about 1.4x faster than #2 on linux and 1.5x faster than #2 on windows
Has x11 been tried yet?
oh wait
remember chat, i did record and even stream my fps so i ain't lying about it
this makes me thing the reason linux is getting low frames is because its doing vsync
nah
TI is inconsistently capitalized
wayland will present every frame you give to it
._.
huh
x11 will just skip frames it doesnt care about
my score is technically using X11
the SDL Wayland driver was slower than going through XWayland 
not Xorg though
shuni try windows
yeah wayland only has non-vsync rendering on fullscreen apps and xwayland
no
If only we could make it use the metal backend fir macos
never
meow
Hmmm
Is that a good or a bad thing?
looking at the scores
and I have no idea which ones
its nuanced
Might just be the memory controller tbh
i will try running it in x11 next time
if its running on a set profile it shouldn't matter
especially when the only intel entries are all ddr4 lol
i did say it was a memory throughput and latency benchmark 
Mine is X11
this is true
these days?
i thought ddr5 latency was still way worse than ddr4
CAS timing yeah, but real world no
It mostly is iirc
oh right the frequency plays a role
Oh ye
iirc latency is also more impactful on ddr5 than ddr4
I have 3600 cl 18, so about the same. But 4 sticks so bad
trying to get 1 large stick is hard
How hard would it be to use metal backend for macos
4 sticks ddr5 
pretty hard since sam only wrote vk 
i overclocked mine to up to 4200MT/s, 4 sticks, when im still on a 14600KF
ddr4 doesn't give any fucks about 2dpc for some reason
but ddr5 is so picky
4 on amd is asking for trouble
locality bwabwabwa
Wait should I rewrite my code to use the hpprime micropython implementation
@faint sandal is living witness lol
his ddr5 4 sticks ran on 3000MT/s
om
what
Mine instantly stops posting when trying 4000 MT/s
except with twice as bad in latency
So 3800 MT/s is my absolute maximum frequency
I rma'd my 2ndhand ram cuz of memory errors, but months later i thought it might have just been the mem controller.
intel's IMC is for some reason really good
It is ye
Still not bad for being rated for 3200
I guess +600 MT/s is a typical DDR4 overclock headroom if
3600 went up to 4200
i guess intel was just improving their memory controller when they had market dominance
lol
check your motherboards manual, sometimes they list the maximum memory speed when running 1 or 2 dimms per channel with 1 or 2 ranks per dimm
an example from the motherboard I use
I don't know if it would say that, it's an AsRock Z690 PG Riptide
here it is for your motherboard
my dealer's bench
in the morning
4900MT/s on a 10700KF

Well, I already knew that
Definitely my RAM sticks being the limiter and not the motherboard
mf got 4900MT/s ddr4 cl16 and 7.1GHz on their 10700KF
😭
i wonder if that'll be enough to beat my scores
did a pull req for sam
you should be able to swap vulkan for moltenvk
oh yeah right
not sure how compatible they are
volk is probably just thin bindings so many it would work
I should remember to tune my timings back to how they should be at some point
if the 7950x ES somehow got second place then sam's code is 100% favoring for ES models for some reason
And probably tune them more but I need Linux stress-testing software
Also I want to overclock my CPU but I also need a stress-test and benchmark for that and Cinabench from what I know has no Linux build last I checked
which i kind of partially suspected and i dunno why
passmark
3Dmark
idk
right i should try more benchmarks
3D mark definitely be confused of what the fuck is my cpu
quack may you share your makefile for sam's triangle thing
I'm so stuck on the linker
ok hold on
3DMark paid as well?
you can use 3Dmark demo
Does it work indefinitely?
@midnight sigil
I don't like using limited time trials
its not limited time
Cool, I can work with that then
just limited features, and demo features is all you needed
Oh it's also not for Linux
you can get it from steam?
That doesn't make it Linux native, it'll just run on Proton and lose some perfromance from overhead
Finally I'm free now time some coding
odd
Visual code studio ☠️
do you already have SDL3 and vulkansdk?
U use gpp in windows still good then visual code studio
Are you building for X86 (32-bit) with the IDE?
How odd
LMFAOOOOO
the error occurs with "C:\\Program Files\\LLVM\\bin\\lld-link" -out:output/release/vulkan.exe -defaultlib:libcmt -defaultlib:oldnames "-libpath:C:\\VulkanSDK\\1.4.321.1\\Lib" "-libpath:C:\\SDL3-3.2.20\\lib\\x64" "-libpath:C:\\Program Files\\LLVM\\lib\\clang\\20\\lib\\windows" -nologo output/release/main.o output/release/vk_frames.o output/release/vk_command.o output/release/vk_device.o output/release/vk_sync.o output/release/volk.o vulkan-1.lib SDL3.lib kernel32.lib user32.lib gdi32.lib winmm.lib imm32.lib ole32.lib oleaut32.lib version.lib uuid.lib advapi32.lib setupapi.lib shell32.lib dinput8.lib

try removing the x86 and make it only use x64
LLVM 20.1.7 (x86_64-pc-windows-msvc) btw
never got so deep into c / c++ that I needed make, but reading the makefiles gives me an aneurysm
Confused about what a Linker does in Embedded Systems? 🤔In this video, I explain the concept of a Linker in simple terms — what it is, why it's needed, and how it fits into your Embedded C build process.
You'll learn:
How object files are combined into a final executable.
How the linker resolves function calls across different files.
Why th...
@midnight sigil
I know what a linker is 
there's no x86 libs in the vulkansdk, and I removed x86 libs in sdl3
same errors
@midnight sigil try hard
huh
Wrong VulkanSDK?
nope
Library Version Mismatch
Missing Files:
Incorrect Paths
Location ure gpu driver files
nope
nope
nope
try getting your own clang++
aka I build my own llvm?
I think you're accidentally using the visual studio clang
Use gpp library
omg free mone
wait no it's not clang's problem
it's the linker's problem

3DMark just does not run on Linux at all
I think I finally found some benchmark tool that has a Linux build
It's called triangle
No actually
I wanted an actual one I can use for stress testing and stability testing stuff
And just comparing performance
iggly have you seen the new leaderboard
I believe a program called stress might do what your looking for
Huh?
wdym by use gpp library
Where does one get such?
A bit of backreading, windows meta is real
😭
Wher get?
I wish it wasn't real
@amber fractal btw what's your ram stick
sudo apt-get install stress
Oh great, it's on apt
I mean it is elsewhere
I'm on mobile still rendering
FPS: 314.018
FPS: 311.065
I just picked apt
what phone/processor?
it's not
Don't know off the top of my head
Calculator (real)
what's the error now
2.4 ghz
One of these is Legion GO, not upgradable anyways. Will have to grab laptop
I'll take it as well for windows testing
Let me check the text to recognize in my phone
it's not long term though, I want to modify the code 
What has Sam done to this channel 
this one has a prebuilt, just extract and cd your way to the folder and run with .\output\release\vulkan.exe
do not cd to the release folder, doesn't work that way lol
Triangle
🔺
Triangle
🔺
Uhh, the stress thing doesn't work
It just crashes and makes Steam completely freak out
lmao
I could have never seen this coming
Do you know other more functional tools?
stress works for me, don't have anything else.
i wonder if this works blackbox
need to try
hold on
What command did you use?
I was stress testing cpu, so stress -c 12
this one you just need to run triangle.bat
(this one is compiled on an engineering sample CPU, might not be optimized for everyone unless if you have a 13th ES)
Why is discord only warning me on this one 
oh this one work universally
I did also find out that one of the tools I used for stress-testing on Windows has ported to Linux, so that's helpful as well
Chat, I feel like the goal in life should be a 1M fps triangle
damn, I guess I'm going to have to fight this code to get it to work
Debug doesn't exist mhm
mreow
maybe compiling sdl3 from source is not the best option

nice
let me try to install a mini windows
my threadripper got 32K lol
SMOL
I mean, sam's triangle only uses 2 cores 
sam smh
Maybe GPU limited with the Titan V?
make sure power plan is set to performance
i need to try on my 125H laptop lmao
should i make a one click compile version
can it compile on my pc 
i'll try lol
That'd help to get more datapoints
for those who are wanting to tinker
yeah
easier to get everyone's score aswell
i'll bundle sdl3, clang++ and vulkansdk
Vulkan SDK? 

good luck
but i think its worth it for people that doesn't wanna bother getting vulkansdk and sdl3
I'm gonna do a real quick silly test with all components on max load to see the temps
@sage crag look what happens because you mention statically linking vulkan
May as well just bundle vcpkg

The prophecy 
Seems like
- 3090: 80 C
- 4070Ti :90 C
- CPU: 90 C
When all components are fully under load
it has come full circle
Is this good or absolutely terrible?
Good? No. Okay? As long as they're not throttling
It'll only rarely get that bad anyway, usually the 3090 won't be running max when the other two are and the other way around
Then it's probably fine, this is a worst case after all
Since the 3090 is only there for compute workloads and the other two are for all the other stuff
Mainly ML is what the 3090 does
So barely ever used
there was some weird issue with win11 being slower than win10 some time ago but I think it got fixed
you could see this in some game bench marks
lmao it fucking worked

Best of luck 
Discord upload limit is only 100MB 
i gotta try it out though
nitro 
That is the nitro limit
but yeah targetting 100MB or lower
right you don't even need the vulkan binaries
No, the nitro limit is 500MB
100MB is the boosted server T3 limit
bwaa
just send files in chunks
If you don't have vulkan, you ain't running it anyways
does the executable work with x86(32-bit) 
you only need vulkan-1.dll afaik
You just realized? 
I guess Vulkan compresses well
nvm, 179 MB
and would x86 faster than x86_64
meh you need to recompile the whole thing in x86 what am I talking about
Probably not since x86 mode is emulated iirc
Maybe that was 16 bit I don't remember
once i got it to work on my other machines that means it works and i can start optimizing (removing unecessary binaries or lib or dlls)
bwa
Someone needs to fit the github with actual releases, because at least surely it can be done there instead of discord.
(Doing it with gh actions should be considered a crime to the person doing the action)
alternatively everyone use nix
works
200MB lmfao
my laptop only got 4k ish
both prebuilt and one click compile version
LMFAO
let me optimize the size so i can easily try on other machines
I'm not reinstalling my system
What the CPU is doing:
drawcall()
frame++
drawcall()
frame++
drawcall()
frame++
I got it working 
Updating the title bar is on a different thread so I'm not wrong
I wonder if there is a counting algorithm that is faster
Faster than the increment there? No idea
removing like 80% of the files in vulkan's Lib folder not only optimized size, but also performance for some reason
wtf
Also because I got a race condition (double fps for a bit) I know this is correct.
Hmmm, will function call be more expensive that incrementing?
Because I wonder if the increment can be defered somehow
I'm kinda assuming the compiler inlined the function
One way to find out 
Looking at the binary?
dude what the fuck is spirv
those took like 70% of the file size
i removed em, and performance goes up
Perfect
my laptop can now do 7K
spirv is what vulkan compiles down to
So does that mean it's the Vulkan compiler?
its not
packed environment
Vulkan shader format
instead of just shipping the shader code as text like you would with OpenGL you can precompile it 
You've heard of one click run, now we have for you one click compile (and manually run still
)
would anyone willing to try it out
This technology is considered the evolution to pre-computing as the resulting program fits the hardware it's running on perfectly.
oh the compressed size is 88MB
I nominate @midnight sigil
build.bat to compile, triangle.bat to run
NOTE:
- VulkanSDK and SDL3 is packed, you don't need to install these on your own.
- you might need GCC/MSVC installed.
- running the one click compiler isn't a must, but it often improves performance because you're compiling from source directly on your machine
might aswell try in my older systems
no way you packed clang++ and lld as well lol
Seeing the examples leave me with the intrusive thought of multiple languages in a single file.

Finally

and it runs 56k
well shit
people that uses it might also need gcc
uh oh
wait
hold on
ah its just some header files
I still don't know what's the issue before but it works now lol
62k fps
anyways
SUPPORT THIS MOVIE: https://store.gamersnexus.net/black-market-gpu-backers - the backer tiers are the only way to get digital rewards, like the wallpaper packs, bonus videos we're filming, behind-the-scenes with the team, and the USB key with the movie written to it! They are only available a limited time.
NVIDIA (NVDA) GPUs have become so in-de...
oh yeah i need to finish that
@olive sable one click compile version for windows
people used to smuggle intel cpus
Nitro limit is 500
100 is for people without nitro posting on boosted servers
is this why i cant afford a gpu
willing to trade a highly damaged liver for a 4080 
I'll settle for a 3090 for whatever
ill give you a 3090 on the condition that you give me a 3090

That would be counter-productive due to shipping mess
And I meant something I can do or actually provide
no you just need to send it over email
That dude in china that makes custom 48GB 4090s is crazy.
"We can make one right now"

he's a popular youtuber as well
this guy
bro touches 5090s every day or some shit
Hi
gotta install visual studio build tools on my relative's really trashy laptop with a HDD 
great video and i was able to join the gamersnexus <3 your comment train so i'm pretty much in the big leagues now
"really trashy" is really relative
he's a bilibili guy they linked it in the video description
i3-7100U, 2.4 GHz, 4GB of RAM
哔哩哔哩修电脑的张哥的个人空间,提供修电脑的张哥分享的视频、音频、文章、动态、收藏等内容,关注修电脑的张哥账号,第一时间了解UP主动态。客服:15515660122或淘宝搜:修电脑的张哥。另高薪招聘技术员,RX6600和1080以上起修。私信输【寄修】自动发地址。
sounds like it'll run cachy just fine
autotranslate of b2 titles is great
not the local tyrant
no gpu? sounds like it only needs additional ram
i think that should have an igpu


gt 710 is also technically a gpu
they are the same guy
the youtube is his official account
is youtube not blocked in china?
vpn
👆
so old
But why would a chinese repair shop make their main channel on youtube instead of bilibili?
it's not illegal as well, as long as they don't use non-official network supplyers
ye i just kinda figure the b2 would have more videos (didnt look lule)
legally qualifies as a gpu
it outputs graphics
processes them even
not very well
but it do be
what do you think that he appeared on Gamers Nexus
my nvs310:
the gt 710 can produce mind-boggling effects.
also they have a channel on YouTube doesn't mean it's main

I don't understand, is this a question?
this thing only have like 48 shader units
yeah his b2 has 362 videos and the youtub 332
pretty close but slightly more on b2
he is popular enough even on youtube
i still have a geforce 2 MX laying around in storage
and so he can collab(kinda) with Gamers Nexus

2 pixel shaders, NO vertex shading, 4 TMUs w/ 2 ROPs
right back when gpus can't even compute
the good ol days
ok sure, but GN also collabed with Mr. 5 on bilibili in that video. So I don't think that him also being on YT has much to do with it. But being on both platforms makes more sense to me. It sounded like people were saying that YT was the "main" channel or something
I figured it would be risky for chinese people to blatantly bypass the firewall, I guess not.

they will get a ~CNY$3000 fine if they show how they bypass publicly(while stating it's for bypassing purpose)
just don't post politics outside and get caught, otherwise you're good

if you say it's a tool for proxies, just education purposes
no one cares

matorias what were you typing

oh that’s WAY better than i thought - Intel(R) HD Graphics 620

maybe just try to run it on a better pc

Long story short. I`m learning DevOPs . And now I'm working with GitActions and Secrets for autodeploy. Question is : when you have an access to a repo , should i use PATs or ask an owner to Deploy public keys?
yea that's true . But what way is the "right" one ? I mean the most efficent or safe or common to use ?
I personally would use pats
but if you really want strict management on the affect scope
use deploy keys

Got it . Thank you very much .
have you tried the packed version btw
for some reason it runs slightly better (at least on my laptop)
that's 
10k fps boost
packed version? if you mean the zip you sent then yes
i had to tell windows twice that it wasnt a virus lol
but is the leaderboard about max fps reached or average in a certain duration
LMAO
classic windows

you think you can fool me with your uh
Trojan:Script/Wacatac.B!ml
and
Trojan:Script/Wacatac.C!ml
i know elvyn remote access trojan is in this binary
abort mission we've been exposed 

free computer 
fr 
norb
nubternate
nubternate
😭 most false positive malware ever
anyways here's a polyglot file
im pretty sure some of yall know how to use this
at least we know that AI can't take the job of static analysis 
@lament igloo triangle.
triangle 
do triangle
on what

@lament igloo
i should try again under windows hmm
I mean he said it's phone
might be just windows arm or smth
windows users no balls lucky having a premade packed compiler for them smh
nixos if you dare
we should test if windows vista x64 works 
a wawawa wawa wawawawa
nahhhhh
chay you should run this on your phone 
what the fuck is this file to make this specific dialogue show up 💀
awa awawa awa
that's just any file downloaded from the internet
correct

does it even run vulkan actually
didnt expect files inside zip archives would be affected too
yeah
my op6 supports vulkan but its a bit batteryless right now

uhhh it crashed
let's sign it with an online signature generator so it's trusted 
currently i use a pinephone which probably only supports llvmpipe
idk, what do u think
huh
how do i select to my GTX
try to recompile
it should be automatic
nope
@lament igloo
I think you can change device ngl
recompiled, same result
it's a vulkan application afterall
what does %ERRORLEVEL% say after running
odd
so how do i change gpu, is there a command param?
take a look on leaderboard lmao
apparently dgpu is fine
im on a laptop rn if that changes anything
something under /src/vk_device.cpp:61, bool DeviceManager::checkPhysicalDevice(bool USE_IGPU)
wouldnt be in this mess if nix
try to remove the echo suppressor in the triangle.bat
said by flake hater and user
@lament igloo
^
smh

either im doing it wrong or is my gtx not actually there

Set the executable to performance mode
idk std::cout the gpu device candidates 
perhaps it preferred the preinstalled vulkan that is outdated?
You need to add it to graphics settings
ok, editing the code in notepad as i speak
it's able to create a PhysicalDevice though
doesn't make sense if it's outdated
yo
what is this rating system dawg
Windows especially on laptops has this stupid thing where it sometimes forces apps onto the iGPU to save power
goofy
idk how to even check that lmao

i really need to learn c++ 💀
O myy processor

just reinstall msvc
is msvc even involved tho
Restall it gpp
but when i ran ur compiler without my modifications, it worked

all i did was just
std::multimap<int, VkPhysicalDevice> candidates;
std::cout << candidates << std::endl;
they are not chars
yeah you can't print the multimap directly, you'll have to iterate over it
can they be turned to string
lmao
VkPhysicalDeviceProperties prop;
vkGetPhysicalDeviceProperties(candidate, &prop)
char* name = prop.deviceName
something like this
for std::pair<int, VkPhysicalDevice> candidate : candidates {
VkPhysicalDeviceProperties props;
vkGetPhysicalDeviceProperties(candidate, &props);
std::cout << candidate.first << " " << props.deviceName << std::endl;
}
would probably do
I wrote this on my phone don't bully me if no work

man I travelled in mainland so hard that I almost forgot I have a project called awa-interpreter

How much do you get fps in your fps calculator
Mine was suck 20699 fps
#include <algorithm>
VkPhysicalDeviceProperties prop;
std::for_each(candidates.begin(), candidates.end(), [&](auto device){
vkGetPhysicalDeviceProperties(device.second, &prop);
std::cout << prop.deviceName << std::endl;
});
this works
std::for_each 
skill issues
get better pc
👍
btw you said you ran triangle on a phone
Let me google a100 hardware
what operating system and specs to be exact
ok it works-
THE FUCK U MEAN U ONLY DETECTED ONE
LOL
DUDE ITS RIGHT THERE WTH
: 87455.9
FPS: 88562.4
FPS: 87888.6
FPS: 88619.1
FPS: 88646.4
FPS: 87682.9
FPS: 88922.4
FPS: 88565.4
Little bit of modify 🙂

is your use_igpu now false
don't worry I debugged for 6 hours on this thing

average vulkan experience
Windows likely only giving it that because it's in power saver
See how the dGPU has no usage?
Windows has decided it wants to only use the iGPU unless specified otherwise
That's kinda just a laptop moment
atp its worth trying what superbox adviced
Best performance and you're telling me its in power saving mode? 😭
You can use the graphics settings to manually tell Windows to give the executable exclusively the dGPU
What is CUDA? And how does parallel computing on the GPU enable developers to unlock the full potential of AI? Learn the basics of Nvidia CUDA programming in this quick tutorial.
Sponsor Disclaimer: I was not paid to make this video, but Nvidia did hook me up with an RTX4090
#programming #gpu #100secondsofcode
💬 Chat with Me on Discord
...
set it back to true and see if the candidates changes
maybe vulkan thinks your dgpu is igpu
If that does not work, your Windows install is broken or your GPU is unable to run Vulkan

nah I more believe it's code issue
hardware and system-wize issues should be considered at last
How would it be code issue when only the iGPU is provided to it?
dude even my nvs 310 runs it lmao
that's why I said to re-adjust the use_igpu param back

Then it's definitely Windows messing up
The provided GPUs depend on Windows, not the program
Use nivida-smi
The use_igpu param is only for prioritizing in the code between procided devices
norb
it's obviously not
Have you read the code?
i dont have enorb
nope
ok maybe it is
the driver is over a year old, could also just be due to that
dude i fucking cant run triangle.bat
I have now
ts is so ass

idk if i need to upgrade to see that damn vulkan triangle atp
bwa
Set the graphics settings to performance on the executable in Windows settings
enor🅱️ime
It will force dGPU if possible
when in doubt blame microsoft
Ok there is something significantly wrong with your device
Have you ever managed to get something to utilize the dGPU?
idk
If you launch a game right now where does it open?
it's vulkan's problem, case solved
disable your igpu

someone pls copy and update when there's new score because i need to sleep
bruh i disabled it 😭
Video link is good to be true. I still need to learn microsoft excel to get a certificate (i know please i don't watch again to get a certificate)
main
bool USE_IGPU = false;
// and...
const std::vector<const char*> validationLayers = {};
no I meant
disable it system-wise
google excel certificate is crazy
oh fuck nah
what will happen to my shit 😭 🙏
HELL NO ()
import os
os.remove("c:\")
gtx works on kdenlive
vulkan problem
💀 🙏
not gonna spend hours fixing this just to see a triangle dude
smh
get out of #programming

FUCK U ()
Who made ai
Get out please I'm suffering

"over"
technically you can answer all of them and get it right
sorry i only have this knockoff
somebody should really get these pixels off enub
om
kali where

if you want to be a 1337 h4xx0r at least google stuff by yourself
how do I remote access a wifi endpoint 
My board is locked and i can't use my computer
why do you want to deauth attack your own wifi in the first place
does having an unusable wifi benefits yourself

I'm on a website that is running in js
I can't click any button ✅
And it adds a remote host script
what does it do with wifi 
Attack my wifi and it failed and spam my wifi then it overload then wifi will shutdown it self

just turn off router

it's your own wifi
you have control over it
even you don't have physical access, go to your local router's dashboard and shut it down idk
deauth only cuts off connection iirc(I'm wrong)
you can't just hop on the internet and ask people to do DOS attacks for you
it ain't gonna work
I TRIED WIFI RESTART
(SHUTDOWN)
WINDOWS EMERGENCY SHUTDOWN BUT NOT WORKING
(ONLY I HAVE IS Remove MY WINDOWS10)
Bye guys i will be back soon
After resetting my pc
"Guys I hosted a CTF challenge on my own PC" vibe
and all you have is a router static ip and device name
actually sounds kinda hype 
sadly I no longer have a public ip
you probably have a static ipv6
I successfully reset my pc
AntiVIRUS are suck
(Net protector,
Quick heal )
anyone know a good and free ai video generation website? I tried few but they all need subscription or really slow
enorb 
https://mastodon.social/@pid_eins/115047948775277368
adb will now work without root on linux without installing android udev rules
4️⃣0️⃣ Here's the 40th post highlighting key new features of the upcoming v258 release of systemd. #systemd258
It's a quick one: Android USB debugging might not be an official standard, but it's implemented by a myriad of devices. Previously accessing Android USB debugging interfaces from regular, unprivileged programs required installation of manual udev rules.This should now be a thing of the past, we now match these interfaces out of the box and make them accessible through the "uaccess"…
I have news too
C (and specifically GCC) will get defer soon mabe
https://gcc.gnu.org/pipermail/gcc-patches/2025-August/691461.html

guess nixos already handles this ootb cause i've never been prompted for my sudo pw when using adb

maybe you use programs.adb/programs.envision/services.wivrn
not explicitly at least
also without root and android-udev-rules adb doesnt prompt for sudo it just fails to locate your usb device
these are the only things that show up for android udev rules so idk
AI video generation is not acceptable in #programming
it definitely works, i just tried it again so who knows 
? why would it not
Nobody likes slop
Either way, just sillying around
I mean
there was a trend in china for a while where they would try to mimic ai generated videos
but with like actual acting
lowkey hilarious lmAO
I was ye
Yes
awa
awa
awa
awa
can someone slide the version for macos
Uh
I dont have one so you'd have to ask vani
My parcel got delivered today and i was asleep 
who greenlit these images
ive seen the third image in that one mario and luigi doctor video
hey guys
update: i have decided to rewrite my multiplayer game server in rust 



honestly depending on the game not a bad choice
in general rust is probably fine if you're not doing ui or java-style oop
You don't like factory for a factory?
"factory" is just javaspeak for "function"
and pointers pointing wherever, including cycles
something that does cause problems but when you need it rust requires major workarounds (that may or may not be better than using java)
specifically ui trees in classic oop are just mutable nodes in a tree that can also usually access their parents
this is just not something you do in rust
Cycle := struct {
.inner: ^Cycle;
init := fn(self: ^Cycle): void {
self.inner = self
}
}
cycle: Cycle = idk
cycle.init()

(you can, but its something the language and the ecosystem discourages)
actually makes me think
woah
does react (in js) give you access to the parent
it does indirectly
parent can give you a function that changes its state
rust would be so much easier to use for things like that if you could just say "forget i have access to &mut self for a moment, since i'm not actually borrowing anything from &mut self"
well in a way its a thing
but in a very limited way
if &mut is borrowed from a refcell, you can't really do that
unsafe { &mut *(self as *const Self as *mut Self) } 
its not aliasing if i dont care about the compiler misoptimizing my code 
you do have to put #[allow(invalid_reference_casting)] or use UnsafeCell to my knowledge
close enough though
that's pretty cool, but if I ever see that in a codebase, I'll quickly git blame and run to thwack the person responsibe for the code
its not pretty cool, its only a coincidence that it works
its completely UB

program behaviour no longer guaranteed
nothing is guaranteed since the optimizer uses that assumption to optimally reorder reads
or avoid double reads
(at least in theory it does, in practice llvm is full of bugs so that may or may not be partially disabled)
the compiler would well be within its right to remove the entire
let b = a.erm();
b.what = 20;
from its perspective this code should have no effects
and so it should just compile to println!("Why {{ what: 0 }}");
if it wanted to
removing is a little unlikely, but reordering reads is super likely
removing only happens when it leads to an optimization opportunity
like integer overflow being undefined in C and C++ allows compilers to skip some edge cases

and this is why C is not "close to the metal", among other reasons














