Wᴀʟʟꜰʟᴏᴡᴇʀ 🥀 (@autom8nerd) header
Wᴀʟʟꜰʟᴏᴡᴇʀ 🥀 (@autom8nerd) profile picture
Brodmann area 10 & 46
Joined October 2024
Photo by Wᴀʟʟꜰʟᴏᴡᴇʀ 🥀 (@autom8nerd)Photo by Wᴀʟʟꜰʟᴏᴡᴇʀ 🥀 (@autom8nerd)Photo by Wᴀʟʟꜰʟᴏᴡᴇʀ 🥀 (@autom8nerd)Photo by Wᴀʟʟꜰʟᴏᴡᴇʀ 🥀 (@autom8nerd)Photo by Wᴀʟʟꜰʟᴏᴡᴇʀ 🥀 (@autom8nerd)Photo by Wᴀʟʟꜰʟᴏᴡᴇʀ 🥀 (@autom8nerd)Photo by Wᴀʟʟꜰʟᴏᴡᴇʀ 🥀 (@autom8nerd)Photo by Wᴀʟʟꜰʟᴏᴡᴇʀ 🥀 (@autom8nerd)Photo by Wᴀʟʟꜰʟᴏᴡᴇʀ 🥀 (@autom8nerd)Photo by Wᴀʟʟꜰʟᴏᴡᴇʀ 🥀 (@autom8nerd)
I am Alone Musk.
1
5
211
Why tf am I still verified?
1
3
159
thanks to all those neurotypicals, who made me feel shitty about myself....fuck you!
3
63
And here, my integral part completes....
62
well, we are back st square one. and hopefully this time, i will leave stronger.
1
5
132
Life has been such a turmoil lately. I don't wanna live but hey, @CounterStrike is there for me so yeah.
1
4
89
now i wanna break into quant. wtf is wrong with me
3
57
So afraid what the people might say, but that's okay since you all are humans...
2
61
it's high time i started learning to implement game theory and dp. it's weird. this problem is weirder than me. lmao. (i miss posting on X)
1
6
147
I believe macos vs Linux has always been an endless debate just like that of learning high level language (like python) or C as a beginner. Nonetheless, I prefer Linux, and I always tell people to learn C as their first language.
Linux is the best OS to start as a developer.
 Windows hides the system.
macOS is too polished and locked down.
 Linux forces you to learn the terminal, processes, and how computers really work.
3
147
Wᴀʟʟꜰʟᴏᴡᴇʀ 🥀 retweeted
making "brain" of engine. started building evaluation func so it actually understands the positions instead of blindly counting nodes. how does it thinks ? +ve score = white winnin -ve = black winnin 2 main things -> Material value (P=100 , Q=900) -> Positional sense using PSTs
made my golang chess engine 3.5x faster by implementing Zobrist Hashing and a Transposition table (TT) . now it can calclulate depth 6 with 119M nodes in just 12.4s down from 44s. here is the breakdown of how i did it . a thread
4
3
1
33
1,316
i'm in dire need of accountability 🥀 and looking back at my posts where I posted logs, it kinda look cringe because i never really did something meaningful. i somehow worked for the sake of logs and nothing else. and now i'm feeling bad. If only I had a friend to check on me daily...
1
4
231
peak larp on X. I'm tired of this.
1
69
Wᴀʟʟꜰʟᴏᴡᴇʀ 🥀 retweeted
Just got my first PR merged into OpenBLAS (OpenMathLib org). Where I added matrix transpose support to GEADD, the routine that computes C = alpha·A + beta·C. Earlier it only supported no-transpose. If you wanted A or C transposed, you had to physically transpose your matrix yourself before calling it. Now it takes a transpose parameter, you just tell it which matrix (A, C, both, or neither) you want transposed, and it handles it internally. The actual logic I implemented lives in the kernel/generic for real and complex number both single and double precision,instead of doing a physical transpose (allocating new memory, copying every element into a rearranged layout) I made it do a logical transpose. Given the matrix is stored in row-major order, transposing it doesn't require touching a single byte of memory I just made it walk through a vertical pointer stride. A pointer that strides vertically instead of horizontally reads the exact same memory as if it were transposed, without taking extra space. I want to clear that, I wasn't contributing through any org or program. I found this as an open issue on the repo. Replied on the thread asking the maintainer if I could pick it up, got a pretty relaxed reply and 4 days later I have merged it successfully. However, It was a bit frustratating CI broke on a queue timeout on one of the architecture tests (RISC-V QEMU emulation), completely unrelated to my actual code. Had to dig through the logs myself, confirm it was an infra timeout and not my bug, and tell the maintainer directly it wasn't coming from my end and it turned out to be a known issue, there was literally another PR that week fixing that exact timeout. Though my contribution is a Small feature on paper, but it touches every precision variant OpenBLAS supports (float, double, complex float, complex double) and runs all the way from the public API down to raw pointer arithmetic in the kernel. First real merged contribution to a widely-used numerical library. Here's the Pr Link PR: github.com/OpenMathLib/OpenB…
9
7
1
58
4,973
my first submission in the daily challenge T-T
2
82
can I even become great on X?
2
4
229
I really feel bad for the one who gifted me premium 🥀
1
4
120