sparrow

wrote a blog: CUDA for Newcomers — a Comprehensive Guide. covers everything from scratch. host, device, kernels, memory, thread hierarchy, and a simple .cu program. @xtrupal/cuda-for-newcomers-a-comprehensive-guide-6fa831f26a9f" target="_blank" rel="nofollow noopener">medium.com/@xtrupal/cuda-…

@yacineMTB ur descendants are going to read some of that and feel proud of their great grandfather

jesus almighty

@_vai0 still alot better than mee

Could be better I guess. Still improving this skill 🫣

@star_stufff i read it as "relativity for retarded engineers" lol

lol arxiv.org/abs/2605.21660

@OpenAI it would be so helpful if u guys show the prompt u used

Today, we share a breakthrough on the planar unit distance problem, a famous open question first posed by Paul Erdős in 1946. For nearly 80 years, mathematicians believed the best possible solutions looked roughly like square grids. An OpenAI model has now disproved that belief, discovering an entirely new family of constructions that performs better. This marks the first time AI has autonomously solved a prominent open problem central to a field of mathematics.

@sama cancer, space, fusion

what problem do you most hope AI will solve in the future? maybe we can help!

@karpathy haha wow

Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.

a golden hour view of the Himalayas

don't avoid leg days it's literally half of your body

btw

Emma Chamberlain wearing custom Mugler by Miguel Castro Freitas at the 2026 Met Gala. #MetGala

hardware stack of sbcc26 winners

@ankkala there is a reason why switch is called a switch and not a splitter

@xtrupal isn't that just a switch

search "why it's not possible to make an ethernet splitter" u will learn a lot.

may our biceps get bigger and our shoulders get wider.

engineering is art

this felt like gta6 to me for some reason

@luthiraabeykoon @karpathy thats so cool honestly

We implemented @karpathy 's MicroGPT fully on FPGA fabric. No GPU. No PyTorch. No CPU inference loop. Just a transformer burned into hardware, generating 50,000+ tokens/sec. The model is small, but the idea is not: inference does not have to live only in software 👇

u can build ur own rack u can build ur own server u can build ur own cluster u can build ur own nas u can run ur own hypervisor u can build ur own cloud u can run ur own vpn u can stream ur own media u can run ur local models u can quantize ur own models u can just do things

some combinations of keystrokes can make you a billionaire