Pranjal

prev gpus @xai

[Pranjal @pranjalssh]

I implemented H100 cuda matmul kernel from scratch, taking inspiration from @Si_Boehm's blog.

Our final kernel outperforms cuBLAS by 7% for N=4096. It fits in a single C++ file without any dependencies.

Full-blown blog post with all details:

https://t.co/NlE0fgoJKS

[Pranjal @pranjalssh]

@elliotarledge Idea is to fix more bugs than it creates

[Pranjal @pranjalssh]

Who's working on LLMs making puzzles by themselves?

[Mechanize @MechanizeWork]

We are now seeking a puzzle maker to help us create puzzles that LLMs can't yet solve.

[Pranjal @pranjalssh]

@real_deep_ml Claude can already do this

[Pranjal @pranjalssh]

Hard times create strong engs
Strong engs create compilers
Compilers create weak engs
Weak engs create hard times

[tender @tenderizzation]

we have become weaker as a species

[Pranjal @pranjalssh]

@ellev3n11 Use subagents aggressively and do long running tasks

[Pranjal @pranjalssh]

@ellev3n11 Use subagents aggressively and do long running tasks

[Pranjal @pranjalssh]

@maharshii Once you go ptx, there's no way back.

[Pranjal @pranjalssh]

@HeinrichKuttler @pmddomingos next blog?

[Pranjal @pranjalssh]

@HeinrichKuttler Daddy Heiner teaching us how this works

[Pranjal @pranjalssh]

@HeinrichKuttler Ok I'll bite

[Pranjal @pranjalssh]

This is a very common misconception. Biggest wins are overlapping.

If you have 2 memory bound kernels back to back reading and writing 1GB, you pay 4GB of HBM cost. If you overlap them while streaming intermediate state of 128MB(fits in L2 cache) - you only pay 2GB HBM cost.

For communication overlapping, benefits are even higher.

[henry tsang @henrylhtsang]

so how much perf gain does megakernel give, if the baseline already uses cuda graphs

[Pranjal @pranjalssh]

Congrats on the launch! SGLang's vision is more ambitious than anyone can imagine

[RadixArk @radixark]

Today, we are thrilled to officially launch RadixArk with $100M in Seed funding at a $400M valuation. The round was led by @Accel and co-led by @sparkcapital.

RadixArk exists to make frontier AI infrastructure open and accessible to everyone. Today, the systems behind the most capable AI models are concentrated in a small number of companies. As a result, most AI teams are forced to rebuild training and inference stacks from scratch, duplicating the same infrastructure work instead of focusing on new models, products, and ideas.

RadixArk was founded to change that. We are building an AI platform that makes it easier for teams to train and serve the best models at scale.

RadixArk comes from the open-source community. We started with SGLang, where many of us are core developers and maintainers, and expanded our work to Miles for large-scale RL and post-training. We will continue contributing to both projects and working with the community to make them the strongest open-source infrastructure foundations for frontier AI.

We would like to thank our long-term partners, contributors, and the broader SGLang community for believing in this mission. We're also grateful to @Accel and @sparkcapital, NVentures (Venture capital arm of @nvidia), Salience Capital, A&E Investment, @HOFCapital, @walden_catalyst, @AMD, LDVP, WTT Fubon Family, @MediaTek, Vocal Ventures, @Sky9Capital and our angel investors @ibab, @LipBuTan1, Hock Tan, @johnschulman2, @soumithchintala, @lilianweng, @oliveur, @Thom_Wolf, @LiamFedus, @robertnishihara, @ericzelikman, @OfficialLoganK, and @multiply_matrix among others.

Thanks for the exclusive interview with @MeghanBobrowsky at @WSJ about our vision.

[Pranjal @pranjalssh]

@HeinrichKuttler Congrats daddy

[Pranjal @pranjalssh]

Best candidates don't use python

[François Chollet @fchollet]

When looking at deep learning profiles, one of the most obvious tells between a mediocre and great candidate is whether they list PyTorch or JAX.

[Pranjal @pranjalssh]

@Guodzh End of an era. GGs

[Guodong Zhang @Guodzh]

Last day at xAI.

Wild journey past three years but excited about next chapter.

Thanks all for the love and support yesterday. So many friends made along the way and I will miss you all!

[Pranjal @pranjalssh]

@dylan522p Attention is all you need