Articles by charles_irl
17

Three types of LLM workloads and how to serve them (modal.com)

1

Host overhead is killing your inference efficiency (modal.com)

3

Quantized Float Exposed (quant.exposed)

46

Against SQL (2021) (scattered-thoughts.net)

2

Length-extension attacks are still a thing (00f.net)

2

The future of Python web services looks GIL-free (baro.dev)

2

Lexical differential highlighting instead of syntax highlighting (wordsandbuttons.online)

1

CReact – JSX for the Cloud (github.com/creact-labs)

38

QUIC and the end of TCP sockets (codemia.io)

1

In C++ modules globally unique module names seem to be unavoidable (nibblestew.blogspot.com)

3

Stupid jj Tricks (arko.net)

5

We reverse-engineered Flash Attention 4 (modal.com)

1

A Tour of eBPF in the Linux Kernel: Observability, Security and Networking (lucavall.in)

7

Categorical Foundations for Cute Layouts (colfax-intl.com)

10

Pocket Casts, You Altered the Deal, So I Will Alter Your App (matthewbrunelle.com)

3

Modal Notebooks: How we built a cloud GPU notebook that boots in seconds (modal.com)

174

Public static void main(String[] args) is dead (mccue.dev)

3

Why Are Event-Driven Systems Hard? (scalablethread.com)

77

Safe C++ proposal is not being continued (sibellavia.lol)

2

The unreasonable effectiveness of modern sort algorithms (github.com/voultapher)

18

Analyzing the memory ordering models of the Apple M1 (sciencedirect.com)

4

Generating diffusion QR codes that work (modal.com)