📦 Code Density - matmat · Scour

KV-CoRE: Benchmarking Data-Dependent Low-Rank Compressibility of KV-Caches in LLMs

arxiv.org·1d

⚡Cache Theory

I Built a 6 BIPS JIT in Five Months

unlikelyemphasis.substack.com·1d·

Discuss: Substack

⚙️RISC-V Microcode

dev.to·1d·

Discuss: DEV

🖥️Modern Terminals

Modern Trends In Floating-Point

semiengineering.com·2d

📐Arithmetic Precision

LLM Inference Benchmarking - Measure What Matters

digitalocean.com·18h

📊Performance Profiling

Balancing FP8 Computation Accuracy and Efficiency on Digital CIM via Shift-Aware On-the-fly Aligned-Mantissa Bitwidth Prediction

arxiv.org·1d

🎯Emulation Accuracy

Info - Intel ARC Alchemist owners thread. Now 75 games tested, retested, and counting. Includes productivity experiences. Feel free to add your results.

forums.anandtech.com·11h

🎯Emulation Accuracy

AI-powered compressed imaging system developed for high-speed scenes

phys.org·1d

⚡Compression Hardware

Both GCC and Clang generate strange/inefficient code

codingmarginalia.blogspot.com·11h·

Discuss: Hacker News, Hacker News

🚀Compiler Optimizations

The GeekBook X14 Pro is one of the lightest laptops I’ve tested, and it packs plenty of power at a reasonable price

techradar.com

·1d

🔌Operating system internals

ChatGPT vs. Logic: Why AI Code is Slower

dev.to·14h·

Discuss: DEV

🌊Streaming Algorithms

Show HN: ZigZag – A Bubble Tea-Inspired TUI Framework for Zig

github.com·6h·

Discuss: Hacker News

Engineering Ethereum's Speed: How we made Ethrex 20x faster

blog.lambdaclass.com·2d

🎯Emulation Accuracy

OSTEP Chapters 6,7

muratbuffalo.blogspot.com·1d·

Discuss: Blogger

⚡CPU Microarchitecture

FreeBSD audio diagnostics and optimization

discoverbsd.com·1d

🔌Operating system internals

AMD's 96-core beast with watercooling engraved into CPU joins car and industrial parts in a 2,000W direct die cooling setup — $12,000 CPU runs at 5.3 GHz, devours 1,300W, and still runs cooler than your gaming PC

tomshardware.com

·21h·

Discuss: Hacker News

🖥️Modern CPU

Why Large Language Models Make Terrible Compilers — And Why the Industry Keeps Trying Anyway

webpronews.com·7h

🏗️Compiler Archaeology

Examining Turbopuffer ANN v3

terencezl.github.io·1d·

Discuss: Hacker News

⚡Compression Hardware

Anthropic's Performance Take-Home: A 65x Optimization (For Dummies)

ikot.blog·3d·

Discuss: Hacker News

🖥️Hardware Architecture

LFortran Compiles Lapack

lfortran.org·23h·

Discuss: Hacker News

🦀Rust Macros

Loading more...