Stories by ashvardanian

Unifying Generative and Dense Retrieval for Sequential Recommendation

Efficient In-Place UTF-16 Unicode Correction with ARM Neon

Scaling data collection for training software engineering agents

Linux on GPU

Amazon's AI Self Sufficiency – Trainium2 Architecture and Networking

Outperforming cuBLAS on H100: A Worklog

Sign operation using VFIXUPIMM in AVX-512

Cerebras 748x faster than Frontier supercomputer in molecular simulation

The Next 31 Years of Developing Unum

Understanding SIMD: Infinite Complexity of Trivial Problems

SIMD Library for Evaluating Elementary Functions, Vectorized Libm and DFT

The Microsoft Azure HBv5 and AMD MI300C

Azure HBv5: 352 AMD Genoa Cores with over 400 GB of HBM3

Waiting for many things at once with io_uring

Is Prefix of String in Table? A Journey into SIMD String Processing

Create Index Externally: Offloading Pgvector Indexing from Postgres

Tenstorrent Wormhole Series Part 6: Vector instruction set

Over-Engineering 5x Faster Set Intersections in SVE2, AVX-512, & Neon

1
2
3
4
5
6
7
8
9
10