Ash Vardanian personal blog

Python, C, Assembly – Faster Cosine Similarity

From pure Python to AVX-512 assembly, optimizing cosine similarity reveals a 2,500x speedup through SIMD, FP16, and VNNI instructions on modern Intel CPUs.
Points
203
HN comments
55
Submissions
1
Cited by
0 commenters

Software EngineeringSystems & Performance2023

On Hacker News

SubmittedTitle on HNPointsComments
Dec 18, 2023Python, C, Assembly – Faster Cosine Similarity by ashvardanian20355

More from Ash Vardanian

All 5 posts →

Related posts from the same era