columnar layout is only faster if reads are sequential | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		antpls on May 2, 2021 \| parent \| context \| favorite \| on: The compiler will optimize that away columnar layout is only faster if reads are sequential

cma on May 3, 2021 [–]

If things work one or a few columns at a time this isn't necessarily true. Prosumer CPU L3 caches are already up to 128MB and will be up to 1GB before long. If one or a few full columns fits fully in cache but all the data doesn't, columnar layout may still be faster in the face of random access within the column.

Consider applying for YC's Winter 2026 batch! Applications are open till Nov 10
Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact