DeepRead
Cover of Data-oriented design

Data-oriented design

Mr Richard Fabian · 2018

Programming and computer science › Software design and clean code

#3,167 most recommended

About the book

Data-oriented design is an approach to writing software that starts from how data is laid out and accessed in memory, rather than from object hierarchies and abstractions. The book's scope is performance-sensitive software such as games and simulations, and it covers the hardware factors, including caches, memory bandwidth and branch behaviour, that shape how fast code runs. Its central claim is that the cost of a program is dominated by how data moves through the memory hierarchy, so a designer should model the data and the transformations applied to it before the code that wraps them. It argues that conventional object-oriented designs, with scattered allocations and per-object behaviour, waste cache capacity and make hardware-friendly optimisation hard to add later. It presents techniques such as splitting fields into contiguous arrays, grouping work by transformation, and reducing indirection, and it judges a design by how well it matches the actual data flow.

Written by Claude from what it knows of this book.

What Hacker News says 100% recommend · 0% critical

02.55201920202022202420262026 peak · 4
This bookAverage book in Software design and clean code

In the commenters’ own words

What Open Library readers say

No Open Library reader has rated it yet.

  • 2 want to read
  • 0 reading
  • 1 have read

Open Library’s page for the book

What critics say, via Wikipedia

No Wikipedia article about this book yet, so no critics to quote.

Reddit communities — coming later

Details

First published
2018
ISBN
9781916478701
Editions known
2 at Open Library
Open Library
OL26362741W

Explore in the mosaic

Read next: its neighbours on the mosaic