Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Optimizing Layout of Recursive Datatypes with Marmoset

Vidush Singhal, Chaitanya Koparkar, Joseph Zullo, Artem Pelenitsyn, Michael Vollmer, Mike Rainey, Ryan Newton, Milind Kulkarni

TL;DR

This work introduces Marmoset, a compiler that optimizes the layouts of algebraic datatypes, with a special focus on producing highly optimized, packed data layouts where recursive structures can be traversed with minimal pointer chasing.

Abstract

While programmers know that the low-level memory representation of data structures can have significant effects on performance, compiler support to optimize the layout of those structures is an under-explored field. Prior work has optimized the layout of individual, non-recursive structures without considering how collections of those objects in linked or recursive data structures are laid out. This work introduces Marmoset, a compiler that optimizes the layouts of algebraic datatypes, with a special focus on producing highly optimized, packed data layouts where recursive structures can be traversed with minimal pointer chasing. Marmoset performs an analysis of how a recursive ADT is used across functions to choose a global layout that promotes simple, strided access for that ADT in memory. It does so by building and solving a constraint system to minimize an abstract cost model, yielding a predicted efficient layout for the ADT. Marmoset then builds on top of Gibbon, a prior compiler for packed, mostly-serial representations, to synthesize optimized ADTs. We show experimentally that Marmoset is able to choose optimal layouts across a series of microbenchmarks and case studies, outperforming both Gibbons baseline approach, as well as MLton, a Standard ML compiler that uses traditional pointer-heavy representations.

Optimizing Layout of Recursive Datatypes with Marmoset

TL;DR

This work introduces Marmoset, a compiler that optimizes the layouts of algebraic datatypes, with a special focus on producing highly optimized, packed data layouts where recursive structures can be traversed with minimal pointer chasing.

Abstract

While programmers know that the low-level memory representation of data structures can have significant effects on performance, compiler support to optimize the layout of those structures is an under-explored field. Prior work has optimized the layout of individual, non-recursive structures without considering how collections of those objects in linked or recursive data structures are laid out. This work introduces Marmoset, a compiler that optimizes the layouts of algebraic datatypes, with a special focus on producing highly optimized, packed data layouts where recursive structures can be traversed with minimal pointer chasing. Marmoset performs an analysis of how a recursive ADT is used across functions to choose a global layout that promotes simple, strided access for that ADT in memory. It does so by building and solving a constraint system to minimize an abstract cost model, yielding a predicted efficient layout for the ADT. Marmoset then builds on top of Gibbon, a prior compiler for packed, mostly-serial representations, to synthesize optimized ADTs. We show experimentally that Marmoset is able to choose optimal layouts across a series of microbenchmarks and case studies, outperforming both Gibbons baseline approach, as well as MLton, a Standard ML compiler that uses traditional pointer-heavy representations.
Paper Structure (13 sections, 3 equations, 6 figures, 1 table, 2 algorithms)

This paper contains 13 sections, 3 equations, 6 figures, 1 table, 2 algorithms.

Table of Contents

  1. Introduction
  2. Dense Representation: The Good, The Bad, and The Pointers
  3. Overview
  4. Running Example
  5. Design
  6. Marmoset's Language
  7. Control-Flow Analysis
  8. Data Flow Analysis
  9. Field Attributes For Code Motion
  10. Field Access Pattern Analysis
  11. Finding a Layout
  12. ILP Constraints
  13. Cost Model

Figures (6)

  • Figure 1: Blog traversal motivating example
  • Figure 2: Showing a dense pointer-free layout with ideal accesses on top. Numbers represent the access order. Out of order accesses (red), incur costly extra traversals over fields in the middle.
  • Figure 3: Showing the unoptimized representation with pointers to allow random access and the optimized layout with favorable access patterns.
  • Figure 4: Simplified grammar for a first-order, monomorphic functional language that Marmoset operates on. The full implementation includes more primitive datatypes, and built in primitives (such as a copy primitive).
  • Figure 5: Control-flow and corresponding field access graphs generated for the running example.
  • ...and 1 more figures