benchmark

Benchmarks for arbitrary precision numbers with Rust implementations. The code is adopted from bigint-benchmark-rs, and see LICENSE for the original license.

Libraries

Library	Version	Notes	Supported Types
dashu	0.4.0	Pure Rust, no_std	integer, float, decimal, rational
rug	1.30.0	Links to libc and GMP	integer, float, rational, complex
rust-gmp	0.5.0	Links to libc and GMP	integer, float, rational
ibig	0.3.6	Pure Rust, no_std	integer
malachite	0.9.2	Pure Rust, LGPL, derived from GMP and FLINT	integer, rational
num	0.4.3	Pure Rust, no_std	integer, rational, complex
ramp	0.7.0	Requires nightly Rust, uses x86_64 assembly	integer
bigdecimal	0.4.10	Pure Rust	decimal
astro-float	0.9.5	Pure Rust, binary float	float (binary)

Tasks

Task	Description	Number Type	Difficulty	Algorithm	Operations
e	n digits of e	Integer	Hard	Binary splitting	addition, multiplication, division, exponentiation, base conversion
e_decimal	n digits of e	Decimal	-	Depends	-
pi	n bits of π	Float (binary)	Medium	Arctangent identity	sqrt, division, atan, multiplication
fib	n-th Fibonnaci number	Integer	Medium	Matrix exponentiation	addition, multiplication, base conversion
fib_hex	n-th Fibonnaci number in hex	Integer	Easy	Matrix exponentiation	addition, multiplication
fib_ratio	n-th modified Fibonnaci number	Rational	-	-	-

Usage examples

• Print results:
  • Integer: cargo run -- --lib dashu --lib num --lib malachite --lib ibig --task e -n 100 print
  • Rational: cargo run -- --lib dashu --lib num --lib malachite --task fib_ratio -n 100 print
  • Decimal Float: cargo run -- --lib dashu --lib bigdecimal --task e_decimal -n 100 print
  • Binary Float (π): cargo run -- --lib dashu --lib astro_float --task pi -n 100 print
  • Binary Float (π, incl. GMP): cargo run --features gmp -- --lib dashu --lib astro_float --lib rug --task pi -n 100 print
• Run the benchmark: change print to exec in the commands above and select a larger n.