parallel ldiv!

using SparseExtra, LinearAlgebra, SparseArrays, BenchmarkTools
const n = 10_000
const A = sprandn(n, n, 5 / n);
const C = A + I
const B = Matrix(sprandn(n, n, 1 / n));
const F = lu(C);
const X = similar(B);

Standard:

@benchmark ldiv!($X, $F, $B)

BenchmarkTools.Trial: 1 sample with 1 evaluation.
 Single result which took 96.309 s (0.00% GC) to evaluate,
 with a memory estimate of 0 bytes, over 0 allocations.

With FLoops.jl:

@benchmark par_solve!($X, $F, $B)

BenchmarkTools.Trial: 1 sample with 1 evaluation.
 Single result which took 25.195 s (0.00% GC) to evaluate,
 with a memory estimate of 1.24 MiB, over 163 allocations.

with manual loops

@benchmark par_ldiv!_t($X, $F, $B)

BenchmarkTools.Trial: 1 sample with 1 evaluation.
 Single result which took 25.395 s (0.00% GC) to evaluate,
 with a memory estimate of 1.24 MiB, over 130 allocations.

This page was generated using Literate.jl.