Update README

Author: YingboMa

README.md

@@ -5,33 +5,18 @@
 
 ---
 
-`RecursiveFactorization.jl` is a package that collects various recursive matrix factorization algorithms.
+`RecursiveFactorization.jl` is a package that collects various recursive matrix
+factorization algorithms.
 
 #### Implemented Algorithms:
 
-- Sivan Toledo's recursive left-looking LU algorithm. DOI: [10.1137/S0895479896297744](https://epubs.siam.org/doi/10.1137/S0895479896297744)
+- Sivan Toledo's recursive left-looking LU algorithm. DOI:
+  [10.1137/S0895479896297744](https://epubs.siam.org/doi/10.1137/S0895479896297744)
 
 #### Performance:
 
-For medium sized mat
-```julia
-julia> using BenchmarkTools
+For small to medium sized matrices, it is beneficial to use
+`RecursiveFactorization` over `OpenBLAS`. The benchmark script is available in
+`perf/lu.jl`
 
-julia> import LinearAlgebra, RecursiveFactorization
-
-julia> A = rand(40, 40);
-
-julia> @btime RecursiveFactorization.lu($A);
-  23.762 μs (19 allocations: 26.50 KiB)
-
-julia> @btime LinearAlgebra.lu($A);
-  63.846 μs (3 allocations: 13.05 KiB)
-
-julia> A = rand(80, 80);
-
-julia> @btime RecursiveFactorization.lu($A);
-  91.609 μs (42 allocations: 102.95 KiB)
-
-julia> @btime LinearAlgebra.lu($A);
-  420.470 μs (4 allocations: 50.83 KiB)
-```
+![lubench](https://user-images.githubusercontent.com/17304743/53050761-1714e800-3468-11e9-916a-148fbb4fbbf8.png)