Chris Elrod

Simple experimental wrapper of small number of Apple Accelerate linear algebra routines using libblastrampoline. Largely used for benchmarking routines using the Apple M1's matrix instructions.

chriselrod

This library provides arrays with columns padded to be a multiple of SIMD-vector width.

chriselrod

Probability models with `y ~ Normal(mu, sigma)` style syntax and reverse mode AD for gradients; WIP. Prioritizes runtime performance.

chriselrod

Pirating base type NTuple{N,Core.VecElement{T}} -- although base methods are NOT overloaded -- and plundering the great work of eschnett's SIMD.jl

chriselrod