/simd-studies

📊 Comparing C Intel vector intrinsics vs. Numba + LLVM's auto-vectorization

Primary LanguageC

Benchmarking SIMD

Comparison of how fast Numba's autovectorization can achieve with respect to Intel C vector intrinsics. Read my blog post for more details on my experience working with Numba! https://laurenar.net/posts/python-simd/

Array Sum

array_sum

Array Sum with Filter

array_sum_filter1

Evaluation

Benchmarked 2 different programs to examine how logical complexity impacts performance.

  1. Sum of all elements in an array
  2. Sum of elements in array if element >= 100

Each program runs over a length 65546 integer array. Times are averaged over 1000s of iterations.