Welcome to my personal blog - a blog empowering developers with tools and how-to to make software faster.

I'm Filipe, a Principal Performance Engineer at Redis Inc (opens new window) and an Open Source Contributor. Opinions expressed here are my own.

Feel free to ping me on twitter (opens new window), github (opens new window), or linkedin (opens new window).

Benchmarking Apple's M1 vector intrinsics SIMD parallelism on common mathematical functions

A couple of weeks ago, during a work conversation about SIMD support on modern ARM processors, I bumped again in the fact that I never had tried ARM's [NEON](https://developer.arm.com/architectures/instruction-sets/simd-isas/neon) instruction set extension. Today was the day.

Hands-on profile driven redudant computation removal

As you will see during this post, even highly optimized software is very likely to perform “redundant” computation during single or consecutive executions. This is either due to way we've designed it (algorithms) or the nature of the values assumed as inputs (dataset).