Computers, High-Performance Algorithms and Software
This is a partial index of my work over the years relating to high-performance computation. This includes my own software efforts and my investigations into the history of computer architecture and benchmarking.
Xmorphia PDE4 : The CPU-oriented implementation of Gray-Scott reaction-diffusion.
The Xmorphia PDE5 Screensaver performs the same calculations on the GPU, and uses it as a video processing filter for the video input (web cam).
Minimal RNN Implementation in Python : A recurrent neural network that models plain text, based on this gist by Andrej Karpathy (@karpathy), but greatly enhanced. I also have a mirror of his RecurrentJS demo.
Computer History and Benchmarking
The SPEC Benchmarks : This page explains each edition of the SPEC CPU benchmarks (both "speed" and "rate" measures, from 1989 through the present), showing how they are computed and how each can be compared to any of the others.
My Functional Computation language is based on a set of recursive definitions starting with a minimal set (three) LISP-like functions. It is meant to be fairly easy to understand, executable on real machines as well as Turing machines and applicable to Goedel numbering.
My annotated version of Goedel's first incompleteness theorem.
32-bit CRC algorithms including precomputed tables.
png-csum-fix : Program that recomputes the CRCs in a PNG file; also allows changing colour table (palette) entries on the fly.
This page was written in the "embarrassingly readable" markup language RHTF, and was last updated on 2018 Aug 27. s.11