Performant Chudnovsky Pi on NodeJS

This project provides an accurate and performant means of generating Pi via the Chudnovsky algorithm using NodeJS.

This project is constantly being improved upon.

All significant variations are benchmarked.

A self check occurs on the first require to ensure system stability.

Jason Giedymin <jasong _:-a-t:-_ apache _dot_ org>

Other Improvements

This project helped find a very fast means of memoization in NodeJS/V8

Make sure you have GMP installed on your system

You'll need GMP installed.

NPM Install

npm install underscore bigint microtime

Performance

Variation of pinode exist as pinode#.js where '#' is the variation number.

Pi1#10kDigits Original x 12.92 ops/sec ±6.36% (36 runs sampled)
Pi2#10kDigits References x 14.76 ops/sec ±6.09% (41 runs sampled)
Pi3#10kDigits Matrix x 14.67 ops/sec ±6.22% (40 runs sampled)
Pi4#10kDigits Cleanup x 23.89 ops/sec ±7.94% (44 runs sampled)
Pi5#10kDigits First-class Functions x 24.11 ops/sec ±8.11% (44 runs sampled)
Pi6#10kDigits Custom Caching x 56.54 ops/sec ±6.73% (60 runs sampled)
Pi7#10kDigits Advanced Custom Caching x 56.71 ops/sec ±6.56% (60 runs sampled)
Pi8#10kDigits Global Advanced Custom Caching x 56.54 ops/sec ±6.57% (60 runs sampled)
Pi9#10kDigits Global Advanced Custom Caching to other functions + hit tracking x 1,729 ops/sec ±3.09% (91 runs sampled)

Fastest is Pi9#10kDigits Global Advanced Custom Caching to other functions + hit tracking

In summary, performance went from ~13 ops/sec (of Pi @ 10k digits) to 1750 ops/sec.