GY-63_MS5611/libraries/randomHelpers
2024-04-13 10:35:57 +02:00
..
.github bulk update GitHub actions 2024-04-13 10:35:57 +02:00
examples 0.2.6 randomHelpers 2023-02-26 13:36:39 +01:00
test 0.2.6 randomHelpers 2023-02-26 13:36:39 +01:00
.arduino-ci.yml 0.2.5 randomHelpers 2022-11-23 13:56:17 +01:00
CHANGELOG.md 0.2.7 randomHelpers 2023-11-16 20:20:16 +01:00
library.json 0.2.7 randomHelpers 2023-11-16 20:20:16 +01:00
library.properties 0.2.7 randomHelpers 2023-11-16 20:20:16 +01:00
LICENSE bulk update GitHub actions 2024-04-13 10:35:57 +02:00
randomHelpers.cpp 0.2.7 randomHelpers 2023-11-16 20:20:16 +01:00
randomHelpers.h 0.2.7 randomHelpers 2023-11-16 20:20:16 +01:00
README.md 0.2.7 randomHelpers 2023-11-16 20:20:16 +01:00

Arduino CI Arduino-lint JSON check GitHub issues

License: MIT GitHub release PlatformIO Registry

randomHelpers

Arduino library for faster generation of random numbers.

Description

This library contains functions that have the goal to deliver random bits faster than the build in random function can, while still using it.

The idea is to have a buffer ( __randomBuffer) which can hold up to 32 bits. When a number of random bits are needed, these are first fetched from the buffer and if the buffer gets empty, it is filled again with a call to the random generator.

This strategy works well with a 32 bits buffer and requests for 1..16 random bits. However above 16 bits the overhead is larger than the gain. So to improve in that range too one could use a faster random function like the one from Marsaglia (included).

Note the gains differ per platform and are more explicit on the (slow) Arduino UNO platform than on a much faster ESP32.

Interface

#include "randomHelpers.h"

generators (sort of)

  • uint32_t Marsaglia() fast PRNG.
  • bool seedMarsaglia(uint32_t a, uint32_t b) seed the Marsaglia PRNG. a and b should not be 0. returns true on success.

getters

  • bool getRandom1() returns 0 or 1, false or true.
  • uint8_t getRandom2() returns 0 .. 3.
  • uint8_t getRandom3() returns 0 .. 7.
  • uint8_t getRandom4() returns 0 .. 15.
  • uint8_t getRandom5() returns 0 .. 31.
  • uint8_t getRandom6() returns 0 .. 63.
  • uint8_t getRandom7() returns 0 .. 127.
  • uint8_t getRandom8() returns 0 .. 255 typically a byte.
  • uint16_t getRandom16() returns 0 .. 65535 (2 bytes).
  • uint32_t getRandom24() returns 0 .. 16777215 (3 bytes), e.g. random RGB colour.
  • uint32_t getRandom32() returns 0 .. 2^32 - 1 (4 bytes) this is the core random generator
  • uint64_t getRandom64() returns 0.. 2^64 - 1 (8 bytes).
  • uint32_t getRandomBits(n) returns 0.. 2^n - 1 This works well for 1..16 bits but above 16 it is slower than the standard way.

Typical wrappers.

  • bool flipCoin() A wrapper around getRandom1().
  • uint8_t throwDice() returns 1..6.

The examples show how to use these and how their performance gain relative to calling random() for every random number.

Performance

To elaborate.

Operation

See examples.

Future

Must

  • improve/update documentation
    • add performance figures
  • wrap all up in a class.
    • rename getRandom64() ==> get64() etc.

Should

  • improve performance getRandomBits(n) for n = 17..31
    • how to preserve bits if idx too small.
  • add JKISS? other RNG's
  • test if the functions are uniform.

Could

  • improve performance getRandomBits(n) for n = 17..31
  • investigate new tricks :)
  • add getRandom9() 10() can be done 3x from 32 bits.
  • 11..16 => 2x from 32 bits
  • add getRandom12() clipping get16 is equally fast.

Wont

  • RANDOM_HELPERS_LIB_VERSION?

Support

If you appreciate my libraries, you can support the development and maintenance. Improve the quality of the libraries by providing issues and Pull Requests, or donate through PayPal or GitHub sponsors.

Thank you,