Faster position lookup

[atesgoral]

Perhaps this doesn't pose an issue since running compiled code on the ESP32 is fast enough, but here's a suggestion, at least for helping with maintainability. Take it with a grain of salt 😅

Calling findPosition() to search for a pixel index during the render loop is inefficient and makes it hard to understand how the LED pixels are laid out:

ikea-led-obegraensad/src/screen.cpp

Line 164 in 6edc2ed

digitalWrite(PIN_DATA, this->renderBuffer_[this->findPosition(row * 16 + col)]);

You could instead change the positions array to be the search results of findPosition().

Here's a fiddle that does this: https://jsfiddle.net/atesgoral/njkx749g/14/

It uses a new lookup table called remapped:

uint8_t remapped[ROWS * COLS] = {
  0x0f, 0x0e, 0x0d, 0x0c, 0x0b, 0x0a, 0x09, 0x08, 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f,
  0x07, 0x06, 0x05, 0x04, 0x03, 0x02, 0x01, 0x00, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17,
  0x27, 0x26, 0x25, 0x24, 0x23, 0x22, 0x21, 0x20, 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37,
  0x2f, 0x2e, 0x2d, 0x2c, 0x2b, 0x2a, 0x29, 0x28, 0x38, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f,
  0x4f, 0x4e, 0x4d, 0x4c, 0x4b, 0x4a, 0x49, 0x48, 0x58, 0x59, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f,
  0x47, 0x46, 0x45, 0x44, 0x43, 0x42, 0x41, 0x40, 0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57,
  0x67, 0x66, 0x65, 0x64, 0x63, 0x62, 0x61, 0x60, 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77,
  0x6f, 0x6e, 0x6d, 0x6c, 0x6b, 0x6a, 0x69, 0x68, 0x78, 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f,
  0x8f, 0x8e, 0x8d, 0x8c, 0x8b, 0x8a, 0x89, 0x88, 0x98, 0x99, 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f,
  0x87, 0x86, 0x85, 0x84, 0x83, 0x82, 0x81, 0x80, 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97,
  0xa7, 0xa6, 0xa5, 0xa4, 0xa3, 0xa2, 0xa1, 0xa0, 0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7,
  0xaf, 0xae, 0xad, 0xac, 0xab, 0xaa, 0xa9, 0xa8, 0xb8, 0xb9, 0xba, 0xbb, 0xbc, 0xbd, 0xbe, 0xbf,
  0xcf, 0xce, 0xcd, 0xcc, 0xcb, 0xca, 0xc9, 0xc8, 0xd8, 0xd9, 0xda, 0xdb, 0xdc, 0xdd, 0xde, 0xdf,
  0xc7, 0xc6, 0xc5, 0xc4, 0xc3, 0xc2, 0xc1, 0xc0, 0xd0, 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7,
  0xe7, 0xe6, 0xe5, 0xe4, 0xe3, 0xe2, 0xe1, 0xe0, 0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7,
  0xef, 0xee, 0xed, 0xec, 0xeb, 0xea, 0xe9, 0xe8, 0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe, 0xff
};

Your render loop just can be:

  for (uint8_t idx = 0; row < ROWS * COLS; idx++)
  {
    digitalWrite(PIN_DATA, this->renderBuffer_[remapped[idx]]);
    digitalWrite(PIN_CLOCK, HIGH);
    digitalWrite(PIN_CLOCK, LOW);
  }

This might require moving rotations to the render loop first.

[letalvoj]

any chance you will open a PR with the proposed changes?