Optimization / Performance cleanup (#264)

* Implemented all changes from #230 and #221

* Fixed off by one mentioned in #216

* Wrapped modulo in i32Portable as mentioned in #216

* issue #207, Allowed making closure builds for debugging, and then tried
to match its inlining

* Removed the legacy api

* Updated the package-lock for the branch

core/core.ts

@@ -184,6 +184,7 @@ export function isGBC(): i32 {
 // Function to return an address to store into save state memory
 // this is to regulate our 20 slots
 // https://docs.google.com/spreadsheets/d/17xrEzJk5-sCB9J2mMJcVnzhbE-XH_NvczVSQH9OHvRk/edit?usp=sharing
+// Inlined because closure compiler inlines
 export function getSaveStateMemoryOffset(offset: i32, saveStateSlot: i32): i32 {
   // 50 bytes per save state memory partiton sli32
   return WASMBOY_STATE_LOCATION + offset + 50 * saveStateSlot;

core/cpu/cbOpcodes.ts

@@ -19,6 +19,7 @@ import { u8Portable, u16Portable } from '../portable/portable';
 // Handle CB Opcodes
 // NOTE: Program stpes and cycles are standardized depending on the register type
 // NOTE: Doing some funny stuff to get around not having arrays or objects
+// Inlined because closure compiler inlines.
 export function handleCbOpcode(cbOpcode: i32): i32 {
   let numberOfCycles: i32 = -1;
   let handledOpcode: boolean = false;

core/cpu/cpu.ts

@@ -152,6 +152,7 @@ export class Cpu {
   }
 }
 
+// Inlined because closure compiler does so
 export function initializeCpu(): void {
   // Reset all stateful Cpu variables
   // Cpu.GBCEnabled is done by core/initialize

core/cpu/instructions.ts

@@ -165,6 +165,7 @@ export function cpARegister(register: u8): void {
   setSubtractFlag(1);
 }
 
+// Inlined because closure compiler inlines
 export function rotateRegisterLeft(register: u8): u8 {
   // RLC register 8-bit
   // Z 0 0 C
@@ -188,6 +189,7 @@ export function rotateRegisterLeft(register: u8): u8 {
   return register;
 }
 
+// Inlined because closure compiler inlines
 export function rotateRegisterRight(register: u8): u8 {
   // RLC register 8-bit
   // Z 0 0 C
@@ -212,6 +214,7 @@ export function rotateRegisterRight(register: u8): u8 {
   return register;
 }
 
+// Inlined because closure compiler inlines
 export function rotateRegisterLeftThroughCarry(register: u8): u8 {
   // RL register 8-bit
   // Z 0 0 C
@@ -240,6 +243,7 @@ export function rotateRegisterLeftThroughCarry(register: u8): u8 {
   return register;
 }
 
+// Inlined because closure compiler inlines
 export function rotateRegisterRightThroughCarry(register: u8): u8 {
   // RR register 8-bit
   // Z 0 0 C
@@ -267,6 +271,7 @@ export function rotateRegisterRightThroughCarry(register: u8): u8 {
   return register;
 }
 
+// Inlined because closure compiler inlines
 export function shiftLeftRegister(register: u8): u8 {
   // SLA register 8-bit
   // Z 0 0 C
@@ -295,6 +300,7 @@ export function shiftLeftRegister(register: u8): u8 {
   return register;
 }
 
+// Inlined because closure compiler inlines
 export function shiftRightArithmeticRegister(register: u8): u8 {
   // SRA register 8-bit
   // Z 0 0 C
@@ -334,6 +340,7 @@ export function shiftRightArithmeticRegister(register: u8): u8 {
   return register;
 }
 
+// Inlined because closure compiler inlines
 export function swapNibblesOnRegister(register: u8): u8 {
   // SWAP register 8-bit
   // Z 0 0 0
@@ -354,6 +361,7 @@ export function swapNibblesOnRegister(register: u8): u8 {
   return register;
 }
 
+// Inlined because closure compiler inlines
 export function shiftRightLogicalRegister(register: u8): u8 {
   // SRA register 8-bit
   // Z 0 0 C

core/cycles.ts

@@ -31,6 +31,7 @@ export function getCycles(): i32 {
   return Cycles.cycles;
 }
 
+// Inlined because closure compiler inlines
 function trackCyclesRan(numberOfCycles: i32): void {
   Cycles.cycles += numberOfCycles;
   if (Cycles.cycles >= Cycles.cyclesPerCycleSet) {
@@ -39,6 +40,7 @@ function trackCyclesRan(numberOfCycles: i32): void {
   }
 }
 
+// Inlined because closure compiler inlines
 export function resetCycles(): void {
   Cycles.cyclesPerCycleSet = 2000000000;
   Cycles.cycleSets = 0;

core/execute.ts

@@ -36,6 +36,7 @@ export function getSteps(): i32 {
   return Execute.steps;
 }
 
+// Inlined because closure compiler inlines
 function trackStepsRan(steps: i32): void {
   Execute.steps += steps;
   if (Execute.steps >= Execute.stepsPerStepSet) {
@@ -44,6 +45,7 @@ function trackStepsRan(steps: i32): void {
   }
 }
 
+// Inlined because closure compiler inlines
 export function resetSteps(): void {
   Execute.stepsPerStepSet = 2000000000;
   Execute.stepSets = 0;

core/graphics/backgroundWindow.ts

@@ -18,8 +18,12 @@ import { TileCache, drawPixelsFromLineOfTile, getTileDataAddress } from './tiles
 import { eightBitLoadFromGBMemory } from '../memory/load';
 import { Memory } from '../memory/memory';
 import { hexLog, checkBitOnByte, setBitOnByte, resetBitOnByte } from '../helpers/index';
-import { u8Portable } from '../portable/portable';
+import { u8Portable, i32Portable } from '../portable/portable';
 
+// NOTE: i32Portable wraps modulo here as somehow it gets converted to a double:
+// https://github.com/torch2424/wasmboy/issues/216
+
+// Inlined because closure compiler inlines
 export function renderBackground(scanlineRegister: i32, tileDataMemoryLocation: i32, tileMapMemoryLocation: i32): void {
   // NOTE: Camera is reffering to what you can see inside the 160x144 viewport of the entire rendered 256x256 map.
 
@@ -45,6 +49,7 @@ export function renderBackground(scanlineRegister: i32, tileDataMemoryLocation:
   drawBackgroundWindowScanline(scanlineRegister, tileDataMemoryLocation, tileMapMemoryLocation, pixelYPositionInMap, 0, scrollX);
 }
 
+// Inlined because closure compiler inlines
 export function renderWindow(scanlineRegister: i32, tileDataMemoryLocation: i32, tileMapMemoryLocation: i32): void {
   // Get our windowX and windowY
   // let windowX: i32 = eightBitLoadFromGBMemory(Graphics.memoryLocationWindowX);
@@ -67,7 +72,9 @@ export function renderWindow(scanlineRegister: i32, tileDataMemoryLocation: i32,
   let pixelYPositionInMap: i32 = scanlineRegister - windowY;
 
   // xOffset is simply a neagative window x
-  let xOffset: i32 = -1 * windowX;
+  // NOTE: This can become negative zero?
+  // https://github.com/torch2424/wasmboy/issues/216
+  let xOffset: i32 = i32Portable(-1 * windowX);
 
   // Draw the Background scanline
   drawBackgroundWindowScanline(scanlineRegister, tileDataMemoryLocation, tileMapMemoryLocation, pixelYPositionInMap, windowX, xOffset);
@@ -178,6 +185,7 @@ function drawBackgroundWindowScanline(
 }
 
 // Function to draw a pixel for the standard GB
+// Inlined because closure compiler inlines
 function drawMonochromePixelFromTileId(
   xPixel: i32,
   yPixel: i32,
@@ -203,7 +211,7 @@ function drawMonochromePixelFromTileId(
   // yPixel = 144. 144 % 8 = 0.
   // 0 Represents last line of pixels in a tile, 1 represents first. 1 2 3 4 5 6 7 0.
   // Because remember, we are counting lines on the display NOT including zero
-  let pixelYInTile: i32 = pixelYPositionInMap % 8;
+  let pixelYInTile: i32 = i32Portable(pixelYPositionInMap % 8);
 
   // Remember to represent a single line of 8 pixels on a tile, we need two bytes.
   // Therefore, we need to times our modulo by 2, to get the correct line of pixels on the tile.
@@ -217,7 +225,7 @@ function drawMonochromePixelFromTileId(
   // Therefore, is pixelX was 2, then really is need to be 5
   // So 2 - 7 = -5, * 1 = 5
   // Or to simplify, 7 - 2 = 5 haha!
-  let pixelXInTile: i32 = pixelXPositionInMap % 8;
+  let pixelXInTile: i32 = i32Portable(pixelXPositionInMap % 8);
   pixelXInTile = 7 - pixelXInTile;
 
   // Now we can get the color for that pixel
@@ -259,6 +267,7 @@ function drawMonochromePixelFromTileId(
 
 // Function to draw a pixel from a tile in C O L O R
 // See above for more context on some variables
+// Inlined because closure compiler inlines
 function drawColorPixelFromTileId(
   xPixel: i32,
   yPixel: i32,
@@ -284,7 +293,7 @@ function drawColorPixelFromTileId(
   let bgMapAttributes: i32 = loadFromVramBank(tileMapAddress, 1);
 
   // See above for explanation
-  let pixelYInTile: i32 = pixelYPositionInMap % 8;
+  let pixelYInTile: i32 = i32Portable(pixelYPositionInMap % 8);
   if (checkBitOnByte(6, bgMapAttributes)) {
     // We are mirroring the tile, therefore, we need to opposite byte
     // So if our pixel was 0 our of 8, it wild become 7 :)
@@ -303,7 +312,7 @@ function drawColorPixelFromTileId(
 
   // Get our X pixel. Need to NOT reverse it if it was flipped.
   // See above, you have to reverse this normally
-  let pixelXInTile: i32 = pixelXPositionInMap % 8;
+  let pixelXInTile: i32 = i32Portable(pixelXPositionInMap % 8);
   if (!checkBitOnByte(5, bgMapAttributes)) {
     pixelXInTile = 7 - pixelXInTile;
   }
@@ -348,6 +357,7 @@ function drawColorPixelFromTileId(
 }
 
 // Function to attempt to draw the tile from the tile cache
+// Inlined because closure compiler inlines
 function drawLineOfTileFromTileCache(
   xPixel: i32,
   yPixel: i32,
@@ -406,7 +416,7 @@ function drawLineOfTileFromTileCache(
   // Calculate when we should do the tileCache calculation again
   if (xPixel >= TileCache.nextXIndexToPerformCacheCheck) {
     TileCache.nextXIndexToPerformCacheCheck = xPixel + 8;
-    let xOffsetTileWidthRemainder: i32 = pixelXPositionInMap % 8;
+    let xOffsetTileWidthRemainder: i32 = i32Portable(pixelXPositionInMap % 8);
     if (xPixel < xOffsetTileWidthRemainder) {
       TileCache.nextXIndexToPerformCacheCheck += xOffsetTileWidthRemainder;
     }
@@ -417,6 +427,7 @@ function drawLineOfTileFromTileCache(
 
 // Function to draw a line of a tile in Color
 // This is for tile rendering shortcuts
+// Inlined because closure compiler inlines
 function drawLineOfTileFromTileId(
   xPixel: i32,
   yPixel: i32,
@@ -427,7 +438,7 @@ function drawLineOfTileFromTileId(
   tileIdFromTileMap: i32
 ): i32 {
   // Get the which line of the tile we are rendering
-  let tileLineY: i32 = pixelYPositionInMap % 8;
+  let tileLineY: i32 = i32Portable(pixelYPositionInMap % 8);
 
   // Now lets find our tileX start and end
   // This is for the case where i = 0, but scroll X was 3.

core/graphics/colors.ts

@@ -50,6 +50,7 @@ export class Colors {
   static obj1Black: i32 = WasmBoyGBColors.obj1Black;
 }
 
+// Inlined because closure compiler inlines
 export function initializeColors(): void {
   setManualColorizationPalette(0);
 
@@ -310,6 +311,7 @@ export function setManualColorizationPalette(colorizationId: i32): void {
 // By checksum of the title
 // https://forums.nesdev.com/viewtopic.php?f=20&t=10226
 // TODO: torch2424 need to find how to get the "disambiguation"
+// Inlined because closure compiler inlines
 export function setHashColorizationPalette(hash: i32): void {
   switch (hash) {
     case 0x88:

core/graphics/graphics.ts

@@ -140,6 +140,7 @@ export function batchProcessGraphics(): void {
   }
 }
 
+// Inlined because closure compiler inlines
 export function initializeGraphics(): void {
   // Reset Stateful Variables
   Graphics.currentCycles = 0;
@@ -295,6 +296,7 @@ function _renderEntireFrame(): void {
 }
 
 // Function to get the start of a RGB pixel (R, G, B)
+// Inlined because closure compiler inlines
 export function getRgbPixelStart(x: i32, y: i32): i32 {
   // Get the pixel number
   // let pixelNumber: i32 = (y * 160) + x;

core/graphics/lcd.ts

@@ -109,6 +109,7 @@ function resetLcd(shouldBlankScreen: boolean): void {
 }
 
 // Pass in the lcd status for performance
+// Inlined because closure compiler inlines
 export function setLcdStatus(): void {
   // Check if the Lcd was disabled
   if (!Lcd.enabled) {

core/graphics/palette.ts

@@ -19,6 +19,7 @@ export class Palette {
   static readonly memoryLocationSpritePaletteTwo: i32 = 0xff49;
 }
 
+// Inlined because closure compiler inlines
 export function initializePalette(): void {
   if (Cpu.GBCEnabled) {
     // GBC Palettes
@@ -38,6 +39,7 @@ export function initializePalette(): void {
 // Simple get pallete color or monochrome GB
 // shouldRepresentColorByColorId is good for debugging tile data for GBC games that don't have
 // monochromePalettes
+// Inlined because closure compiler inlines
 export function getMonochromeColorFromPalette(
   colorId: i32,
   paletteMemoryLocation: i32,
@@ -135,6 +137,7 @@ export function getColorizedGbHexColorFromPalette(colorId: i32, paletteMemoryLoc
   return hexColor;
 }
 
+// Inlined because closure compiler inlines
 export function writeColorPaletteToMemory(offset: i32, value: i32): void {
   // FF68
   //  Bit 0-5   Index (00-3F)
@@ -160,6 +163,7 @@ export function writeColorPaletteToMemory(offset: i32, value: i32): void {
 // Functions to Handle Write to pallete data registers
 // http://gbdev.gg8.se/wiki/articles/Video_Display#FF68_-_BCPS.2FBGPI_-_CGB_Mode_Only_-_Background_Palette_Index
 // Function to handle incrementing the pallete index if required
+// Inlined because closure compiler inlines
 function incrementPaletteIndexIfSet(paletteIndex: i32, offset: i32): void {
   // Check ther auto increment box
   if (checkBitOnByte(7, paletteIndex)) {
@@ -217,6 +221,7 @@ export function loadPaletteByteFromWasmMemory(paletteIndexByte: i32, isSprite: b
 }
 
 // Function to store a byte to our Gbc Palette memory
+// Inlined because closure compiler inlines
 export function storePaletteByteInWasmMemory(paletteIndexByte: i32, value: i32, isSprite: boolean): void {
   // Clear the top two bits to just get the bottom palette Index
   let paletteIndex: i32 = paletteIndexByte & 0x3f;

core/graphics/priority.ts

@@ -14,10 +14,12 @@ export function addPriorityforPixel(x: i32, y: i32, colorId: i32 = 0, hasGbcBgPr
   store<u8>(BG_PRIORITY_MAP_LOCATION + getPixelStart(x, y), <u8>bgPriorityByte);
 }
 
+// Inlined because closure compiler inlines
 export function getPriorityforPixel(x: i32, y: i32): u8 {
   return load<u8>(BG_PRIORITY_MAP_LOCATION + getPixelStart(x, y));
 }
 
+// Inlined because closure compiler inlines
 export function clearPriorityMap(): void {
   for (let y: i32 = 0; y < 144; y++) {
     for (let x: i32 = 0; x < 160; x++) {
@@ -26,6 +28,7 @@ export function clearPriorityMap(): void {
   }
 }
 
+// Inlined because closure compiler inlines
 function getPixelStart(x: i32, y: i32): i32 {
   // Get the pixel number
   return y * 160 + x;

core/graphics/sprites.ts

@@ -17,6 +17,7 @@ import { getPriorityforPixel } from './priority';
 import { eightBitLoadFromGBMemory } from '../memory/load';
 import { checkBitOnByte, setBitOnByte, resetBitOnByte, hexLog } from '../helpers/index';
 
+// Inlined because closure compiler inlines
 export function renderSprites(scanlineRegister: i32, useLargerSprites: boolean): void {
   // Need to loop through all 40 sprites to check their status
   // Going backwards since lower sprites draw over higher ones

core/graphics/tiles.ts

@@ -23,6 +23,7 @@ export class TileCache {
   static nextXIndexToPerformCacheCheck: i32 = -1;
 }
 
+// Inlined because closure compiler inlines
 export function resetTileCache(): void {
   TileCache.tileId = -1;
   TileCache.nextXIndexToPerformCacheCheck = -1;
@@ -155,6 +156,7 @@ export function drawPixelsFromLineOfTile(
   return pixelsDrawn;
 }
 
+// Inlined because closure compiler inlines
 export function getTilePixelStart(outputLineX: i32, outputLineY: i32, outputWidth: i32): i32 {
   // Finally Lets place a pixel in memory
   let pixelStart: i32 = outputLineY * outputWidth + outputLineX;

core/index.ts

@@ -86,23 +86,3 @@ export {
 export { getLY, drawBackgroundMapToWasmMemory, drawTileDataToWasmMemory, drawOamToWasmMemory } from './debug/debug-graphics';
 export { getDIV, getTIMA, getTMA, getTAC } from './debug/debug-timer';
 export { updateDebugGBMemory } from './debug/debug-memory';
-export {
-  update,
-  emulationStep,
-  getAudioQueueIndex,
-  resetAudioQueue,
-  wasmMemorySize,
-  wasmBoyInternalStateLocation,
-  wasmBoyInternalStateSize,
-  gameBoyInternalMemoryLocation,
-  gameBoyInternalMemorySize,
-  videoOutputLocation,
-  frameInProgressVideoOutputLocation,
-  gameboyColorPaletteLocation,
-  gameboyColorPaletteSize,
-  backgroundMapLocation,
-  tileDataMap,
-  soundOutputLocation,
-  gameBytesLocation,
-  gameRamBanksLocation
-} from './legacy';