transform to switch case for ppu_tick()

1 files changed, 321 insertions, 114 deletions

diff --git a/ppu.c b/ppu.c
index aff5166..02de73a 100644
--- a/ppu.c
+++ b/ppu.c
@@ -17,7 +17,8 @@ static void ppu_reset(struct nes_state *state) {
 	memset(ppu, 0, sizeof(struct ppu_state));
 }
 
-static void ppu_write(struct nes_state *state, uint32_t offset, uint8_t value) {
+__attribute__((always_inline, hot))
+static inline void ppu_write(struct nes_state *state, uint32_t offset, uint8_t value) {
 	struct ppu_state *ppu = &state->ppu;
 
 	switch(offset & 7) {
@@ -73,7 +74,7 @@ static void ppu_write(struct nes_state *state, uint32_t offset, uint8_t value) {
 				state->mapper.ciram_write(state, addr, value);
 			} else if(addr < 0x4000) {
 				uint32_t pal_addr = addr & 0x1f;
-				if((pal_addr & 0x13) == 0x10) {
+				if((pal_addr & 3) == 0) {
 					pal_addr &= ~0x10;
 				}
 				ppu->palette[pal_addr] = value;
@@ -83,7 +84,8 @@ static void ppu_write(struct nes_state *state, uint32_t offset, uint8_t value) {
 	}
 }
 
-static uint8_t ppu_read(struct nes_state *state, uint32_t offset) {
+__attribute__((always_inline, hot))
+static inline uint8_t ppu_read(struct nes_state *state, uint32_t offset) {
 	struct ppu_state *ppu = &state->ppu;
 	uint8_t result = ppu->open_bus;
 
@@ -122,7 +124,8 @@ static uint8_t ppu_read(struct nes_state *state, uint32_t offset) {
 	return result;
 }
 
-static void ppu_evaluate_sprites(struct nes_state *state) {
+__attribute__((always_inline, hot))
+static inline void ppu_evaluate_sprites(struct nes_state *state) {
 	struct ppu_state *ppu = &state->ppu;
 	uint8_t sprite_height = (ppu->reg_ctrl & 0x20) ? 16 : 8;
 	uint8_t n = 0;
@@ -155,7 +158,8 @@ static void ppu_evaluate_sprites(struct nes_state *state) {
 	ppu->sprite_count = n;
 }
 
-static void ppu_fetch_sprite_patterns(struct nes_state *state) {
+__attribute__((always_inline, hot))
+static inline void ppu_fetch_sprite_patterns(struct nes_state *state) {
 	struct ppu_state *ppu = &state->ppu;
 	for(uint8_t i = 0; i < ppu->sprite_count; i++) {
 		uint8_t *s = ppu->secondary_oam + i * 4;
@@ -194,7 +198,8 @@ static void ppu_fetch_sprite_patterns(struct nes_state *state) {
 	}
 }
 
-static void ppu_render_pixel(struct nes_state *state) {
+__attribute__((always_inline, hot))
+static inline void ppu_render_pixel(struct nes_state *state) {
 	uint8_t bg_pixel = 0;
 	uint8_t bg_palette = 0;
 	uint8_t sp_pixel = 0;
@@ -256,172 +261,374 @@ static void ppu_render_pixel(struct nes_state *state) {
 	ppu->pixels[y * 256 + x] = final_color;
 }
 
+
 __attribute__((hot, flatten))
 static void ppu_tick(struct nes_state *state) {
 	struct ppu_state *ppu = &state->ppu;
 
-	uint32_t dot = ppu->dot;
-	uint32_t scanline = ppu->scanline;
+	for(uint32_t ppu_loops = 0; ppu_loops < 3; ++ppu_loops) {
+
+		uint32_t dot = ppu->dot;
+		uint32_t scanline = ppu->scanline;
+
+		uint8_t rendering = (ppu->reg_mask & 0x18);
+
+#if 1
+if(rendering) {
+	switch(scanline) {
+		case 0 ... 239: {
+			// if(rendering && (dot >= 1 && dot <= 256)) {
+			// }
+
+			switch(dot) {
+				case 1 ... 256:
+					ppu_render_pixel(state);
+
+
+					if(dot == 256) {
+						if((ppu->vram_addr & 0x7000) != 0x7000) {
+							ppu->vram_addr += 0x1000;
+						} else {
+							ppu->vram_addr &= ~0x7000;
+							uint32_t y = (ppu->vram_addr & 0x03e0) >> 5;
+							if(y == 29) {
+								y = 0;
+								ppu->vram_addr ^= 0x0800;
+							} else if(y == 31) {
+								y = 0;
+							} else {
+								y++;
+							}
+							ppu->vram_addr = (ppu->vram_addr & ~0x03e0) | (y << 5);
+						}
+					}
 
-	uint8_t rendering = (ppu->reg_mask & 0x18) != 0;
+				__attribute__((fallthrough));
+				case 321 ... 336: {
+
+					if(rendering && ((dot >= 1 && dot <= 256) || (dot >= 321 && dot <= 336))) {
+						if(ppu->reg_mask & 0x10) {
+							for(uint32_t i = 0; i < ppu->sprite_count; i++) {
+								if(ppu->sprite_positions[i] > 0) {
+									ppu->sprite_positions[i]--;
+								} else {
+									ppu->sprite_shift_lo[i] <<= 1;
+									ppu->sprite_shift_hi[i] <<= 1;
+								}
+							}
+						}
+
+						ppu->bg_shift_pattern_low <<= 1;
+						ppu->bg_shift_pattern_high <<= 1;
+						ppu->bg_shift_attrib_low <<= 1;
+						ppu->bg_shift_attrib_high <<= 1;
+					}
 
-	if(rendering && scanline < 240 && dot >= 1 && dot <= 256) {
-		ppu_render_pixel(state);
-	}
+					switch(dot % 8) {
+						case 1: {
+							uint32_t nt_addr = 0x2000 | (ppu->vram_addr & 0x0fff);
+							ppu->bg_next_tile_id = state->mapper.ciram_read(state, nt_addr);
+						} break;
+						case 3: {
+							uint32_t attr_addr = 0x23c0 | (ppu->vram_addr & 0x0c00) |
+														((ppu->vram_addr >> 4) & 0x38) |
+														((ppu->vram_addr >> 2) & 0x07);
+							uint8_t attr = state->mapper.ciram_read(state, attr_addr & 0x0fff);
+							uint8_t shift = ((ppu->vram_addr >> 4) & 4) | (ppu->vram_addr & 2);
+							ppu->bg_next_tile_attrib = (attr >> shift) & 3;
+						} break;
+						case 5: {
+							uint32_t base = (ppu->reg_ctrl & 0x10) ? 0x1000 : 0x0000;
+							uint32_t tile = ppu->bg_next_tile_id;
+							uint32_t fine_y = (ppu->vram_addr >> 12) & 7;
+							uint32_t addr_lsb = (base + tile * 16 + fine_y) & 0x1fff;
+							ppu->bg_next_tile_lsb = state->mapper.chr_read(state, addr_lsb);
+						} break;
+						case 7: {
+							uint32_t base = (ppu->reg_ctrl & 0x10) ? 0x1000 : 0x0000;
+							uint32_t tile = ppu->bg_next_tile_id;
+							uint32_t fine_y = (ppu->vram_addr >> 12) & 7;
+							uint32_t addr_msb = (base + tile * 16 + fine_y + 8) & 0x1fff;
+							ppu->bg_next_tile_msb = state->mapper.chr_read(state, addr_msb);
+
+						} break;
+						case 0: {
+							ppu->bg_shift_pattern_low = (ppu->bg_shift_pattern_low & 0xff00) | ppu->bg_next_tile_lsb;
+							ppu->bg_shift_pattern_high = (ppu->bg_shift_pattern_high & 0xff00) | ppu->bg_next_tile_msb;
+
+							uint8_t a = ppu->bg_next_tile_attrib;
+							ppu->bg_shift_attrib_low = (ppu->bg_shift_attrib_low & 0xff00) | ((a & 1) ? 0xff : 0x00);
+							ppu->bg_shift_attrib_high = (ppu->bg_shift_attrib_high & 0xff00) | ((a & 2) ? 0xff : 0x00);
+
+							if((ppu->vram_addr & 0x001f) == 31) {
+								ppu->vram_addr &= ~0x001f;
+								ppu->vram_addr ^= 0x0400;
+							} else {
+								ppu->vram_addr++;
+							}
+						} break;
+					}
+				} break;
 
-	if(rendering && ((dot >= 2 && dot <= 257) || (dot >= 322 && dot <= 337))) {
+				case 257: {
+					ppu->vram_addr = (ppu->vram_addr & ~0x041f) | (ppu->temp_addr & 0x041f);
+					ppu_evaluate_sprites(state);
+					break;
+				}
 
-		if(ppu->reg_mask & 0x10) {
-			for(uint32_t i = 0; i < ppu->sprite_count; i++) {
-				if(ppu->sprite_positions[i] > 0) {
-					ppu->sprite_positions[i]--;
-				} else {
-					ppu->sprite_shift_lo[i] <<= 1;
-					ppu->sprite_shift_hi[i] <<= 1;
+				case 340: {
+					ppu_fetch_sprite_patterns(state);
+					break;
 				}
 			}
-		}
+		} break;
 
-		ppu->bg_shift_pattern_low <<= 1;
-		ppu->bg_shift_pattern_high <<= 1;
-		ppu->bg_shift_attrib_low <<= 1;
-		ppu->bg_shift_attrib_high <<= 1;
+		// case 241: {
+		// 	if(dot == 1) {
+		// 		ppu->reg_status |= 0x80;
+		// 		if(ppu->reg_ctrl & 0x80) {
+		// 			state->nmi_pending = 1;
+		// 		}
+		// 	}
+		// } break;
+
+		// case 261: {
+		// 	if(dot == 1) {
+		// 		ppu->reg_status &= ~0x80;
+		// 		ppu->reg_status &= ~0x40;
+		// 		ppu->sprite_zero_hit_possible = 0;
+		// 	}
+
+		// 	if(dot >= 280 && dot <= 304) {
+		// 		ppu->vram_addr = (ppu->vram_addr & ~0x7be0) | (ppu->temp_addr & 0x7be0);
+		// 	}
+
+		// 	if(dot == 340) {
+		// 		ppu_fetch_sprite_patterns(state);
+		// 	}
+		// } break;
+
+		// // Handle the frame rendering
+		// if(++dot > 340) {
+		// 	dot = 0;
+		// 	scanline++;
+		// 	if(scanline > 261) {
+		// 		scanline = 0;
+		// 		ppu->frame_ready = 1;
+		// 	}
+		// }
+
+		// ppu->dot = dot;
+		// ppu->scanline = scanline;
 	}
+}
+
+#else
+
+		// if(ppu->even_frame && (ppu->reg_mask & 0x18)) {
+		// 	// skip this dot
+		// 	// call mapper_tick here.
+		// 	ppu->dot++;
+		// }
+
+
+		if(rendering && scanline < 240 && dot >= 1 && dot <= 256) {
+			ppu_render_pixel(state);
+		}
 
-	if(scanline < 240 || scanline == 261) {
 		if(rendering && ((dot >= 1 && dot <= 256) || (dot >= 321 && dot <= 336))) {
-			switch(dot % 8) {
-				case 1: {
-					uint32_t nt_addr = 0x2000 | (ppu->vram_addr & 0x0fff);
-					ppu->bg_next_tile_id = state->mapper.ciram_read(state, nt_addr);
-					break;
+
+			if(ppu->reg_mask & 0x10) {
+				for(uint32_t i = 0; i < ppu->sprite_count; i++) {
+					if(ppu->sprite_positions[i] > 0) {
+						ppu->sprite_positions[i]--;
+					} else {
+						ppu->sprite_shift_lo[i] <<= 1;
+						ppu->sprite_shift_hi[i] <<= 1;
+					}
 				}
-				case 3: {
-					uint32_t attr_addr = 0x23c0 | (ppu->vram_addr & 0x0c00) | ((ppu->vram_addr >> 4) & 0x38) | ((ppu->vram_addr >> 2) & 0x07);
-					uint8_t attr = state->mapper.ciram_read(state, attr_addr & 0x0fff);
-					uint8_t shift = ((ppu->vram_addr >> 4) & 4) | (ppu->vram_addr & 2);
-					ppu->bg_next_tile_attrib = (attr >> shift) & 3;
-					break;
+			}
+
+			ppu->bg_shift_pattern_low <<= 1;
+			ppu->bg_shift_pattern_high <<= 1;
+			ppu->bg_shift_attrib_low <<= 1;
+			ppu->bg_shift_attrib_high <<= 1;
+		}
+
+
+		if(scanline < 240 || scanline == 261) {
+			if(rendering && ((dot >= 1 && dot <= 256) || (dot >= 321 && dot <= 336))) {
+				switch(dot % 8) {
+					case 1: {
+						uint32_t nt_addr = 0x2000 | (ppu->vram_addr & 0x0fff);
+						ppu->bg_next_tile_id = state->mapper.ciram_read(state, nt_addr);
+						break;
+					}
+					case 3: {
+						uint32_t attr_addr = 0x23c0 | (ppu->vram_addr & 0x0c00) | ((ppu->vram_addr >> 4) & 0x38) | ((ppu->vram_addr >> 2) & 0x07);
+						uint8_t attr = state->mapper.ciram_read(state, attr_addr & 0x0fff);
+						uint8_t shift = ((ppu->vram_addr >> 4) & 4) | (ppu->vram_addr & 2);
+						ppu->bg_next_tile_attrib = (attr >> shift) & 3;
+						break;
+					}
+					case 5: {
+						uint32_t base = (ppu->reg_ctrl & 0x10) ? 0x1000 : 0x0000;
+						uint32_t tile = ppu->bg_next_tile_id;
+						uint32_t fine_y = (ppu->vram_addr >> 12) & 7;
+						uint32_t addr_lsb = (base + tile * 16 + fine_y) & 0x1fff;
+						ppu->bg_next_tile_lsb = state->mapper.chr_read(state, addr_lsb);
+						break;
+					}
+					case 7: {
+						uint32_t base = (ppu->reg_ctrl & 0x10) ? 0x1000 : 0x0000;
+						uint32_t tile = ppu->bg_next_tile_id;
+						uint32_t fine_y = (ppu->vram_addr >> 12) & 7;
+						uint32_t addr_msb = (base + tile * 16 + fine_y + 8) & 0x1fff;
+						ppu->bg_next_tile_msb =  state->mapper.chr_read(state, addr_msb);
+						break;
+					}
+					case 0: {
+						ppu->bg_shift_pattern_low = (ppu->bg_shift_pattern_low & 0xff00) | ppu->bg_next_tile_lsb;
+						ppu->bg_shift_pattern_high = (ppu->bg_shift_pattern_high & 0xff00) | ppu->bg_next_tile_msb;
+
+						uint8_t a = ppu->bg_next_tile_attrib;
+						ppu->bg_shift_attrib_low = (ppu->bg_shift_attrib_low & 0xff00) | ((a & 1) ? 0xff : 0x00);
+						ppu->bg_shift_attrib_high = (ppu->bg_shift_attrib_high & 0xff00) | ((a & 2) ? 0xff : 0x00);
+
+						if((ppu->vram_addr & 0x001f) == 31) {
+							ppu->vram_addr &= ~0x001f;
+							ppu->vram_addr ^= 0x0400;
+						} else {
+							ppu->vram_addr++;
+						}
+
+						break;
+					}
 				}
-				case 5: {
-					uint32_t base = (ppu->reg_ctrl & 0x10) ? 0x1000 : 0x0000;
-					uint32_t tile = ppu->bg_next_tile_id;
-					uint32_t fine_y = (ppu->vram_addr >> 12) & 7;
-					uint32_t addr_lsb = (base + tile * 16 + fine_y) & 0x1fff;
-					ppu->bg_next_tile_lsb = state->mapper.chr_read(state, addr_lsb);
-					break;
+			}
+
+			if(rendering) {
+				if(dot == 256) {
+					if((ppu->vram_addr & 0x7000) != 0x7000) {
+						ppu->vram_addr += 0x1000;
+					} else {
+						ppu->vram_addr &= ~0x7000;
+						uint32_t y = (ppu->vram_addr & 0x03e0) >> 5;
+						if(y == 29) {
+							y = 0;
+							ppu->vram_addr ^= 0x0800;
+						} else if(y == 31) {
+							y = 0;
+						} else {
+							y++;
+						}
+						ppu->vram_addr = (ppu->vram_addr & ~0x03e0) | (y << 5);
+					}
 				}
-				case 7: {
-					uint32_t base = (ppu->reg_ctrl & 0x10) ? 0x1000 : 0x0000;
-					uint32_t tile = ppu->bg_next_tile_id;
-					uint32_t fine_y = (ppu->vram_addr >> 12) & 7;
-					uint32_t addr_msb = (base + tile * 16 + fine_y + 8) & 0x1fff;
-					ppu->bg_next_tile_msb =  state->mapper.chr_read(state, addr_msb);
-					break;
+
+				if(dot == 257) {
+					ppu->vram_addr = (ppu->vram_addr & ~0x041f) | (ppu->temp_addr & 0x041f);
 				}
-				case 0: {
-					ppu->bg_shift_pattern_low = (ppu->bg_shift_pattern_low & 0xff00) | ppu->bg_next_tile_lsb;
-					ppu->bg_shift_pattern_high = (ppu->bg_shift_pattern_high & 0xff00) | ppu->bg_next_tile_msb;
 
-					uint8_t a = ppu->bg_next_tile_attrib;
-					ppu->bg_shift_attrib_low = (ppu->bg_shift_attrib_low & 0xff00) | ((a & 1) ? 0xff : 0x00);
-					ppu->bg_shift_attrib_high = (ppu->bg_shift_attrib_high & 0xff00) | ((a & 2) ? 0xff : 0x00);
+				if(UNLIKELY(scanline == 261) && dot >= 280 && dot <= 304) {
+					ppu->vram_addr = (ppu->vram_addr & ~0x7be0) | (ppu->temp_addr & 0x7be0);
+				}
 
-					if((ppu->vram_addr & 0x001f) == 31) {
-						ppu->vram_addr &= ~0x001f;
-						ppu->vram_addr ^= 0x0400;
-					} else {
-						ppu->vram_addr++;
-					}
+				if(dot == 257 && LIKELY(scanline < 240)) {
+					ppu_evaluate_sprites(state);
+				}
 
-					break;
+				if(dot == 340 && (LIKELY(scanline < 240) || UNLIKELY(scanline == 261))) {
+					ppu_fetch_sprite_patterns(state);
 				}
 			}
 		}
+#endif
 
-		if(rendering) {
-			if(dot == 256) {
-				if((ppu->vram_addr & 0x7000) != 0x7000) {
-					ppu->vram_addr += 0x1000;
-				} else {
-					ppu->vram_addr &= ~0x7000;
-					uint32_t y = (ppu->vram_addr & 0x03e0) >> 5;
-					if(y == 29) {
-						y = 0;
-						ppu->vram_addr ^= 0x0800;
-					} else if(y == 31) {
-						y = 0;
-					} else {
-						y++;
-					}
-					ppu->vram_addr = (ppu->vram_addr & ~0x03e0) | (y << 5);
+
+
+// TEST SWITCH CODE
+
+	switch(scanline) {
+		case 241: {
+			if(dot == 1) {
+				ppu->reg_status |= 0x80;
+				if(ppu->reg_ctrl & 0x80) {
+					state->nmi_pending = 1;
 				}
 			}
+			break;
+		}
 
-			if(dot == 257) {
-				ppu->vram_addr = (ppu->vram_addr & ~0x041f) | (ppu->temp_addr & 0x041f);
+		case 261: {
+			if(dot == 1) {
+				ppu->reg_status &= ~0x80;
+				ppu->reg_status &= ~0x40;
+				ppu->sprite_zero_hit_possible = 0;
 			}
 
-			if(scanline == 261 && dot == 304) { //>= 280 && dot <= 304) {
+			if(dot >= 280 && dot <= 304) {
 				ppu->vram_addr = (ppu->vram_addr & ~0x7be0) | (ppu->temp_addr & 0x7be0);
 			}
 
-			if(dot == 257 && scanline < 240) {
-				ppu_evaluate_sprites(state);
-			}
-
-			if(dot == 340 && scanline < 240) {
+			if(dot == 340) {
 				ppu_fetch_sprite_patterns(state);
 			}
+			break;
 		}
 	}
 
-	if(scanline == 241 && dot == 1) {
-		ppu->reg_status |= 0x80;
-		if(ppu->reg_ctrl & 0x80) {
-			state->nmi_pending = 1;
-		}
-	}
-
-	if(scanline == 261 && dot == 1) {
-		ppu->reg_status &= ~0x80;
-		ppu->reg_status &= ~0x40;
-		ppu->sprite_zero_hit_possible = 0;
-	}
 
-	dot++;
-	if(dot > 340) {
-		dot = 0;
-		scanline++;
-		if(scanline > 261) {
-			scanline = 0;
-			ppu->frame_ready = 1;
+// TEST SWITCH CODE
+
+		// if(UNLIKELY(scanline == 241) && dot == 1) {
+		// 	ppu->reg_status |= 0x80;
+		// 	if(ppu->reg_ctrl & 0x80) {
+		// 		state->nmi_pending = 1;
+		// 	}
+		// }
+
+		// if(UNLIKELY(scanline == 261) && dot == 1) {
+		// 	ppu->reg_status &= ~0x80;
+		// 	ppu->reg_status &= ~0x40;
+		// 	ppu->sprite_zero_hit_possible = 0;
+		// }
+
+		dot++;
+		if(dot > 340) {
+			dot = 0;
+			scanline++;
+			if(scanline > 261) {
+				scanline = 0;
+				ppu->frame_ready = 1;
+			}
 		}
-	}
 
-	ppu->dot = dot;
-	ppu->scanline = scanline;
+		ppu->dot = dot;
+		ppu->scanline = scanline;
+	}
 }
 
-static void ppu_dma_4014(struct nes_state *state, uint8_t page) {
+__attribute__((always_inline, hot))
+static inline void ppu_dma_4014(struct nes_state *state, uint8_t page) {
 	uint32_t base = page << 8;
 
 	// Add 1 or 2 idle cycles depending on current CPU cycle
 	uint8_t idle_cycles = (state->cycles & 1) ? 1 : 2;
 	for(uint8_t i = 0; i < idle_cycles; i++) {
 		state->cycles++;
-		ppu_tick(state); ppu_tick(state); ppu_tick(state);
+		ppu_tick(state);
 	}
 
 	for(uint32_t i = 0; i < 256; i++) {
 		uint32_t addr = base + i;
 
 		state->cycles++;
-		ppu_tick(state); ppu_tick(state); ppu_tick(state);
+		ppu_tick(state);
 		uint8_t value = memory_read_dma(state, addr);
 
 		state->cycles++;
-		ppu_tick(state); ppu_tick(state); ppu_tick(state);
+		ppu_tick(state);
 		// ppu_write_2004(state, value);
 		ppu_write(state, 4, value);
 	}