transform to switch case for ppu_tick()

15 files changed, 668 insertions, 414 deletions

diff --git a/addmapper.py b/addmapper.py
new file mode 100755
index 0000000..8848e87
--- /dev/null
+++ b/addmapper.py
@@ -0,0 +1,67 @@
+#!/usr/bin/env python3
+import sys
+
+def generate_mapper_files(mapper_id, submapper_id=0):
+    # Format: submapper (1 hex digit) + mapper_id (3 hex digits, zero-padded)
+    full_id = f"{submapper_id:01x}{int(mapper_id):03x}"
+    filename = f"mapper_{full_id}"
+
+    # Generate C file content
+    c_content = f"""
+
+static void {filename}_init(struct nes_state *state) {{
+}}
+
+static uint8_t {filename}_prg_read(struct nes_state *state, uint32_t addr) {{
+}}
+
+static void {filename}_prg_write(struct nes_state *state, uint32_t addr, uint8_t value) {{
+}}
+
+static uint8_t {filename}_chr_read(struct nes_state *state, uint32_t addr) {{
+}}
+
+static void {filename}_chr_write(struct nes_state *state, uint32_t addr, uint8_t value) {{
+}}
+
+static uint8_t {filename}_ciram_read(struct nes_state *state, uint32_t addr) {{
+}}
+
+static void {filename}_ciram_write(struct nes_state *state, uint32_t addr, uint8_t value) {{
+}}
+
+static void {filename}_tick(struct nes_state *state) {{
+}}
+"""
+
+    # Generate H file content
+    h_content = f"""
+
+struct {filename} {{
+}};
+"""
+
+    # Generate table entry
+    table_entry = f"""	{{ 0x{int(mapper_id):x}, {filename}_prg_read, {filename}_prg_write, {filename}_chr_read, {filename}_chr_write, {filename}_ciram_read, {filename}_ciram_write, {filename}_tick, {filename}_init }},"""
+
+    # Write files
+    with open(f"{filename}.c", "w") as f:
+        f.write(c_content)
+
+    with open(f"{filename}.h", "w") as f:
+        f.write(h_content)
+
+    print(f"Generated files: {filename}.c, {filename}.h")
+    print("\nTable entry (copy-paste this into your mapper table):")
+    print(table_entry)
+
+if __name__ == "__main__":
+    if len(sys.argv) < 2:
+        print("Usage: python newmapper.py <mapper_id> [submapper_id=0]")
+        print("Example: python newmapper.py 123 0  # Creates mapper_0123.*")
+        sys.exit(1)
+
+    mapper_id = sys.argv[1]
+    submapper_id = int(sys.argv[2]) if len(sys.argv) > 2 else 0
+    generate_mapper_files(mapper_id, submapper_id)
+
diff --git a/build.sh b/build.sh
index 348700d..5487439 100755
--- a/build.sh
+++ b/build.sh
@@ -11,7 +11,7 @@ CFLAGS+="-fno-stack-protector -fno-PIE -no-pie -fno-strict-aliasing -ffunction-s
 CFLAGS+="-fno-exceptions -fno-rtti -fno-use-cxa-atexit -fno-non-call-exceptions "
 CFLAGS+="-Wall -Wextra "
 CFLAGS+="-Wno-unused-parameter -Wno-sign-compare -Wno-trigraphs -Wno-maybe-uninitialized "
-CFLAGS+="-Wno-unused-variable -Wno-unused-const-variable -Wno-unused-function -Wno-write-strings "
+CFLAGS+="-Wno-unused-variable -Wno-unused-const-variable -Wno-unused-function -Wno-write-strings -Wno-missing-field-initializers "
 
 LDFLAGS="-Wl,--gc-sections "
 
@@ -20,7 +20,7 @@ INCLUDE_PATHS="-Ibase -I.."
 
 # Linux-specific includes and libraries
 LINUX_INCLUDE="-I/usr/include/pipewire-0.3 -I/usr/include/spa-0.2"
-LINUX_LIBS="-lpipewire-0.3 -lXi -lX11 -lGL -lm -ldl -pthread -lglfw"
+LINUX_LIBS="-lpipewire-0.3 -lXi -lX11 -lGL -lm -ldl -pthread -lglfw -larchive"
 
 # Windows-specific includes and libraries
 # WINDOWS_INCLUDE=""
diff --git a/cpu.c b/cpu.c
index eb7c837..263ee6e 100644
--- a/cpu.c
+++ b/cpu.c
@@ -92,13 +92,8 @@ static void cpu_tick(struct nes_state *state) {
 
 	uint8_t opcode;
 
-	// if(cpu->pc <= 0x90cc || cpu->pc >= 0x90e6) {
-	// 	printf("%5.5d %4.4x: ", line++, cpu->pc);
-	// 	opcode = memory_read(state, cpu->pc++);
-	// 	printf("%2.2x   a:%2.2x x:%2.2x y:%2.2x p:%2.2x sp:%2.2x  cycle: %ld\n", opcode, cpu->a, cpu->x, cpu->y, pack_flags(cpu), cpu->sp, state->cycles);
-	// } else {
-		opcode = memory_read(state, cpu->pc++);
-	// }
+	// printf("%5.5d %4.4x: ", line++, cpu->pc);
+	opcode = memory_read(state, cpu->pc++);
+	// printf("%2.2x   a:%2.2x x:%2.2x y:%2.2x p:%2.2x sp:%2.2x  cycle: %ld\n", opcode, cpu->a, cpu->x, cpu->y, pack_flags(cpu), cpu->sp, state->cycles);
 	opcode_lut[opcode](state);
-
 }
diff --git a/ines2.c b/ines2.c
index 9fb743d..dc0e57d 100644
--- a/ines2.c
+++ b/ines2.c
@@ -1,5 +1,6 @@
 
-
+#include <archive.h>
+#include <archive_entry.h>
 
 // iNES header fields
 #define INES_HEADER_SIZE	16
@@ -27,14 +28,89 @@
 #define MIRROR_FOUR_SCREEN	2
 
 
-static int ines2_load(struct nes_state *state, char *path) {
+static uint8_t *ines2_read_entire_file(const char *path, size_t *out_size) {
 	FILE *f = fopen(path, "rb");
-	if(!f) {
+	if(!f) return 0;
+
+	fseek(f, 0, SEEK_END);
+	size_t size = ftell(f);
+	fseek(f, 0, SEEK_SET);
+
+	uint8_t *buffer = (uint8_t *)malloc(size);
+	size_t read_size = fread(buffer, 1, size, f);
+	fclose(f);
+
+	if(read_size != size) {
+		free(buffer);
+		return 0;
+	}
+
+	*out_size = size;
+	return buffer;
+}
+
+uint8_t *ines2_unzip_file_to_memory(const char *zip_path, size_t *out_size) {
+	struct archive *a = archive_read_new();
+	struct archive_entry *entry;
+	uint8_t *buffer = 0;
+	size_t size = 0;
+
+	archive_read_support_format_zip(a);
+
+	if(archive_read_open_filename(a, zip_path, 10240) != ARCHIVE_OK) {
+		fprintf(stderr, "libarchive: failed to open zip: %s\n", archive_error_string(a));
+		archive_read_free(a);
+		return 0;
+	}
+
+	if(archive_read_next_header(a, &entry) != ARCHIVE_OK) {
+		fprintf(stderr, "libarchive: no valid file found in zip\n");
+		archive_read_free(a);
+		return 0;
+	}
+
+	const char *name = archive_entry_pathname(entry);
+
+	size = archive_entry_size(entry);
+	buffer = (uint8_t*)malloc(size);
+	if(!buffer) {
+		fprintf(stderr, "libarchive: malloc failed\n");
+		archive_read_free(a);
+		return 0;
+	}
+
+	ssize_t read_size = archive_read_data(a, buffer, size);
+	if(read_size != (ssize_t)size) {
+		fprintf(stderr, "libarchive: failed to read full file\n");
+		free(buffer);
+		buffer = 0;
+	}
+
+
+	archive_read_free(a);
+
+	if(buffer && out_size) {
+		*out_size = size;
+	}
+	return buffer;
+}
+
+static int ines2_load(struct nes_state *state, const char *path) {
+	uint8_t *data = 0;
+	size_t size = 0;
+
+	if(strstr(path, ".zip")) {
+		data = ines2_unzip_file_to_memory(path, &size);
+	} else {
+		data = ines2_read_entire_file(path, &size);
+	}
+
+	if(!data || size < INES_HEADER_SIZE) {
 		return -1;
 	}
 
-	uint8_t header[INES_HEADER_SIZE];
-	fread(header, 1, INES_HEADER_SIZE, f);
+	uint8_t *ptr = data;
+	uint8_t *header = ptr; ptr += INES_HEADER_SIZE;
 
 	uint8_t prg_lsb = header[INES_PRG_SIZE_LSB];
 	uint8_t chr_lsb = header[INES_CHR_SIZE_LSB];
@@ -44,21 +120,18 @@ static int ines2_load(struct nes_state *state, char *path) {
 	uint32_t prg_size = (prg_msb << 8 | prg_lsb) * PRG_ROM_UNIT;
 	uint32_t chr_size = (chr_msb << 8 | chr_lsb) * CHR_ROM_UNIT;
 
-	// NES 2.0 exponential format (only if lower byte is 0x0f)
-	if(prg_lsb == 0x0f) {
-		prg_size = 1 << header[INES_PRG_EXP];
-	}
-	if(chr_lsb == 0x0f) {
-		chr_size = 1 << header[INES_CHR_EXP];
-	}
+	if(prg_lsb == 0x0f) prg_size = 1 << header[INES_PRG_EXP];
+	if(chr_lsb == 0x0f) chr_size = 1 << header[INES_CHR_EXP];
+
+	uint8_t mapper_low   = (header[INES_FLAGS6] >> 4);
+	uint8_t mapper_high  = (header[INES_FLAGS7] & 0xf0);
+	uint8_t mapper_ext   = (header[INES_PRG_CHR_MSB] & 0x0f);
+	uint8_t mapper_upper = header[8] & 0x0f;
+	uint8_t submapper    = (header[8] >> 4);
 
-	// Extract mapper
-	uint8_t mapper_low  = (header[INES_FLAGS6] >> 4);
-	uint8_t mapper_high = (header[INES_FLAGS7] & 0xf0);
-	uint8_t mapper_ext  = (header[INES_PRG_CHR_MSB] & 0x0f);
-	state->ines.mapper = mapper_low | mapper_high | (mapper_ext << 8);
+	state->ines.mapper    = mapper_low | mapper_high | (mapper_upper << 8);
+	state->ines.submapper = submapper;
 
-	// Extract mirroring
 	if(header[INES_FLAGS6] & FLAG6_FOUR_SCREEN) {
 		state->ines.mirroring = MIRROR_FOUR_SCREEN;
 	} else if(header[INES_FLAGS6] & FLAG6_VERTICAL_MIRROR) {
@@ -70,21 +143,20 @@ static int ines2_load(struct nes_state *state, char *path) {
 	state->ines.prg_size = prg_size;
 	state->ines.chr_size = chr_size;
 
-	// Skip trainer if present
 	if(header[INES_FLAGS6] & FLAG6_TRAINER) {
-		fseek(f, TRAINER_SIZE, SEEK_CUR);
+		ptr += TRAINER_SIZE;
 	}
 
-	// Read PRG
-	printf("prgsize_read: %ld\n", fread(state->prg_rom, 1, prg_size, f));
+	memcpy(state->prg_rom, ptr, prg_size);
+	ptr += prg_size;
 
-	// Read CHR if present
 	if(chr_size > 0) {
-		printf("chrsize_read: %ld\n", fread(state->chr_rom, 1, chr_size, f));
+		memcpy(state->chr_rom, ptr, chr_size);
 	}
 
-	fclose(f);
+	free(data);
 	return 0;
 }
 
 
+
diff --git a/mapper.c b/mapper.c
index a7954b1..09d9a62 100644
--- a/mapper.c
+++ b/mapper.c
@@ -1,54 +1,54 @@
 
 
-#include "mapper_001.c"
-// #include "mapper_mmc1.c"
-// #include "mapper_uxrom.c"
-#include "mapper_066.c"
-
-struct mapper_entry {
-	int		id;
-	uint8_t	(*prg_read)(struct nes_state *state, uint32_t addr);
-	void		(*prg_write)(struct nes_state *state, uint32_t addr, uint8_t value);
-	uint8_t	(*chr_read)(struct nes_state *state, uint32_t addr);
-	void		(*chr_write)(struct nes_state *state, uint32_t addr, uint8_t value);
-	uint8_t	(*ciram_read)(struct nes_state *state, uint32_t addr);
-	void		(*ciram_write)(struct nes_state *state, uint32_t addr, uint8_t value);
-	void		(*tick)(struct nes_state *state);
-	void		(*init)(struct nes_state *state);
-};
+#include "mapper_0000.c"
+#include "mapper_0042.c"
+
+
+static uint8_t mapper_default_ciram_read(struct nes_state *state, uint32_t addr) {
+	if(state->ines.mirroring == 0) { // Horizontal
+		addr = (addr & 0x800) | (addr & 0x3ff);
+	} else { // Vertical (default fallback)
+		addr = addr & 0x7ff;
+	}
 
+	return state->ciram[addr];
+}
+
+static void mapper_default_ciram_write(struct nes_state *state, uint32_t addr, uint8_t value) {
+	if(state->ines.mirroring == 0) {
+		addr = (addr & 0x800) | (addr & 0x3ff);
+	} else {
+		addr = addr & 0x7ff;
+	}
+	state->ciram[addr] = value;
+}
+
+static void mapper_default_tick(struct nes_state *state) { // No IRQ or timing logic needed
+}
+
+/*
+ *  Mapper numbers are constructed like this:  (submapper << 12 | mapperid) both in hex.
+ *
+ *  NOTE(peter): Mapper 0 always has to be first!
+ */
 static struct mapper_entry mapper_table[] = {
-	{  0, mapper_001_prg_read, mapper_001_prg_write, mapper_001_chr_read, mapper_001_chr_write, mapper_001_ciram_read, mapper_001_ciram_write, mapper_001_tick, mapper_001_init },
-	{ 66, mapper_066_prg_read, mapper_066_prg_write, mapper_066_chr_read, mapper_066_chr_write, mapper_066_ciram_read, mapper_066_ciram_write, mapper_066_tick, mapper_066_init },
-	// { 1, mapper_mmc1_read, ... }, etc
+/* Mapper:   0 */	{ 0x00, mapper_0000_prg_read, mapper_0000_prg_write, mapper_0000_chr_read, mapper_0000_chr_write, mapper_default_ciram_read, mapper_default_ciram_write, mapper_default_tick, mapper_0000_init },
+/* Mapper:  66 */	{ 0x42, mapper_0042_prg_read, mapper_0042_prg_write, mapper_0042_chr_read, mapper_0042_chr_write, mapper_default_ciram_read, mapper_default_ciram_write, mapper_default_tick, mapper_0042_init },
 };
 
+
 static void mapper_setup(struct nes_state *state) {
 	uint32_t mapper = state->ines.mapper;
 	for(uint32_t i = 0; i < sizeof(mapper_table)/sizeof(mapper_table[0]); i++) {
 		if(mapper_table[i].id == mapper) {
-			state->mapper.prg_read		= mapper_table[i].prg_read;
-			state->mapper.prg_write		= mapper_table[i].prg_write;
-			state->mapper.chr_read		= mapper_table[i].chr_read;
-			state->mapper.chr_write		= mapper_table[i].chr_write;
-			state->mapper.ciram_read	= mapper_table[i].ciram_read;
-			state->mapper.ciram_write	= mapper_table[i].ciram_write;
-			state->mapper.tick			= mapper_table[i].tick;
-			state->mapper.init			= mapper_table[i].init;
+			state->mapper = mapper_table[i];
 			state->mapper.init(state);
 			return;
 		}
 	}
 
-	// NOTE(peter): Not sure how safe this is, but it sure is funny...
 	printf("Unsupported mapper %d, falling back to NROM\n", mapper);
-	state->mapper.prg_read		= mapper_001_prg_read;
-	state->mapper.prg_write		= mapper_001_prg_write;
-	state->mapper.chr_read		= mapper_001_chr_read;
-	state->mapper.chr_write		= mapper_001_chr_write;
-	state->mapper.ciram_read	= mapper_001_ciram_read;
-	state->mapper.ciram_write	= mapper_001_ciram_write;
-	state->mapper.tick			= mapper_001_tick;
-	state->mapper.init			= mapper_001_init;
+	state->mapper = mapper_table[0];
 	state->mapper.init(state);
 }
+
diff --git a/mapper.h b/mapper.h
index 46db0a4..e7e1437 100644
--- a/mapper.h
+++ b/mapper.h
@@ -1,20 +1,20 @@
 
-#include "mapper_066.h"
+#include "mapper_0042.h"
 
 
-
-struct mapper {
-	void (*init)(struct nes_state *state);
-	uint8_t (*prg_read)(struct nes_state *state, uint32_t addr);
-	void (*prg_write)(struct nes_state *state, uint32_t addr, uint8_t value);
-	uint8_t (*chr_read)(struct nes_state *state, uint32_t addr);
-	void (*chr_write)(struct nes_state *state, uint32_t addr, uint8_t value);
-	uint8_t (*ciram_read)(struct nes_state *state, uint32_t addr);
-	void (*ciram_write)(struct nes_state *state, uint32_t addr, uint8_t value);
-	void (*tick)(struct nes_state *state);
+struct mapper_entry {
+	int		id;
+	uint8_t	(*prg_read)(struct nes_state *state, uint32_t addr);
+	void		(*prg_write)(struct nes_state *state, uint32_t addr, uint8_t value);
+	uint8_t	(*chr_read)(struct nes_state *state, uint32_t addr);
+	void		(*chr_write)(struct nes_state *state, uint32_t addr, uint8_t value);
+	uint8_t	(*ciram_read)(struct nes_state *state, uint32_t addr);
+	void		(*ciram_write)(struct nes_state *state, uint32_t addr, uint8_t value);
+	void		(*tick)(struct nes_state *state);
+	void		(*init)(struct nes_state *state);
 };
 
 union mapper_data {
-	struct mapper_066 m066;
+	struct mapper_0042 m0042;
 };
 
diff --git a/mapper_0000.c b/mapper_0000.c
new file mode 100644
index 0000000..e075de5
--- /dev/null
+++ b/mapper_0000.c
@@ -0,0 +1,24 @@
+
+
+static void mapper_0000_init(struct nes_state *state) {
+	// Nothing to initialize for 001
+}
+
+__attribute__((hot))
+static uint8_t mapper_0000_prg_read(struct nes_state *state, uint32_t addr) {
+	uint32_t prg_size = state->ines.prg_size;
+
+	uint32_t mask = (state->ines.prg_size == 16384) ? 0x3fff : 0x7fff;
+	return state->prg_rom[addr & mask];
+}
+
+static void mapper_0000_prg_write(struct nes_state *state, uint32_t addr, uint8_t value) {
+}
+
+__attribute__((hot))
+static uint8_t mapper_0000_chr_read(struct nes_state *state, uint32_t addr) {
+	return state->chr_rom[addr];
+}
+
+static void mapper_0000_chr_write(struct nes_state *state, uint32_t addr, uint8_t value) {
+}
diff --git a/mapper_001.c b/mapper_001.c
deleted file mode 100644
index 4704444..0000000
--- a/mapper_001.c
+++ /dev/null
@@ -1,46 +0,0 @@
-
-
-static void mapper_001_init(struct nes_state *state) {
-	// Nothing to initialize for 001
-}
-
-static uint8_t mapper_001_prg_read(struct nes_state *state, uint32_t addr) {
-	uint32_t prg_size = state->ines.prg_size;
-
-	uint32_t mask = (state->ines.prg_size == 16384) ? 0x3fff : 0x7fff;
-	return state->prg_rom[addr & mask];
-}
-
-static void mapper_001_prg_write(struct nes_state *state, uint32_t addr, uint8_t value) {
-}
-
-static uint8_t mapper_001_chr_read(struct nes_state *state, uint32_t addr) {
-	return state->chr_rom[addr];
-}
-
-static void mapper_001_chr_write(struct nes_state *state, uint32_t addr, uint8_t value) {
-}
-
-static uint8_t mapper_001_ciram_read(struct nes_state *state, uint32_t addr) {
-	if(state->ines.mirroring == 0) { // Horizontal
-		addr = (addr & 0x800) | (addr & 0x3ff);
-	} else { // Vertical (default fallback)
-		addr = addr & 0x7ff;
-	}
-
-	return state->ciram[addr];
-}
-
-static void mapper_001_ciram_write(struct nes_state *state, uint32_t addr, uint8_t value) {
-	if(state->ines.mirroring == 0) {
-		addr = (addr & 0x800) | (addr & 0x3ff);
-	} else {
-		addr = addr & 0x7ff;
-	}
-
-	state->ciram[addr] = value;
-}
-
-static void mapper_001_tick(struct nes_state *state) {
-}
-
diff --git a/mapper_0042.c b/mapper_0042.c
new file mode 100644
index 0000000..4cf4c86
--- /dev/null
+++ b/mapper_0042.c
@@ -0,0 +1,30 @@
+
+static void mapper_0042_init(struct nes_state *state) {
+	state->map.m0042.prg_offset = 0;
+	state->map.m0042.chr_offset = 0;
+}
+
+static uint8_t mapper_0042_prg_read(struct nes_state *state, uint32_t addr) {
+	if(addr >= 0x8000) {
+		uint32_t base = state->map.m0042.prg_offset;
+		return state->prg_rom[base + (addr - 0x8000)];
+	}
+	return 0;
+}
+
+static void mapper_0042_prg_write(struct nes_state *state, uint32_t addr, uint8_t value) {
+	if(addr >= 0x8000) {
+		uint32_t prg_bank = (value >> 4) & 3;
+		uint32_t chr_bank = (value >> 0) & 3;
+
+		state->map.m0042.prg_offset = prg_bank * 0x8000;
+		state->map.m0042.chr_offset = chr_bank * 0x2000;
+	}
+}
+
+static uint8_t mapper_0042_chr_read(struct nes_state *state, uint32_t addr) {
+	return state->chr_rom[state->map.m0042.chr_offset + addr];
+}
+
+static void mapper_0042_chr_write(struct nes_state *state, uint32_t addr, uint8_t value) {
+}
diff --git a/mapper_066.h b/mapper_0042.h
index d78f8b9..cb201d9 100644
--- a/mapper_066.h
+++ b/mapper_0042.h
@@ -1,5 +1,5 @@
 
-struct mapper_066 {
+struct mapper_0042 {
 	uint32_t prg_offset;
 	uint32_t chr_offset;
 };
diff --git a/mapper_066.c b/mapper_066.c
deleted file mode 100644
index 7a2b542..0000000
--- a/mapper_066.c
+++ /dev/null
@@ -1,58 +0,0 @@
-
-static void mapper_066_init(struct nes_state *state) {
-	state->map.m066.prg_offset = 0;
-	state->map.m066.chr_offset = 0;
-}
-
-static uint8_t mapper_066_prg_read(struct nes_state *state, uint32_t addr) {
-	if(addr >= 0x8000) {
-		uint32_t base = state->map.m066.prg_offset;
-		return state->prg_rom[base + (addr - 0x8000)];
-	}
-	return 0;
-}
-
-static void mapper_066_prg_write(struct nes_state *state, uint32_t addr, uint8_t value) {
-	if(addr >= 0x8000) {
-		uint32_t prg_bank = (value >> 4) & 3;
-		uint32_t chr_bank = (value >> 0) & 3;
-
-		state->map.m066.prg_offset = prg_bank * 0x8000;
-		state->map.m066.chr_offset = chr_bank * 0x2000;
-	}
-}
-
-static uint8_t mapper_066_chr_read(struct nes_state *state, uint32_t addr) {
-	return state->chr_rom[state->map.m066.chr_offset + addr];
-}
-
-static void mapper_066_chr_write(struct nes_state *state, uint32_t addr, uint8_t value) {
-}
-
-static uint8_t mapper_066_ciram_read(struct nes_state *state, uint32_t addr) {
-	uint32_t mirrored = addr & 0x0fff;
-
-	if(state->ines.mirroring == 0) { // Horizontal
-		mirrored = (mirrored & 0x800) | (mirrored & 0x3ff);
-	} else { // Vertical (default fallback)
-		mirrored = mirrored & 0x7ff;
-	}
-
-	return state->ciram[mirrored];
-}
-
-static void mapper_066_ciram_write(struct nes_state *state, uint32_t addr, uint8_t value) {
-	uint32_t mirrored = addr & 0x0fff;
-
-	if(state->ines.mirroring == 0) {
-		mirrored = (mirrored & 0x800) | (mirrored & 0x3ff);
-	} else {
-		mirrored = mirrored & 0x7ff;
-	}
-
-	state->ciram[mirrored] = value;
-}
-
-static void mapper_066_tick(struct nes_state *state) {
-	// No IRQ or timing logic needed
-}
diff --git a/memory.c b/memory.c
index e0ed85a..28c51ad 100644
--- a/memory.c
+++ b/memory.c
@@ -1,97 +1,61 @@
 
 
 
+__attribute__((hot))
 static uint8_t memory_read(struct nes_state *restrict state, uint32_t offset) {
 	state->cycles++;
-	ppu_tick(state); ppu_tick(state); ppu_tick(state);
+	ppu_tick(state);
 
-	if(offset > 0xffff) {
-		printf("%x\n", offset);
-	}
-
-	// state->ram[0x301] = 0x1;
-	if(offset < 0x2000) {
+	if(LIKELY(offset < 0x2000)) {
 		return state->ram[offset & 0x07ff];
 	} else if(offset < 0x4000) {
 		return ppu_read(state, offset);
-		// switch(offset & 7) {
-		// 	case 2: return ppu_read_2002(state);
-		// 	case 4: return ppu_read_2004(state);
-		// 	case 7: return ppu_read_2007(state);
-		// 	default: return 0;
-		// }
-	} else if(offset < 0x4020) {
-		// TODO: APU and I/O reads
-		return 0;
-	} else if(offset >= 0x6000) {
+	// } else if(offset < 0x4020) {
+	// 	// TODO: APU and I/O reads
+	// 	return 0;
+	} else if(LIKELY(offset >= 0x6000)) {
 		return state->mapper.prg_read(state, offset);
-	} else {
-		return 0;
 	}
+	return 0;
 }
 
+__attribute__((hot))
 static void memory_write(struct nes_state *restrict state, uint32_t offset, uint8_t value) {
 	state->cycles++;
-	ppu_tick(state); ppu_tick(state); ppu_tick(state);
-
-	if(offset > 0xffff) {
-		printf("%x\n", offset);
-	}
+	ppu_tick(state);
 
-	if(offset < 0x2000) {
+	if(LIKELY(offset < 0x2000)) {
 		state->ram[offset & 0x07ff] = value;
 	} else if(offset < 0x4000) {
 		ppu_write(state, offset, value);
-		// switch(offset & 7) {
-		// 	case 0: ppu_write_2000(state, value); break;
-		// 	case 1: ppu_write_2001(state, value); break;
-		// 	case 3: ppu_write_2003(state, value); break;
-		// 	case 4: ppu_write_2004(state, value); break;
-		// 	case 5: ppu_write_2005(state, value); break;
-		// 	case 6: ppu_write_2006(state, value); break;
-		// 	case 7: ppu_write_2007(state, value); break;
-		// 	default: break;
-		// }
 	} else if(offset == 0x4014) {
 		ppu_dma_4014(state, value);
-	} else if(offset < 0x4020) {
-		// TODO: APU and I/O writes
+	// } else if(offset < 0x4020) {
+	// 	// TODO: APU and I/O writes
 	} else if(offset >= 0x6000) {
 		state->mapper.prg_write(state, offset, value);
 	}
 }
 
+__attribute__((hot))
 static uint8_t memory_read_dma(struct nes_state *restrict state, uint32_t offset) {
 	// Do not tick CPU/PPU/APU — caller handles timing
-	if(offset < 0x2000) {
+	if(LIKELY(offset < 0x2000)) {
 		return state->ram[offset & 0x07ff];
-	} else if(offset < 0x4000) {
-		// PPU register reads are ignored during DMA
-		return 0;
-	} else if(offset < 0x4020) {
-		// APU and I/O — usually ignored or blocked during DMA
-		return 0;
 	} else if(offset >= 0x6000) {
 		return state->mapper.prg_read(state, offset);
-	} else {
-		return 0;
 	}
+	return 0;
 }
 
+__attribute__((hot))
 static uint8_t memory_read_dummy(struct nes_state *restrict state, uint32_t offset) {
 	state->cycles++;
-	ppu_tick(state); ppu_tick(state); ppu_tick(state);
+	ppu_tick(state);
 
-	if(offset < 0x2000) {
-		return 0;
-	} else if(offset < 0x4000) {
+	if(UNLIKELY(offset >= 0x2000 && offset < 0x4000)) {
 		return ppu_read(state, offset);
-	} else if(offset < 0x4020) {
-		return 0;
-	} else if(offset >= 0x6000) {
-		return state->mapper.prg_read(state, offset);
-	} else {
-		return 0;
 	}
+	return 0;
 }
 
diff --git a/mknes.c b/mknes.c
index 0ca6749..9a91376 100644
--- a/mknes.c
+++ b/mknes.c
@@ -13,6 +13,9 @@
 #define WINDOW_WIDTH 320 * 3
 #define WINDOW_HEIGHT 240 * 3
 
+#define LIKELY(x)   __builtin_expect(!!(x), 1)
+#define UNLIKELY(x) __builtin_expect(!!(x), 0)
+
 #ifndef restrict
 #  if defined(__cplusplus)
 #    define restrict __restrict
@@ -113,7 +116,7 @@ int main(int argc, char **argv) {
 	// ines2_load(&nstate, "data/nrom/Excitebike (Japan, USA).nes");
 	// ines2_load(&nstate, "data/nrom/Ice Climber (USA, Europe, Korea).nes");
 	// ines2_load(&nstate, "data/nrom/Kung Fu (Japan, USA).nes");
-	ines2_load(&nstate, "data/nrom/Super Mario Bros. (World) (HVC-SM).nes");
+	// ines2_load(&nstate, "data/nrom/Super Mario Bros. (World) (HVC-SM).nes");
 	// ines2_load(&nstate, "data/nrom/Urban Champion (World).nes");
 	// ines2_load(&nstate, "data/nrom/Wrecking Crew (World).nes");
 	// ines2_load(&nstate, "data/nrom/scanline.nes");
@@ -128,6 +131,9 @@ int main(int argc, char **argv) {
 	// ines2_load(&nstate, "data/nrom/raster_demos/RasterTest3d.NES");
 	// ines2_load(&nstate, "data/nrom/raster_demos/RasterTest3e.NES");
 	// ines2_load(&nstate, "data/nrom/NEStress.NES");
+	// ines2_load(&nstate, "data/tv.nes");
+	ines2_load(&nstate, "data/Super Mario Bros. (World) (HVC-SM).zip");
+	// ines2_load(&nstate, "data/Super Mario Bros. + Duck Hunt (USA).zip");
 
 	mapper_setup(&nstate);
 	uint32_t lo = nstate.mapper.prg_read(&nstate, 0xfffc);
@@ -172,7 +178,7 @@ int main(int argc, char **argv) {
 			timer_start(timer);
 
 			while(!glfwWindowShouldClose(window)) {
-				// timer_wait(timer);
+				timer_wait(timer);
 				glfwPollEvents();
 
 // //
@@ -183,21 +189,21 @@ int main(int argc, char **argv) {
 				nstate.ppu.frame_ready = 0;
 				frames++;
 
-				// uint32_t * restrict dst = buffer;
-				// uint8_t * restrict src = nstate.ppu.pixels;
-				// for(uint32_t y = 0; y < 240; ++y) {
-				// 	for(uint32_t x = 0; x < 256; ++x) {
-				// 		uint8_t val = *src++;
-				// 		if(val >= 64) val = 0;
-				// 		dst[x] = nes_palette[val];
-				// 	}
-				// 	dst += BUFFER_WIDTH;
-				// }
-				// apply_phosphor_decay();
-				// render_frame();
+				uint32_t * restrict dst = buffer;
+				uint8_t * restrict src = nstate.ppu.pixels;
+				for(uint32_t y = 0; y < 240; ++y) {
+					for(uint32_t x = 0; x < 256; ++x) {
+						uint8_t val = *src++;
+						if(val >= 64) val = 0;
+						dst[x] = nes_palette[val];
+					}
+					dst += BUFFER_WIDTH;
+				}
+				apply_phosphor_decay();
+				render_frame();
 				glfwSwapBuffers(window);
 			}
-printf("%d\n", frames);
+printf("total frames: %6.6d  total cycles: %ld\n", frames, nstate.cycles);
 			glfwDestroyWindow(window);
 		} else {
 			fprintf(stderr, "Failed to create window\n");
diff --git a/mknes.h b/mknes.h
index ed22652..b8bf87f 100644
--- a/mknes.h
+++ b/mknes.h
@@ -21,19 +21,33 @@
 #define MIRROR_FOURSCREEN 2
 
 struct ppu_state {
-	uint8_t pixels[256 * 240];
-
-	uint8_t oam[256];
+	uint32_t scanline;
+	uint32_t dot;
+	uint32_t bg_shift_pattern_low;
+	uint32_t bg_shift_pattern_high;
+	uint32_t bg_shift_attrib_low;
+	uint32_t bg_shift_attrib_high;
+	uint8_t bg_next_tile_id;
+	uint8_t bg_next_tile_attrib;
+	uint8_t bg_next_tile_lsb;
+	uint8_t bg_next_tile_msb;
 	uint8_t oam_addr;
 	uint8_t oam_data;
+
+	uint8_t pixels[256 * 240] __attribute__((aligned(64)));
+	uint8_t oam[256];
 	uint8_t secondary_oam[32];
+	uint8_t palette[0x20];
+	uint8_t sprite_indexes[8];
+	uint32_t sprite_patterns[8];
+	uint8_t sprite_positions[8];
+	uint8_t sprite_priorities[8];
+	uint8_t sprite_shift_lo[8];
+	uint8_t sprite_shift_hi[8];
 
 	uint8_t reg_ctrl;
 	uint8_t reg_mask;
 	uint8_t reg_status;
-	uint8_t reg_scroll[2];
-	uint8_t reg_addr[2];
-	uint8_t reg_latch;
 
 	uint32_t vram_addr;
 	uint32_t temp_addr;
@@ -44,33 +58,11 @@ struct ppu_state {
 
 	uint8_t open_bus;
 
-	uint8_t palette[0x20];
-
-	uint32_t scanline;
-	uint32_t dot;
-	uint8_t even_frame;
 	uint8_t frame_ready;
 
-	uint8_t sprite_indexes[8];
 	uint8_t sprite_zero_hit_possible;
 	uint8_t sprite_count;
-	uint32_t sprite_patterns[8];
-	uint8_t sprite_positions[8];
-	uint8_t sprite_priorities[8];
-	uint8_t sprite_shift_lo[8];
-	uint8_t sprite_shift_hi[8];
-
-	uint32_t bg_shift_pattern_low;
-	uint32_t bg_shift_pattern_high;
-	uint32_t bg_shift_attrib_low;
-	uint32_t bg_shift_attrib_high;
-
-	uint8_t bg_next_tile_id;
-	uint8_t bg_next_tile_attrib;
-	uint8_t bg_next_tile_lsb;
-	uint8_t bg_next_tile_msb;
-};
-
+} __attribute__((packed, aligned(64)));
 
 struct cpu_state {
 	uint32_t pc;	// Program Counter
@@ -81,38 +73,39 @@ struct cpu_state {
 	uint8_t p;		// Processor Status Flags (this can be expanded with separate flags if needed)
 	uint8_t n;		// Negative Flag
 	uint8_t v;		// Overflow Flag
-	// uint8_t b;		// Break Flag (Set by BRK instruction)
+	// uint8_t b;		// Break Flag (Set by BRK instruction) -- does not exist outside the stack
 	uint8_t d;		// Decimal Flag
 	uint8_t i;		// Interrupt Disable Flag
 	uint8_t z;		// Zero Flag
 	uint8_t c;		// Carry Flag
 // --
 	uint8_t die;	// KIL instruction found!
-};
+} __attribute__((packed, aligned(64)));
 
 
 struct ines_state {
 	uint32_t mapper;
+	uint32_t submapper;
 	uint8_t mirroring;	// 0 = H, 1 = V, 2 = 4-screen
 	uint32_t prg_size;
 	uint32_t chr_size;
 };
 
 struct nes_state {
+	size_t cycles;
 	struct ines_state ines;
 	struct cpu_state cpu;
-	struct ppu_state ppu;
-	struct mapper mapper;
-	union mapper_data map;
-	size_t cycles;
 	uint8_t irq_pending;
 	uint8_t nmi_pending;
-	uint8_t ram[0x800];
-	uint8_t sram[0x2000];
-	uint8_t ciram[0x1000];				// NOTE(peter): Originally 0x800 bytes, but extended as it should work for up to fourway, this is optimization, reality is 2kb, but there is no side-effects, so this is fine!
-	uint8_t prg_rom[4 * 1024 * 1024];
-	uint8_t chr_rom[4 * 1024 * 1024];
-};
+	struct ppu_state ppu;
+	struct mapper_entry mapper;
+	union mapper_data map;
+	uint8_t ram[0x800] __attribute__((aligned(64)));
+	uint8_t sram[0x2000] __attribute__((aligned(64)));
+	uint8_t ciram[0x1000] __attribute__((aligned(64)));				// NOTE(peter): Originally 0x800 bytes, but extended as it should work for up to fourway, this is optimization, reality is 2kb, but there is no side-effects, so this is fine!
+	uint8_t prg_rom[4 * 1024 * 1024] __attribute__((aligned(64)));
+	uint8_t chr_rom[4 * 1024 * 1024] __attribute__((aligned(64)));
+} __attribute__((aligned(64)));
 
 
 __attribute__((aligned(4096))) static uint32_t nes_palette[64] = {
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);
 	}