diff --git a/include/cart.h b/include/cart.h
index f991e20..88fdfea 100644
--- a/include/cart.h
+++ b/include/cart.h
@@ -24,4 +24,8 @@ bool cart_load(char *cart);
 rom_header *get_rom_header();
 
 u8 cart_read(u16 address);
-void cart_write(u16 address, u8 value);
\ No newline at end of file
+void cart_write(u16 address, u8 value);
+
+bool cart_need_save();
+bool cart_battery_load();
+bool cart_battery_save();
\ No newline at end of file
diff --git a/include/gamepad.h b/include/gamepad.h
new file mode 100644
index 0000000..8460056
--- /dev/null
+++ b/include/gamepad.h
@@ -0,0 +1,22 @@
+#pragma once
+
+#include <common.h>
+
+typedef struct {
+    bool start;
+    bool select;
+    bool a;
+    bool b;
+    bool up;
+    bool down;
+    bool left;
+    bool right;
+} gamepad_state;
+
+void gamepad_init();
+bool gamepad_button_sel();
+bool gamepad_dir_sel();
+void gamepad_set_sel(u8 value);
+
+gamepad_state *gamepad_get_state();
+u8 gamepad_get_output();
\ No newline at end of file
diff --git a/include/lcd.h b/include/lcd.h
index b6c260c..921fa55 100644
--- a/include/lcd.h
+++ b/include/lcd.h
@@ -41,7 +41,7 @@ lcd_context *lcd_get_context();
 #define LCDC_LCD_ENABLE (BIT(lcd_get_context()->lcdc, 7))
 
 #define LCDS_MODE ((lcd_mode)(lcd_get_context()->lcds & 0b11))
-#define LCDS_MODE_SET(mode) { lcd_get_context()->lcds &= ~0b11; lcd_get_context()->lcdc |= mode; }
+#define LCDS_MODE_SET(mode) { lcd_get_context()->lcds &= ~0b11; lcd_get_context()->lcds |= mode; }
 #define LCDS_LYC (BIT(lcd_get_context()->lcds, 2))
 #define LCDS_LYC_SET(b) BIT_SET(lcd_get_context()->lcds, 2, b)
 
diff --git a/include/ppu.h b/include/ppu.h
index 101949f..aa005de 100644
--- a/include/ppu.h
+++ b/include/ppu.h
@@ -7,23 +7,73 @@ static const int TICKS_PER_LINE = 456;
 static const int YRES = 144;
 static const int XRES = 160;
 
+u32 target_frame_time;
+
+typedef enum {
+    FS_TILE,
+    FS_DATA0,
+    FS_DATA1,
+    FS_IDLE,
+    FS_PUSH
+} fetch_state;
+
+typedef struct _fifo_entry {
+    struct _fifo_entry *next;
+    u32 value;
+} fifo_entry;
+
+typedef struct {
+    fifo_entry *head;
+    fifo_entry *tail;
+    u32 size;  //32 bit color
+} fifo;
+
+typedef struct {
+    fetch_state cur_fetch_state;
+    fifo pixel_fifo;
+    u8 line_x;
+    u8 pushed_x;
+    u8 fetch_x;
+    u8 bgw_fetch_data[3];
+    u8 fetch_entry_data[6]; //oam data
+    u8 map_y;
+    u8 map_x;
+    u8 tile_y;
+    u8 fifo_x;
+} pixel_fifo_context;
+
 typedef struct {
     u8 y;
     u8 x;
     u8 tile;
 
-    unsigned f_cgb_pn : 3;
-    unsigned f_cgb_vram_bank : 1;
-    unsigned f_pn : 1;
-    unsigned f_x_flip : 1;
-    unsigned f_y_flip : 1;
-    unsigned f_bgp : 1;
+    u8 f_cgb_pn : 3;
+    u8 f_cgb_vram_bank : 1;
+    u8 f_pn : 1;
+    u8 f_x_flip : 1;
+    u8 f_y_flip : 1;
+    u8 f_bgp : 1;
 } oam_entry;
 
+typedef struct _oam_line_entry {
+    oam_entry entry;
+    struct _oam_line_entry *next;
+} oam_line_entry;
+
 typedef struct {
     oam_entry oam_ram[40];
     u8 vram[0x2000];
 
+    u8 line_sprite_count; // 0 to 10 sprites
+    oam_line_entry *line_sprites; //linked list of sprites
+    oam_line_entry line_entry_array[10]; //memory to use for list
+    
+    u8 fetched_entry_count;
+    oam_entry fetched_entries[3]; //entries fetched during pipeline
+    u8 window_line;
+
+    pixel_fifo_context pfc;
+
     u32 current_frame;
     u32 line_ticks;
     u32 *video_buffer;
diff --git a/lib/cart.c b/lib/cart.c
index f6c0247..36eadb9 100644
--- a/lib/cart.c
+++ b/lib/cart.c
@@ -6,10 +6,39 @@ typedef struct {
     u32 rom_size;
     u8 *rom_data;
     rom_header *header;
+
+    //mbc1 data
+    bool ram_enabled;
+    bool ram_banking;
+
+    u8 *rom_bank_x;
+    u8 banking_mode;
+
+    u8 rom_bank_value;
+    u8 ram_bank_value;
+
+    u8 *ram_bank;
+    u8 *ram_banks[16];
+
+    //battery
+    bool battery;
+    bool need_save;
 } cart_context;
 
 static cart_context ctx;
 
+bool cart_need_save () {
+    return ctx.need_save;
+}
+
+bool cart_mbc1() {
+    return BETWEEN(ctx.header->type, 1, 3);
+}
+
+bool cart_battery() {
+    return ctx.header->type == 3;
+}
+
 static const char *ROM_TYPES[] = {
     "ROM ONLY",
     "MBC1",
@@ -132,6 +161,23 @@ const char *cart_lic_name() {
     return "UNKNOWN";
 }
 
+void cart_setup_bamking() {
+    for (int i=0; i<16; i++) {
+        ctx.ram_banks[1] = 0;
+
+        if((ctx.header->ram_size == 2 && i == 0) || 
+            (ctx.header->ram_size == 3 && i < 4) ||
+            (ctx.header->ram_size == 5 && i < 8) ||
+            (ctx.header->ram_size == 4)) {
+            ctx.ram_banks[i] = malloc(0x2000);
+            memset(ctx.ram_banks[i], 0, 0x2000);
+        }
+    }
+
+    ctx.ram_bank = ctx.ram_banks[0];
+    ctx.rom_bank_x = ctx.rom_data + 0x4000;
+}
+
 bool cart_load(char *cart) {
     snprintf(ctx.filename, sizeof(ctx.filename), "%s", cart);
     FILE *fp = fopen(cart, "r");
@@ -153,6 +199,8 @@ bool cart_load(char *cart) {
 
     ctx.header = (rom_header *)(ctx.rom_data + 0x100);
     ctx.header->title[15] = 0;
+    ctx.battery = cart_battery();
+    ctx.need_save = false;
 
     printf("Cartridge Loaded:\n");
     printf("\t Title    : %s\n", ctx.header->title);
@@ -162,6 +210,8 @@ bool cart_load(char *cart) {
     printf("\t LIC Code : %2.2X (%s)\n", ctx.header->lic_code, cart_lic_name());
     printf("\t ROM Vers : %2.2X\n", ctx.header->version);
 
+    cart_setup_bamking();
+
     u16 x = 0;
     for (u16 i=0X0134; i<=0X014C; i++) {
         x = x - ctx.rom_data[i] -1;
@@ -172,17 +222,119 @@ bool cart_load(char *cart) {
     if((x & 0xFF) != ctx.header->checksum) {
         fprintf(stderr, "WARNING!!! The header checksum does not match! ROM may be corrupt or invalid!\n");
     }
+
+    if(ctx.battery) {
+        cart_battery_load();
+    }
     
     return true;
 }
 
 u8 cart_read(u16 address){
-    //for now ROM ONLY type supported...
+    if(!cart_mbc1() || address < 0x4000) {
+        return ctx.rom_data[address];
+    }
+
+    if((address & 0xE000) == 0xA000) {
+        if (!ctx.ram_enabled) {
+            return 0xFF;
+        }
+
+        if(!ctx.ram_bank) {
+            return 0xFF;
+        }
+
+        return ctx.ram_bank[address - 0xA000];
+    }
+
+    return ctx.rom_bank_x[address - 0x4000];
 
-    return ctx.rom_data[address];
 }
 
 void cart_write(u16 address, u8 value){
-    printf("UNSUPPORTED cart_write(%04X)\n", address);
-    //NO_IMPL
+    if(!cart_mbc1()) {
+        return;
+    }
+
+    if(address < 0x2000) {
+        ctx.ram_enabled = ((value & 0xF) == 0xA);
+    }
+
+    if((address & 0xE000) == 0x2000) {
+        //rom bank
+        if(value == 0) {
+            value = 1;
+        }
+
+        value &= 0b11111;
+        ctx.rom_bank_value = value;
+        ctx.rom_bank_x = ctx.rom_data + (0x4000 * ctx.rom_bank_value);
+    }
+
+    if((address & 0xE000) == 0x4000) {
+        //ram bank number
+        ctx.ram_bank_value = value & 0xb11;
+        if(ctx.ram_banking) {
+            if(cart_need_save) {
+                cart_battery_save();
+            }
+            ctx.ram_bank = ctx.ram_banks[ctx.ram_bank_value];
+        }
+    }
+
+    if((address & 0xE000) == 0x6000) {
+        //bamking mode select
+
+        ctx.banking_mode = value & 1;
+
+        ctx.ram_banking = ctx.banking_mode;
+
+        if (ctx.ram_banking) {
+            if(cart_need_save) {
+                cart_battery_save();
+            }
+            ctx.ram_bank = ctx.ram_banks[ctx.ram_bank_value];
+        }
+    }
+
+    if((address & 0xE000) == 0xA000) {
+        if(!ctx.ram_enabled)
+            return;
+
+        if(!ctx.ram_bank)
+            return;
+
+        ctx.ram_bank[address - 0xA000] = value;
+
+        if(ctx.battery) {
+            ctx.need_save = true;
+        }
+    }
+}
+
+bool cart_battery_load(){
+    char fn[1048];
+    sprintf(fn, "%s.battery", ctx.filename);
+    FILE *fp = fopen(fn, "rb");
+
+    if(!fp) {
+        fprintf(stderr, "unable to open: %s\n", fn);
+    }
+
+    fread(ctx.ram_bank, 0x2000, 1, fp);
+    fclose(fp);
+}
+
+bool cart_battery_save(){
+    char fn[1048];
+    sprintf(fn, "%s.battery", ctx.filename);
+    FILE *fp = fopen(fn, "wb");
+
+    if(!fp) {
+        fprintf(stderr, "unable to open: %s\n", fn);
+    }
+
+    fwrite(ctx.ram_bank, 0x2000, 1, fp);
+    fclose(fp);
+
 }
\ No newline at end of file
diff --git a/lib/cpu_fetch.c b/lib/cpu_fetch.c
index 0d5e3a6..c52f146 100644
--- a/lib/cpu_fetch.c
+++ b/lib/cpu_fetch.c
@@ -75,7 +75,7 @@ void fetch_data() {
             return;
         case AM_R_A8:
             ctx.fetched_data = bus_read(ctx.regs.pc);
-            emu_cycles(2);
+            emu_cycles(1);
             ctx.regs.pc++;
             return;
         case AM_A8_R:
diff --git a/lib/cpu_proc.c b/lib/cpu_proc.c
index 0681363..588f323 100644
--- a/lib/cpu_proc.c
+++ b/lib/cpu_proc.c
@@ -70,7 +70,6 @@ static void proc_halt(cpu_context *ctx) {
 
 static void proc_stop(cpu_context *ctx) {
     printf("CPU STOP!\n");
-    NO_IMPL
 }
 
 static void proc_rlca(cpu_context *ctx) {
@@ -384,7 +383,6 @@ static void proc_pop(cpu_context *ctx) {
 
     u16 n = (hi << 8) | lo;
     cpu_set_reg(ctx->cur_inst->reg_1, n);
-    emu_cycles(2);
 
     if (ctx->cur_inst->reg_1 == RT_AF) {
         cpu_set_reg(ctx->cur_inst->reg_1, n & 0xFFF0);
@@ -397,6 +395,8 @@ static void proc_push(cpu_context *ctx) {
     emu_cycles(1);
     stack_push((val & 0xFF));
     emu_cycles(1);
+
+    emu_cycles(1);
 }
 
 static void proc_ldh(cpu_context *ctx) {
@@ -423,8 +423,8 @@ static void proc_ld(cpu_context *ctx) {
             emu_cycles(1);
         } else {
             bus_write(ctx->mem_dest, ctx->fetched_data);
-            emu_cycles(1);
         }
+        emu_cycles(1);
         return;
     }
 
diff --git a/lib/gamepad.c b/lib/gamepad.c
new file mode 100644
index 0000000..8491168
--- /dev/null
+++ b/lib/gamepad.c
@@ -0,0 +1,65 @@
+#include <gamepad.h>
+#include <string.h>
+
+typedef struct {
+    bool button_sel;
+    bool dir_sel;
+    gamepad_state controller;
+} gamepad_context;
+
+static gamepad_context ctx = {0};
+
+void gamepad_init();
+
+bool gamepad_button_sel(){
+    return ctx.button_sel;
+}
+
+bool gamepad_dir_sel(){
+    return ctx.dir_sel;
+}
+
+void gamepad_set_sel(u8 value){
+    ctx.button_sel = value & 0x20;
+    ctx.dir_sel = value & 0x10;
+}
+
+gamepad_state *gamepad_get_state(){
+    return &ctx.controller;
+}
+
+u8 gamepad_get_output() {
+    u8 output = 0xCF;
+
+    if(!gamepad_button_sel()) {
+        if (gamepad_get_state()->start) {
+            output &= ~(1 << 3);
+        }
+        if (gamepad_get_state()->select) {
+            output &= ~(1 << 2);
+        }
+        if (gamepad_get_state()->a) {
+            output &= ~(1 << 0);
+        }
+        if (gamepad_get_state()->b) {
+            output &= ~(1 << 1);
+        }
+    }
+
+    if(!gamepad_dir_sel()) {
+        if (gamepad_get_state()->left) {
+            output &= ~(1 << 1);
+        }
+        if (gamepad_get_state()->right) {
+            output &= ~(1 << 0);
+        }
+        if (gamepad_get_state()->up) {
+            output &= ~(1 << 2);
+        }
+        if (gamepad_get_state()->down) {
+            output &= ~(1 << 3);
+        }
+    }
+
+    return output;
+}
\ No newline at end of file
diff --git a/lib/io.c b/lib/io.c
index 45fc128..1c5a3e2 100644
--- a/lib/io.c
+++ b/lib/io.c
@@ -3,10 +3,14 @@
 #include <cpu.h>
 #include <dma.h>
 #include <lcd.h>
+#include <gamepad.h>
 
 static char serial_data[2];
 
 u8 io_read(u16 address){
+    if(address == 0xFF00) {
+        return gamepad_get_output();
+    }
     if(address == 0xFF01) {
         return serial_data[0];
     }
@@ -27,14 +31,24 @@ u8 io_read(u16 address){
         return cpu_get_int_flags();
     }
 
+    if(BETWEEN(address, 0xFF10, 0xFF3F)) {
+        //ignore sound
+        return 0;
+    }
+
     printf("UNSUPPORTED io_read(%04X)\n", address);
     return 0;
 }
 
 
 void io_write(u16 address, u8 value){
+    if(address == 0xFF00) {
+        gamepad_set_sel(value);
+        return;
+    }
     if(address == 0xFF01) {
         serial_data[0] = value;
+        printf("%c", value);
         return;
     }
 
@@ -58,6 +72,11 @@ void io_write(u16 address, u8 value){
         return;
     }
 
+    if(BETWEEN(address, 0xFF10, 0xFF3F)) {
+        //ignore sound
+        return;
+    }
+
     printf("UNSUPPORTED io_write(%04X)\n", address);
 }
 
diff --git a/lib/ppu.c b/lib/ppu.c
index 83f7a8c..152e69f 100644
--- a/lib/ppu.c
+++ b/lib/ppu.c
@@ -3,6 +3,9 @@
 #include <string.h>
 #include <ppu_sm.h>
 
+void pipeline_fifo_reset();
+void pipeline_process();
+
 static ppu_context ctx;
 
 ppu_context *ppu_get_context() {
@@ -14,6 +17,17 @@ void ppu_init() {
     ctx.line_ticks = 0;
     ctx.video_buffer = malloc(YRES * XRES * sizeof(u32));
 
+    ctx.pfc.line_x = 0;
+    ctx.pfc.pushed_x = 0;
+    ctx.pfc.fetch_x = 0;
+    ctx.pfc.pixel_fifo.size = 0;
+    ctx.pfc.pixel_fifo.head = ctx.pfc.pixel_fifo.tail = NULL;
+    ctx.pfc.cur_fetch_state = FS_TILE;
+
+    ctx.line_sprites = 0;
+    ctx.fetched_entry_count = 0;
+    ctx.window_line = 0;
+
     lcd_init();
     LCDS_MODE_SET(MODE_OAM);
 
diff --git a/lib/ppu_pipeline.c b/lib/ppu_pipeline.c
new file mode 100644
index 0000000..4ba19d0
--- /dev/null
+++ b/lib/ppu_pipeline.c
@@ -0,0 +1,271 @@
+#include <ppu.h>
+#include <lcd.h>
+#include <bus.h>
+
+bool window_visible() {
+    return LCDC_WIN_ENABLE && lcd_get_context()->win_x >= 0 &&
+        lcd_get_context()->win_x <= 166 && lcd_get_context()->win_y >= 0 &&
+        lcd_get_context()->win_y < YRES;
+}
+
+void pixel_fifo_push(u32 value) {
+    fifo_entry *next = malloc(sizeof(fifo_entry));
+    next->next = NULL;
+    next->value = value;
+
+    if(!ppu_get_context()->pfc.pixel_fifo.head) {
+        //first entry
+        ppu_get_context()->pfc.pixel_fifo.head = ppu_get_context()->pfc.pixel_fifo.tail = next;
+    } else {
+        ppu_get_context()->pfc.pixel_fifo.tail->next = next;
+        ppu_get_context()->pfc.pixel_fifo.tail = next;
+    }
+    
+    ppu_get_context()->pfc.pixel_fifo.size++;
+}
+
+u32 pixel_fifo_pop() {
+    if(ppu_get_context()->pfc.pixel_fifo.size <= 0) {
+        fprintf(stderr, "ERR in pixel FIFO\n");
+    }
+
+    fifo_entry *popped = ppu_get_context()->pfc.pixel_fifo.head;
+    ppu_get_context()->pfc.pixel_fifo.head = popped->next;
+    ppu_get_context()->pfc.pixel_fifo.size--;
+
+    u32 val = popped->value;
+    free(popped);
+    return val;
+}
+
+u32 fetch_sprite_pixels(int bit, u32 color, u8 bg_color) {
+    for (int i = 0; i < ppu_get_context()->fetched_entry_count; i++) {
+        int sp_x = (ppu_get_context()->fetched_entries[i].x - 8) +
+            ((lcd_get_context()->scroll_x % 8));
+
+        if (sp_x + 8 < ppu_get_context()->pfc.fifo_x) {
+            //past this pixel already
+            continue;
+        }
+
+        int offset = ppu_get_context()->pfc.fifo_x - sp_x;
+
+        if (offset < 0 || offset > 7) {
+            continue;
+        }
+
+        bit = 7 - offset;
+
+        if (ppu_get_context()->fetched_entries[i].f_x_flip) {
+            bit = offset;
+        }
+
+        u8 lo = !!(ppu_get_context()->pfc.fetch_entry_data[i * 2] & (1 << bit));
+        u8 hi = !!(ppu_get_context()->pfc.fetch_entry_data[(i * 2)+1] & (1 << bit)) << 1;
+
+        bool bg_priority = ppu_get_context()->fetched_entries[i].f_bgp;
+
+        if (!(hi|lo)) {
+            //transparant
+            continue;
+        }
+
+        if(!bg_priority || bg_color == 0) {
+            color = (ppu_get_context()->fetched_entries[i].f_pn) ?
+                lcd_get_context()->sp2_colors[hi|lo] : lcd_get_context()->sp1_colors[hi|lo];
+
+            if(hi | lo) {
+                break;
+            }
+        }
+    }
+
+    return color;
+}
+
+bool pipeline_fifo_add() {
+    if(ppu_get_context()->pfc.pixel_fifo.size > 8) {
+        //fifo is full!
+        return false;
+    }
+
+    int x = ppu_get_context()->pfc.fetch_x - (8 - lcd_get_context()->scroll_x % 8);
+
+    for (int i=0; i<8; i++) {
+        int bit = 7 - i;
+        u8 lo = !!(ppu_get_context()->pfc.bgw_fetch_data[1] & (1 << bit));
+        u8 hi = !!(ppu_get_context()->pfc.bgw_fetch_data[2] & (1 << bit)) << 1;
+        u32 color = lcd_get_context()->bg_colors[hi | lo];
+
+        if (!LCDC_BGW_ENABLE) {
+            color = lcd_get_context()->bg_colors[0];
+        }
+
+        if(LCDC_OBJ_ENABLE) {
+            color = fetch_sprite_pixels(bit, color, hi | lo);
+        }
+
+        if (x >= 0) {
+            pixel_fifo_push(color);
+            ppu_get_context()->pfc.fifo_x++;
+        }
+    }
+
+    return true;
+}
+
+void pipeline_load_sprite_tile() {
+    oam_line_entry *le = ppu_get_context()->line_sprites;
+
+    while(le) {
+        int sp_x = (le->entry.x - 8) + (lcd_get_context()->scroll_x % 8);
+
+        if ((sp_x >= ppu_get_context()->pfc.fetch_x && sp_x < ppu_get_context()->pfc.fetch_x + 8) ||
+            ((sp_x + 8) >= ppu_get_context()->pfc.fetch_x && (sp_x + 8) < ppu_get_context()->pfc.fetch_x + 8)) {
+                ppu_get_context()->fetched_entries[ppu_get_context()->fetched_entry_count++] = le->entry;
+        }
+
+        le = le->next;
+
+        if (!le || ppu_get_context()->fetched_entry_count >= 3) {
+            //max checking 3 sprites on pixels
+            break;
+        }
+    }
+}
+
+void pipeline_load_sprite_data(u8 offset) {
+    int cur_y = lcd_get_context()->ly;
+    u8 sprite_height = LCDC_OBJ_HEIGHT;
+
+    for (int i = 0; i < ppu_get_context()->fetched_entry_count; i++) {
+        u8 ty = ((cur_y + 16) - ppu_get_context()->fetched_entries[i].y) * 2;
+
+        if (ppu_get_context()->fetched_entries[i].f_y_flip) {
+            ty = ((sprite_height * 2) - 2) - ty;
+        }
+
+        u8 tile_index = ppu_get_context()->fetched_entries[i].tile;
+
+        if (sprite_height == 16) {
+            tile_index &= ~(1);
+        }
+
+        ppu_get_context()->pfc.fetch_entry_data[(i * 2) + offset] =
+            bus_read(0x8000 + (tile_index * 16) + ty + offset);
+    }
+}
+
+void pipeline_load_window_tile() {
+    if(!window_visible()) 
+        return;
+        
+    u8 window_y = lcd_get_context()->win_y;
+
+    if (ppu_get_context()->pfc.fetch_x + 7 >= lcd_get_context()->win_x &&
+        ppu_get_context()->pfc.fetch_x + 7 < lcd_get_context()->win_x + YRES + 14) {
+        if(lcd_get_context()->ly >= window_y && lcd_get_context()->ly < window_y + XRES) {
+            u8 w_tile_y = ppu_get_context()->window_line / 8;
+
+
+            ppu_get_context()->pfc.bgw_fetch_data[0] = bus_read(LCDC_WIN_MAP_AREA + 
+                    ((ppu_get_context()->pfc.fetch_x + 7 - lcd_get_context()->win_x) / 8) +
+                    (w_tile_y * 32));
+
+            if(LCDC_BGW_DATA_AREA == 0x8800){
+                ppu_get_context()->pfc.bgw_fetch_data[0] += 128;
+            }
+        }
+    }
+}
+
+void pipeline_fetch() {
+    switch(ppu_get_context()->pfc.cur_fetch_state) {
+        case FS_TILE: {
+            ppu_get_context()->fetched_entry_count = 0;
+            if (LCDC_BGW_ENABLE) {
+                ppu_get_context()->pfc.bgw_fetch_data[0] = bus_read(LCDC_BG_MAP_AREA +
+                    (ppu_get_context()->pfc.map_x / 8) + 
+                    (((ppu_get_context()->pfc.map_y / 8)) * 32));
+
+                if(LCDC_BGW_DATA_AREA == 0x8800) {
+                    ppu_get_context()->pfc.bgw_fetch_data[0] += 128;
+                }
+
+                pipeline_load_window_tile();
+            }
+
+            if(LCDC_OBJ_ENABLE && ppu_get_context()->line_sprite_count) {
+                pipeline_load_sprite_tile();
+            }
+
+            ppu_get_context()->pfc.cur_fetch_state = FS_DATA0;
+            ppu_get_context()->pfc.fetch_x += 8;
+        } break;
+
+        case FS_DATA0: {
+            ppu_get_context()->pfc.bgw_fetch_data[1] = bus_read(LCDC_BGW_DATA_AREA +
+                (ppu_get_context()->pfc.bgw_fetch_data[0] * 16) +
+                ppu_get_context()->pfc.tile_y);
+
+            pipeline_load_sprite_data(0);
+
+            ppu_get_context()->pfc.cur_fetch_state = FS_DATA1;
+        } break;
+
+        case FS_DATA1: {
+            ppu_get_context()->pfc.bgw_fetch_data[2] = bus_read(LCDC_BGW_DATA_AREA +
+                (ppu_get_context()->pfc.bgw_fetch_data[0] * 16) +
+                ppu_get_context()->pfc.tile_y + 1);
+
+            pipeline_load_sprite_data(1);
+
+            ppu_get_context()->pfc.cur_fetch_state = FS_IDLE;
+        } break;
+
+        case FS_IDLE: {
+            ppu_get_context()->pfc.cur_fetch_state = FS_PUSH;
+        } break;
+
+        case FS_PUSH: {
+            if (pipeline_fifo_add()) {
+                ppu_get_context()->pfc.cur_fetch_state = FS_TILE;
+            }
+        } break;
+    }
+}
+
+void pipeline_push_pixel() {
+    if (ppu_get_context()->pfc.pixel_fifo.size > 8) {
+        u32 pixel_data = pixel_fifo_pop();
+
+        if(ppu_get_context()->pfc.line_x >= (lcd_get_context()->scroll_x % 8)) {
+            ppu_get_context()->video_buffer[ppu_get_context()->pfc.pushed_x +
+                (lcd_get_context()->ly * XRES)] = pixel_data;
+
+                ppu_get_context()->pfc.pushed_x++;
+        }
+
+        ppu_get_context()->pfc.line_x++;
+    }
+}
+
+void pipeline_process() {
+    ppu_get_context()->pfc.map_y = (lcd_get_context()->ly + lcd_get_context()->scroll_y);
+    ppu_get_context()->pfc.map_x = (ppu_get_context()->pfc.fetch_x + lcd_get_context()->scroll_x);
+    ppu_get_context()->pfc.tile_y = ((lcd_get_context()->ly + lcd_get_context()->scroll_y) % 8) * 2;
+
+    if(!(ppu_get_context()->line_ticks & 1)) {
+        pipeline_fetch();
+    }
+
+    pipeline_push_pixel();
+}
+
+void pipeline_fifo_reset() {
+    while(ppu_get_context()->pfc.pixel_fifo.size) {
+        pixel_fifo_pop();
+    }
+
+    ppu_get_context()->pfc.pixel_fifo.head = 0;
+}
\ No newline at end of file
diff --git a/lib/ppu_sm.c b/lib/ppu_sm.c
index b64e3e8..8953c85 100644
--- a/lib/ppu_sm.c
+++ b/lib/ppu_sm.c
@@ -2,8 +2,17 @@
 #include <lcd.h>
 #include <cpu.h>
 #include <interrupts.h>
+#include <string.h>
+#include <cart.h>
+
+void pipeline_fifo_reset();
+void pipeline_process();
 
 void increment_ly() {
+    if (window_visible() && lcd_get_context()->ly >= lcd_get_context()->win_y && lcd_get_context()->ly < lcd_get_context()->win_y + YRES) {
+        ppu_get_context()->window_line++;
+    }
+
     lcd_get_context()->ly++;
 
     if(lcd_get_context()->ly == lcd_get_context()->ly_compare) {
@@ -17,15 +26,93 @@ void increment_ly() {
     }
 }
 
+void load_line_sprites() {
+    int cur_y = lcd_get_context()->ly;
+
+    u8 sprite_height = LCDC_OBJ_HEIGHT;
+    memset(ppu_get_context()->line_entry_array, 0, sizeof(ppu_get_context()->line_entry_array));
+    for (int i=0; i<40; i++) {
+        oam_entry e = ppu_get_context()->oam_ram[i];
+        if (!e.x) {
+            //x = 0 means not visible
+            continue;
+        }
+
+        if (ppu_get_context()->line_sprite_count >= 10) {
+            //max of 10 sprites per line
+            break;
+        }
+
+        if(e.y <= cur_y + 16 && e.y + sprite_height > cur_y + 16) {
+            //this sprite is on the current line
+            oam_line_entry *entry = &ppu_get_context()->line_entry_array[
+                ppu_get_context()->line_sprite_count++
+            ];
+
+            entry->entry = e;
+            entry->next = NULL;
+
+            if(!ppu_get_context()->line_sprites ||
+                    ppu_get_context()->line_sprites->entry.x > e.x) {
+                entry->next = ppu_get_context()->line_sprites;
+                ppu_get_context()->line_sprites = entry;
+                continue;
+            }
+
+            //do some sorting
+
+            oam_line_entry *le = ppu_get_context()->line_sprites;
+            oam_line_entry *prev = le;
+
+            while(le) {
+                if (le->entry.x > e.x) {
+                    prev->next = entry;
+                    entry->next = le;
+                    break;
+                }
+
+                if(!le->next) {
+                    le->next = entry;
+                    break;
+                }
+
+                prev = le;
+                le = le->next;
+            }
+        }
+    }
+}
+
 void ppu_mode_oam() {
     if(ppu_get_context()->line_ticks >= 80) {
         LCDS_MODE_SET(MODE_XFER);
+
+        ppu_get_context()->pfc.cur_fetch_state = FS_TILE;
+        ppu_get_context()->pfc.line_x = 0;
+        ppu_get_context()->pfc.fetch_x = 0;
+        ppu_get_context()->pfc.pushed_x = 0;
+        ppu_get_context()->pfc.fifo_x = 0;
+    }
+
+    if(ppu_get_context()->line_ticks == 1) {
+        //read oam on the first tick only...
+        ppu_get_context()->line_sprites = 0;
+        ppu_get_context()->line_sprite_count = 0;
+
+        load_line_sprites();
     }
 }
 
 void ppu_mode_xfer() {
-    if(ppu_get_context()->line_ticks >= 80 + 172) {
+    pipeline_process();
+
+    if(ppu_get_context()->pfc.pushed_x >= XRES) {
+        pipeline_fifo_reset();
         LCDS_MODE_SET(MODE_HBLANK);
+
+        if(LCDS_STAT_INT(SS_HBLANK)) {
+            cpu_request_interrupt(IT_LCD_STAT);
+        }
     }
 }
 
@@ -36,13 +123,14 @@ void ppu_mode_vblank() {
         if(lcd_get_context()->ly >= LINES_PER_FRAME) {
             LCDS_MODE_SET(MODE_OAM);
             lcd_get_context()->ly = 0;
+            ppu_get_context()->window_line = 0;
         }
 
         ppu_get_context()->line_ticks = 0;
     }
 }
 
-static u32 target_frame_time = 1000/60;
+u32 target_frame_time = 1000/60;
 static long prev_frame_time = 0;
 static long start_timer = 0;
 static long frame_count = 0;
@@ -75,6 +163,9 @@ void ppu_mode_hblank() {
                 start_timer = end;
                 frame_count = 0;
                 printf("FPS: %ld\n", fps);
+                if(cart_need_save()){
+                    cart_battery_save();
+                }
             }
 
             frame_count++;
diff --git a/lib/ui.c b/lib/ui.c
index 1412216..493dbb7 100644
--- a/lib/ui.c
+++ b/lib/ui.c
@@ -1,5 +1,7 @@
 #include <ui.h>
 #include <emu.h>
+#include <ppu.h>
+#include <gamepad.h>
 
 #include <SDL.h>
 #include <SDL_ttf.h>
@@ -23,6 +25,19 @@ void ui_init(){
     printf("TTF INIT\n");
 
     SDL_CreateWindowAndRenderer(SCREEN_WIDTH, SCREEN_HEIGHT, 0, &sdlWindow, &sdlRenderer);
+
+    screen = SDL_CreateRGBSurface(0, SCREEN_WIDTH,
+                                    SCREEN_HEIGHT, 32,
+                                    0x00FF0000,
+                                    0x0000FF00,
+                                    0x000000FF,
+                                    0xFF000000);
+    sdlTexture = SDL_CreateTexture(sdlRenderer,
+                                            SDL_PIXELFORMAT_ARGB8888,
+                                            SDL_TEXTUREACCESS_STREAMING,
+                                            SCREEN_WIDTH,
+                                            SCREEN_HEIGHT);
+
     SDL_CreateWindowAndRenderer(16 * 8 * scale, 32 * 8 * scale, 0, &sdlDebugWindow, &sdlDebugRenderer);
     debugScreen = SDL_CreateRGBSurface(0, (16 * 8 * scale) + (16 * scale),
                                             (32 * 8 * scale) + (64 * scale), 32,
@@ -95,15 +110,73 @@ void update_debug_window() {
 }
 
 void ui_update() {
+    SDL_Rect rc;
+    rc.x = rc.y = 0;
+    rc.w = rc.h = 2048;
+
+    u32 *video_buffer = ppu_get_context()->video_buffer;
+
+    for(int line_num = 0; line_num < YRES; line_num++) {
+        for(int x = 0; x < XRES; x++) {
+            rc.x = x * scale;
+            rc.y = line_num * scale;
+            rc.w = scale;
+            rc.h = scale;
+
+            SDL_FillRect(screen, &rc, video_buffer[x + (line_num * XRES)]);
+        }
+    }
+
+    SDL_UpdateTexture(sdlTexture, NULL, screen->pixels, screen->pitch);
+    SDL_RenderClear(sdlRenderer);
+    SDL_RenderCopy(sdlRenderer, sdlTexture, NULL, NULL);
+    SDL_RenderPresent(sdlRenderer);
+
     update_debug_window();
 }
 
+static bool ff = false;
+
+void ui_on_key(bool down, u32 key_code) {
+    if(key_code == SDLK_SPACE && down == true) {
+        ff = !ff;
+        if(ff){
+            target_frame_time = 1000/300;
+        } else {
+            target_frame_time = 1000/60;
+        }
+        printf("target frame time: %d\n", target_frame_time);
+    }
+    switch(key_code){
+        case SDLK_z: gamepad_get_state()->b = down; break;
+        case SDLK_x: gamepad_get_state()->a = down; break;
+        case SDLK_RETURN: gamepad_get_state()->start = down; break;
+        case SDLK_TAB: gamepad_get_state()->select = down; break;
+
+        case SDLK_UP: gamepad_get_state()->up = down; break;
+        case SDLK_DOWN: gamepad_get_state()->down = down; break;
+        case SDLK_LEFT: gamepad_get_state()->left = down; break;
+        case SDLK_RIGHT: gamepad_get_state()->right = down; break;
+    }
+}
+
+
 void ui_handle_events(){
     SDL_Event e;
     while(SDL_PollEvent(&e) > 0) {
+
+        if(e.type == SDL_KEYDOWN) {
+            ui_on_key(true, e.key.keysym.sym);
+        }
+
+        if(e.type == SDL_KEYUP) {
+            ui_on_key(false, e.key.keysym.sym);
+        }
+
         if(e.type == SDL_WINDOWEVENT && e.window.event == SDL_WINDOWEVENT_CLOSE) {
             emu_get_context()->die = true;
         }
+
     }
 }
 
diff --git a/roms/01-read_timing.gb b/roms/01-read_timing.gb
new file mode 100644
index 0000000..d0836aa
Binary files /dev/null and b/roms/01-read_timing.gb differ
diff --git a/roms/02-write_timing.gb b/roms/02-write_timing.gb
new file mode 100644
index 0000000..cfe7c0f
Binary files /dev/null and b/roms/02-write_timing.gb differ
diff --git a/roms/03-modify_timing.gb b/roms/03-modify_timing.gb
new file mode 100644
index 0000000..28f1ae6
Binary files /dev/null and b/roms/03-modify_timing.gb differ
diff --git a/roms/instr_timing.gb b/roms/instr_timing.gb
new file mode 100644
index 0000000..61d2b20
Binary files /dev/null and b/roms/instr_timing.gb differ