diff --git a/.vscode/.cortex-debug.peripherals.state.json b/.vscode/.cortex-debug.peripherals.state.json
index 0637a08..462db0e 100644
--- a/.vscode/.cortex-debug.peripherals.state.json
+++ b/.vscode/.cortex-debug.peripherals.state.json
@@ -1 +1 @@
-[]
\ No newline at end of file
+[{"node":"PIO0","expanded":true,"format":0,"pinned":false}]
\ No newline at end of file
diff --git a/.vscode/settings.json b/.vscode/settings.json
index 6e3a574..62d4392 100644
--- a/.vscode/settings.json
+++ b/.vscode/settings.json
@@ -18,5 +18,6 @@
     },
     "cmake.buildBeforeRun": true,
     "C_Cpp.default.configurationProvider": "ms-vscode.cmake-tools",
-    "C_Cpp.errorSquiggles": "disabled"
+    "C_Cpp.errorSquiggles": "disabled",
+    "cortex-debug.variableUseNaturalFormat": false
 }
\ No newline at end of file
diff --git a/CMakeLists.txt b/CMakeLists.txt
index f0e1f48..0d7dd67 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -9,6 +9,8 @@ pico_generate_pio_header(clacer ${CMAKE_CURRENT_SOURCE_DIR}/src/tft.pio)
 target_sources(clacer PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/src/main.c)
 file(GLOB SOURCES "src/*.c")
 target_sources(clacer PRIVATE ${SOURCES})
+file(GLOB SOURCES "src/*.pio")
+target_sources(clacer PRIVATE ${SOURCES})
 target_link_libraries(clacer pico_stdlib pico_time hardware_pio hardware_dma)
 set_target_properties(clacer PROPERTIES RUNTIME_OUTPUT_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}/build)
 pico_add_extra_outputs(clacer)
diff --git a/src/main.c b/src/main.c
index 614ca33..70fa9ff 100644
--- a/src/main.c
+++ b/src/main.c
@@ -15,40 +15,35 @@ const struct tft_config lcd_config = {
 	.gpio_rst = 20,
 	.gpio_bl = 16,
 	.width = 320,
-	.height = 480,
+	.height = 480
 };
 
 int main()
 {
-	set_sys_clock_khz(200 * 1000, true);
+	set_sys_clock_khz(100 * 1000, true);
 
     uint offset = pio_add_program(lcd_config.pio, &tft_program);
-    tft_program_init(lcd_config.pio, lcd_config.sm, offset, lcd_config.gpio_din, lcd_config.gpio_clk, 1.f);
+    tft_program_init(lcd_config.pio, lcd_config.sm, offset, lcd_config.gpio_din, lcd_config.gpio_clk, 10.f);
 	
 	int dma_chan = dma_claim_unused_channel(true);
 	dma_channel_config c = dma_channel_get_default_config(dma_chan);
 	channel_config_set_transfer_data_size(&c, DMA_SIZE_32);
+	channel_config_set_read_increment(&c, true);
 	channel_config_set_dreq(&c, DREQ_PIO0_TX0);
 	dma_channel_configure(
 		dma_chan,
 		&c,
-		&pio0_hw->txf[lcd_config.sm],
+		&pio0->txf[lcd_config.sm],
 		NULL,
-		4096,
+		320*480/12,
 		false
 	);
-	
+	dled_start(25);	
 	tft_init(&lcd_config, dma_chan);
-	
-	while (true) {
-		sleep_ms(100);
- 		absolute_time_t start = get_absolute_time();	
-		tft_fill(0b0000000000000000);
-		tft_fill(0b1111111111111111);
-		tft_fill(0b0000000000000000);
-		absolute_time_t end = get_absolute_time();
-		volatile int64_t diff = absolute_time_diff_us(start, end);
-	}
+	sleep_ms(100);
+	tft_fill(16000);
+ 	dma_channel_wait_for_finish_blocking(dma_chan);
+	while (true);
 	return 0;
 }
 void dled_start(uint pin){
diff --git a/src/tft.c b/src/tft.c
index b1a8dcc..be9b10a 100644
--- a/src/tft.c
+++ b/src/tft.c
@@ -17,8 +17,9 @@ void tft_write_cmd(PIO pio, uint sm, const uint8_t *cmd, size_t count) {
     if (count >= 2) {
         tft_wait_idle(pio, sm);
         tft_set_dc_cs(1, 0);
-        for (size_t i = 0; i < count - 1; ++i)
+        for (size_t i = 0; i < count - 1; ++i) {
             tft_put(pio, sm, *cmd++);
+        }
     }
     tft_wait_idle(pio, sm);
     tft_set_dc_cs(1, 1);
@@ -28,14 +29,14 @@ void lcd_init(PIO pio, uint sm, const uint8_t *init_seq) {
     const uint8_t *cmd = init_seq;
     while (*cmd) {
         tft_write_cmd(pio, sm, cmd + 2, *cmd);
-        //sleep_ms(*(cmd + 1) * 5);
-        sleep_ms(200);
+        sleep_ms(*(cmd + 1) * 5);
+//        sleep_ms(200);
         cmd += *cmd + 2;
     }
 }
 
 void tft_start_pixels(PIO pio, uint sm) {
-    uint8_t cmd = 0x2c; // RAMWR
+    const uint8_t cmd = 0x2c; // RAMWR
     tft_write_cmd(pio, sm, &cmd, 1);
     tft_set_dc_cs(1, 0);
 }
@@ -55,18 +56,19 @@ void tft_init(const struct tft_config* config, int dma_channel) {
 
     gpio_put(tft_cfg.gpio_cs, 1);
     gpio_put(tft_cfg.gpio_rst, 1);
-    lcd_init(tft_cfg.pio, tft_cfg.sm, st7789_init_seq);
+    //lcd_init(tft_cfg.pio, tft_cfg.sm, st7789_init_seq);
     gpio_put(tft_cfg.gpio_bl, 1);
 }
 void tft_fill(uint16_t color) {
-	tft_start_pixels(tft_cfg.pio, tft_cfg.sm);
-	size_t buf_size = 4096;
-	uint16_t* pix_buf = malloc(buf_size * sizeof(uint16_t));	
+    uint32_t color_left = ((uint32_t)color) << 16;
+    uint32_t two = 0;
+    two |= color_left;
+    two |= color;
+	//tft_start_pixels(tft_cfg.pio, tft_cfg.sm);
+    uint32_t pix_buf[320*480/12]; 
 	size_t i;
-	for(i = 0; i < buf_size; i++) {
-		pix_buf[i] = color;
+	for(i = 0; i < 320*480/12; i++) {
+		pix_buf[i] = two;
 	}
-	tft_push(tft_cfg.pio, tft_cfg.sm, dma_chan, pix_buf, buf_size);	
-	tft_push(tft_cfg.pio, tft_cfg.sm, dma_chan, pix_buf, buf_size);	
-	free(pix_buf);
+    dma_channel_set_read_addr(dma_chan, pix_buf, true);
 }
\ No newline at end of file
diff --git a/src/tft.h b/src/tft.h
index 33275c3..5507965 100644
--- a/src/tft.h
+++ b/src/tft.h
@@ -42,4 +42,5 @@ void lcd_init(PIO pio, uint sm, const uint8_t *init_seq);
 void tft_start_pixels(PIO pio, uint sm);
 void tft_init(const struct tft_config* config, int dma_channel);
 void tft_fill(uint16_t color);
+void tft_push(PIO pio, uint sm, int dma_chan, uint32_t pixels[]);
 #endif // TFT_H
\ No newline at end of file
diff --git a/src/tft.pio b/src/tft.pio
index e9f13cf..6a8a534 100644
--- a/src/tft.pio
+++ b/src/tft.pio
@@ -41,6 +41,7 @@ static inline void tft_program_init(PIO pio, uint sm, uint offset, uint data_pin
 
 // Making use of the narrow store replication behaviour on RP2040 to get the
 // data left-justified (as we are using shift-to-left to get MSB-first serial)
+// => command gets sent 4 times
 
 static inline void tft_put(PIO pio, uint sm, uint8_t x) {
     while (pio_sm_is_tx_fifo_full(pio, sm))
@@ -48,18 +49,6 @@ static inline void tft_put(PIO pio, uint sm, uint8_t x) {
     *(volatile uint8_t*)&pio->txf[sm] = x;
 }
 
-static inline void tft_push(PIO pio, uint sm, int dma_chan, uint16_t* pixels, size_t size) {
-    size_t new_size = size / 2;
-    uint32_t* x = malloc(new_size * sizeof(uint32_t));
-    
-    for (size_t i = 0, j = 0; i < size; i += 2, j++) {
-        uint32_t combinedValue = ((uint32_t)pixels[i] << 16) | pixels[i + 1];
-        x[j] = combinedValue;
-    }
-    dma_channel_set_read_addr(dma_chan, x, true);
-    sleep_ms(100);
-    free(x);
-}
 
 // SM is done when it stalls on an empty FIFO