diff --git a/components/display/core/gds.c b/components/display/core/gds.c
index c80f18d8..6cec45d2 100644
--- a/components/display/core/gds.c
+++ b/components/display/core/gds.c
@@ -61,9 +61,22 @@ void GDS_ClearWindow( struct GDS_Device* Device, int x1, int y1, int x2, int y2,
 	// driver can provide onw optimized clear window
 	if (Device->ClearWindow) {
 		Device->ClearWindow( Device, x1, y1, x2, y2, Color );
-	// cheap optimization on boundaries
-	if (x1 == 0 && x2 == Device->Width - 1 && y1 % 8 == 0 && (y2 + 1) % 8 == 0) {
-		memset( Device->Framebuffer + (y1 / 8) * Device->Width, 0, (y2 - y1 + 1) / 8 * Device->Width );
+	} else if (Device->Depth == 1) {
+		// single shot if we erase all screen
+		if (x2 - x1 == Device->Width - 1 && y2 - y1 == Device->Height - 1) {
+			memset( Device->Framebuffer, Color, Device->FramebufferSize );
+		} else {
+			uint8_t _Color = Color == GDS_COLOR_BLACK ? 0: 0xff;
+			uint8_t Width = Device->Width;
+			// try to do byte processing as much as possible
+			for (int c = x1; c <= x2; c++) {
+				int r = y1;
+				while (r & 0x07 && r <= y2) GDS_DrawPixelFast( Device, c, r++, Color);
+				for (; (r >> 3) < (y2 >> 3); r++) Device->Framebuffer[(r >> 3) * Width + c] = _Color;
+				// memset(Device->Framebuffer + (r >> 3) * Width, _Color, (y2 >> 3) - (r >> 3));
+				while (r <= y2) GDS_DrawPixelFast( Device, c, r++, Color);
+			}
+		}	
 	} else {
 		for (int y = y1; y <= y2; y++) {
 			for (int x = x1; x <= x2; x++) {
diff --git a/components/display/core/gds.h b/components/display/core/gds.h
index 88156279..2c3a9e26 100644
--- a/components/display/core/gds.h
+++ b/components/display/core/gds.h
@@ -4,6 +4,15 @@
 #include <stdint.h>
 #include <stdbool.h>
 
+/* NOTE for drivers:
+ The build-in DrawPixel(Fast), DrawCBR and ClearWindow are optimized for 1 bit 
+ and 4 bits screen depth. For any other type of screen, DrawCBR and ClearWindow
+ default to use DrawPixel, which is very sub-optimal. For such other depth, you 
+ must supply the DrawPixelFast. The built-in 1 bit depth function are only for 
+ screen with vertical framing (1 byte = 8 lines). For example SSD1326 in 
+ monochrome mode is not such type of screen, SH1106 and SSD1306 are
+*/ 
+
 enum { 	GDS_COLOR_L0 = 0, GDS_COLOR_L1, GDS_COLOR_L2, GDS_COLOR_L3, GDS_COLOR_L4, GDS_COLOR_L5, GDS_COLOR_L6, GDS_COLOR_L7, 
 		GDS_COLOR_L8, GDS_COLOR_L9, GDS_COLOR_L10, GDS_COLOR_L11, GDS_COLOR_L12, GDS_COLOR_L13, GDS_COLOR_L14, GDS_COLOR_L15,
 };
diff --git a/components/display/core/gds_private.h b/components/display/core/gds_private.h
index 806f5a90..3e2676ff 100644
--- a/components/display/core/gds_private.h
+++ b/components/display/core/gds_private.h
@@ -93,7 +93,7 @@ struct GDS_Device {
     bool FontForceMonospace;
 
 	// various driver-specific method
-	// must provide
+	// must always provide 
 	bool (*Init)( struct GDS_Device* Device);
 	void (*SetContrast)( struct GDS_Device* Device, uint8_t Contrast );
 	void (*DisplayOn)( struct GDS_Device* Device );
@@ -101,8 +101,9 @@ struct GDS_Device {
 	void (*SetHFlip)( struct GDS_Device* Device, bool On );
 	void (*SetVFlip)( struct GDS_Device* Device, bool On );
 	void (*Update)( struct GDS_Device* Device );
-	// may provide for optimization
+	// must provide for depth other than 1 (vertical) and 4 (may provide for optimization)
 	void (*DrawPixelFast)( struct GDS_Device* Device, int X, int Y, int Color );
+	// may provide for optimization
 	void (*DrawRGB16)( struct GDS_Device* Device, int x, int y, int Width, int Height, int RGB_Mode, uint16_t **Image );
 	void (*ClearWindow)( struct GDS_Device* Device, int x1, int y1, int x2, int y2, int Color );