commit 4aa0753

Faster sprite drawing

 1@@ -525,24 +525,42 @@ screen_deo_sprite(void)
 2 	const int row_start = flipx ? 0 : 7, row_delta = flipx ? 1 : -1;
 3 	const int col_start = flipy ? 7 : 0, col_delta = flipy ? -1 : 1;
 4 	const int layer = ctrl & 0x40, layer_mask = layer ? 0x3 : 0xc;
 5-	const int is_2bpp = ctrl & 0x80;
 6-	const int addr_incr = rMA << (is_2bpp ? 2 : 1);
 7 	const int stride = screen_wmar2;
 8 	const int blend = ctrl & 0xf;
 9 	const Uint8 opaque_mask = blend % 5;
10 	const Uint8 *table = blend_lut[blend][layer >> 6];
11-	for(i = 0; i <= rML; i++, x += dx, y += dy, rA += addr_incr) {
12-		const Uint16 x0 = x + 8, y0 = y + 8;
13-		if(x0 + 8 >= stride || y0 + 8 >= screen_hmar2) continue;
14-		Uint8 *dst = screen_layers + y0 * stride + x0;
15-		const Uint8 *col = &ram[rA + col_start];
16-		for(j = 0; j < 8; j++, dst += stride, col += col_delta) {
17-			Uint8 *d = dst;
18-			const int ch1 = *col, ch2 = is_2bpp ? col[8] : 0;
19-			for(int k = 0, row = row_start; k < 8; k++, d++, row += row_delta) {
20-				const int color = ((ch1 >> row) & 1) | (((ch2 >> row) & 1) << 1);
21-				if(opaque_mask || color)
22-					*d = (*d & layer_mask) | table[color];
23+	if(ctrl & 0x80) {
24+		const int addr_incr = rMA << 2;
25+		for(i = 0; i <= rML; i++, x += dx, y += dy, rA += addr_incr) {
26+			const Uint16 x0 = x + 8, y0 = y + 8;
27+			if(x0 + 8 >= stride || y0 + 8 >= screen_hmar2) continue;
28+			Uint8 *dst = screen_layers + y0 * stride + x0;
29+			const Uint8 *col = &ram[rA + col_start];
30+			for(j = 0; j < 8; j++, dst += stride, col += col_delta) {
31+				Uint8 *d = dst;
32+				const int ch1 = *col, ch2 = col[8];
33+				for(int k = 0, row = row_start; k < 8; k++, d++, row += row_delta) {
34+					const int color = ((ch1 >> row) & 1) | (((ch2 >> row) & 1) << 1);
35+					if(opaque_mask || color)
36+						*d = (*d & layer_mask) | table[color];
37+				}
38+			}
39+		}
40+	} else {
41+		const int addr_incr = rMA << 1;
42+		for(i = 0; i <= rML; i++, x += dx, y += dy, rA += addr_incr) {
43+			const Uint16 x0 = x + 8, y0 = y + 8;
44+			if(x0 + 8 >= stride || y0 + 8 >= screen_hmar2) continue;
45+			Uint8 *dst = screen_layers + y0 * stride + x0;
46+			const Uint8 *col = &ram[rA + col_start];
47+			for(j = 0; j < 8; j++, dst += stride, col += col_delta) {
48+				Uint8 *d = dst;
49+				const int ch1 = *col;
50+				for(int k = 0, row = row_start; k < 8; k++, d++, row += row_delta) {
51+					const int color = (ch1 >> row) & 1;
52+					if(opaque_mask || color)
53+						*d = (*d & layer_mask) | table[color];
54+				}
55 			}
56 		}
57 	}