Patch #4829 from Fredrik Öhrn - Coldfire assembler blit_display() function for H300 and move the palette data into IRAM. Each of these gives about a 1fps speed improvement.

git-svn-id: svn://svn.rockbox.org/rockbox/trunk@9038 a1c6a512-1295-4272-9138-f99709370657

1 files changed, 155 insertions, 0 deletions

diff --git a/apps/plugins/pacbox/pacbox_cf.S b/apps/plugins/pacbox/pacbox_cf.S
new file mode 100644
index 0000000..7dc6140
--- /dev/null
+++ b/apps/plugins/pacbox/pacbox_cf.S
@@ -0,0 +1,155 @@
+/***************************************************************************
+ *             __________               __   ___.
+ *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
+ *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
+ *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
+ *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
+ *                     \/            \/     \/    \/            \/
+ * $Id$
+ *
+ * Copyright (c) 2006 Fredrik Ohrn
+ *
+ * All files in this archive are subject to the GNU General Public License.
+ * See the file COPYING in the source tree root for full license agreement.
+ *
+ * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
+ * KIND, either express or implied.
+ *
+ ****************************************************************************/
+
+#include "pacbox.h"
+
+	.section .text
+	.align  2
+	.global blit_display
+
+/*
+
+ 0.75 scaling - display 3 out of 4 pixels = 216x168
+ Skipping pixel #2 out of 4 seems to give the most legible display
+ 
+void blit_display(fb_data* lcd_framebuffer, unsigned char* vbuf)
+{
+    next_dst=&lcd_framebuffer[YOFS*LCD_WIDTH+XOFS+((ScreenHeight*3)/4)-1];
+    for (y=ScreenHeight-1;y >= 0; y--) {
+        if ((y & 3) != 1) {
+            dst = (next_dst--);
+            for (x=0;x<ScreenWidth;x++) {
+                if ((x & 3) == 1) { vbuf++; }
+                else {
+                   *dst = palette[*(vbuf++)];
+                   dst+=LCD_WIDTH;
+                }
+            }
+        } else {
+            vbuf+=ScreenWidth;
+        }
+    }
+}
+*/
+
+/* FIXME: ScreenWidth & ScreenHeight are defined in arcade.h, but all
+          hell breaks loose if you include it... */
+
+#define ScreenWidth 224
+#define ScreenHeight 288
+
+#define SHRUNK_WIDTH ((ScreenWidth/4)*3)
+#define SHRUNK_HEIGHT ((ScreenHeight/4)*3)
+
+blit_display:
+	lea	-28(%sp), %sp
+	movem.l	%d2-%d6/%a2, (%sp)
+
+	move.l	32(%sp), %a0	// lcd_framebuffer
+	move.l	36(%sp), %a2	// vbuf
+
+	lea	palette, %a1
+
+	lea.l	((YOFS * LCD_WIDTH + XOFS + SHRUNK_HEIGHT) * 2)(%a0), %a0
+
+	move.l	#(ScreenHeight / 4), %d0
+	move.l	#(ScreenWidth / 4), %d1
+	moveq.l	#0, %d2
+
+y_loop:
+	move.l	%d1, %d5
+x_loop1:
+	move.l	(%a2)+, %d3
+
+	move.b	%d3, %d2
+	move.w	(%a1, %d2.l * 2), %d6
+	move.w	%d6, (2 * LCD_WIDTH * 2)(%a0)
+
+	lsr.l	#8, %d3
+	move.b	%d3, %d2
+	move.w	(%a1, %d2.l * 2), %d6
+	move.w	%d6, (1 * LCD_WIDTH * 2)(%a0)
+
+	swap.w	%d3
+	move.b	%d3, %d2
+	move.w	(%a1, %d2.l * 2), (%a0)
+
+	lea.l	3 * LCD_WIDTH * 2(%a0), %a0
+
+	subq.l	#1, %d5
+	bne.s	x_loop1
+
+	suba.l	#(LCD_WIDTH * SHRUNK_WIDTH + 1) * 2, %a0
+	lea.l	ScreenWidth(%a2), %a2	// Skip 1 line
+
+	move.l	%d1, %d5
+x_loop2:
+	move.l	(%a2)+, %d3
+
+	move.b	%d3, %d2
+	move.w	(%a1, %d2.l * 2), %d6
+	move.w	%d6, 2 * LCD_WIDTH * 2(%a0)
+
+	lsr.l	#8, %d3
+	move.b	%d3, %d2
+	move.w	(%a1, %d2.l * 2), %d6
+	move.w	%d6, 1 * LCD_WIDTH * 2(%a0)
+
+	swap.w	%d3
+	move.b	%d3, %d2
+	move.w	(%a1, %d2.l * 2), (%a0)
+
+	lea.l	3 * LCD_WIDTH * 2(%a0), %a0
+
+	subq.l	#1, %d5
+	bne.s	x_loop2
+
+	suba.l	#(LCD_WIDTH * SHRUNK_WIDTH + 1) * 2, %a0
+
+	move.l	%d1, %d5
+x_loop3:
+	move.l	(%a2)+, %d3
+
+	move.b	%d3, %d2
+	move.w	(%a1, %d2.l * 2), %d6
+	move.w	%d6, 2 * LCD_WIDTH * 2(%a0)
+
+	lsr.l	#8, %d3
+	move.b	%d3, %d2
+	move.w	(%a1, %d2.l * 2), %d6
+	move.w	%d6, 1 * LCD_WIDTH * 2(%a0)
+
+	swap.w	%d3
+	move.b	%d3, %d2
+	move.w	(%a1, %d2.l * 2), (%a0)
+
+	lea.l	3 * LCD_WIDTH * 2(%a0), %a0
+
+	subq.l	#1, %d5
+	bne.s	x_loop3
+
+	suba.l	#(LCD_WIDTH * SHRUNK_WIDTH + 1) * 2, %a0
+
+	subq.l	#1, %d0
+	bne	y_loop
+
+	movem.l	(%sp), %d2-%d6/%a2
+	lea.l	28(%sp), %sp
+	rts
+