ArDrone SDK 1.8 added

[mardrone] / mardrone / ARDrone_SDK_Version_1_8_20110726 / ARDroneLib / VLIB / Platform / arm9 / video_config.h

diff --git a/mardrone/ARDrone_SDK_Version_1_8_20110726/ARDroneLib/VLIB/Platform/arm9/video_config.h b/mardrone/ARDrone_SDK_Version_1_8_20110726/ARDroneLib/VLIB/Platform/arm9/video_config.h
new file mode 100644 (file)
index 0000000..a8beacd
--- /dev/null
+++ b/mardrone/ARDrone_SDK_Version_1_8_20110726/ARDroneLib/VLIB/Platform/arm9/video_config.h
@@ -0,0 +1,33 @@
+#ifndef _VIDEO_CONFIG_ARM9_H_
+#define _VIDEO_CONFIG_ARM9_H_
+
+# define CACHE_DLINESIZE (32)
+
+/* Default configuration for P5+ platform */
+
+#define VIDEO_DCT_INTERRUPT_ENABLE    (0)
+
+#define VIDEO_DCT_USE_INTRAM          (0)
+
+#define DEFAULT_QUANTIZATION          (6)
+
+/*
+  Help on finding a good MAX_NUM_MACRO_BLOCKS_PER_CALL on P5P
+  Maximum is 10 macroblocks because P5P DCT can only handle 64 blocks at once
+  This parameter is important and has to be adapted to your resolution
+  (CIF, VGA,... ) in order to minimize waiting time between two calls.
+  For example case like a call with a lot of blocks followed by a call
+  with a small number of blocks need to be avoided.
+  Ex: In QVGA a value of 6 is not optimal because there's 20 macroblocks per GOB
+  so there will be 3 calls with 6 macroblocks and a call with 2 macroblocks
+  (you can check with RTMON to monitor the waiting time represendted by event VIDEO_VLIB_DCT_WAIT_UEVENT)
+  In the QVGA case this is a better choice to choose a value of 7 to have two calls for 7 macroblocks and
+  a call for 6 macroblocks (of course 10 seems to be the better value - 2 calls - from an hardware DCT point of view).
+*/
+#define MAX_NUM_MACRO_BLOCKS_PER_CALL (6)
+
+#define DEFAULT_INTERNAL_STREAM_SIZE  (1024 * 8)
+
+#define VLIB_ALLOC_ALIGN              CACHE_DLINESIZE
+
+#endif // _VIDEO_CONFIG_ARM9_H_