Some optimizations on integer affine estimation
1. Adds a limit on number of candidate samples used for the
estimation.
2. Adds a limit on max mv magnitude for use in the least-squares
3. Makes some of the internal variables 32-bit.
Impact on coding efficiency in the noise range.
Change-Id: I8c1c3216368ceb2e3548660a3b8c159df54a8312
diff --git a/aom_ports/mem.h b/aom_ports/mem.h
index 4789d29..7316399 100644
--- a/aom_ports/mem.h
+++ b/aom_ports/mem.h
@@ -46,6 +46,14 @@
(((value) < 0) ? -ROUND_POWER_OF_TWO(-(value), (n)) \
: ROUND_POWER_OF_TWO((value), (n)))
+/* Shift down with rounding for use when n >= 0, value >= 0 for (64 bit) */
+#define ROUND_POWER_OF_TWO_64(value, n) \
+ (((value) + ((((int64_t)1 << (n)) >> 1))) >> (n))
+/* Shift down with rounding for signed integers, for use when n >= 0 (64 bit) */
+#define ROUND_POWER_OF_TWO_SIGNED_64(value, n) \
+ (((value) < 0) ? -ROUND_POWER_OF_TWO_64(-(value), (n)) \
+ : ROUND_POWER_OF_TWO_64((value), (n)))
+
#define ALIGN_POWER_OF_TWO(value, n) \
(((value) + ((1 << (n)) - 1)) & ~((1 << (n)) - 1))
