Add rhadd builtin

rhadd = (x+y+1)>>1

Implemented as:
(x>>1) + (y>>1) + ((x&1)|(y&1))

This prevents us having to do assembly addition and overflow detection

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 188477

2 files changed, 10 insertions, 0 deletions

diff --git a/libclc/generic/lib/integer/rhadd.cl b/libclc/generic/lib/integer/rhadd.cl
new file mode 100644
index 00000000000..c985870f7c7
--- /dev/null
+++ b/libclc/generic/lib/integer/rhadd.cl
@@ -0,0 +1,4 @@
+#include <clc/clc.h>
+
+#define __CLC_BODY <rhadd.inc>
+#include <clc/integer/gentype.inc>
diff --git a/libclc/generic/lib/integer/rhadd.inc b/libclc/generic/lib/integer/rhadd.inc
new file mode 100644
index 00000000000..3d607687480
--- /dev/null
+++ b/libclc/generic/lib/integer/rhadd.inc
@@ -0,0 +1,6 @@
+//rhadd = (x+y+1)>>1
+//This can be simplified to x>>1 + y>>1 + (1 if either x or y have the 1s bit set)
+//This saves us having to do any checks for overflow in the addition sums
+_CLC_OVERLOAD _CLC_DEF __CLC_GENTYPE rhadd(__CLC_GENTYPE x, __CLC_GENTYPE y) {
+    return (x>>(__CLC_GENTYPE)1)+(y>>(__CLC_GENTYPE)1)+((x&(__CLC_GENTYPE)1)|(y&(__CLC_GENTYPE)1));
+}