Simplify rotate implementation a bit..

Much more understandable/readable as a result, and probably more efficient.

2 files changed, 37 insertions, 21 deletions

diff --git a/generic/include/clc/integer/gentype.inc b/generic/include/clc/integer/gentype.inc
index 005b9af..dd7d061 100644
--- a/generic/include/clc/integer/gentype.inc
+++ b/generic/include/clc/integer/gentype.inc
@@ -2,7 +2,9 @@
 #define GENTYPE char
 #define UGENTYPE uchar
 #define SGENTYPE char
+#define SCALAR 1
 #include BODY
+#undef SCALAR
 #undef GENTYPE
 #undef UGENTYPE
 #undef SGENTYPE
@@ -50,7 +52,9 @@
 #define GENTYPE uchar
 #define UGENTYPE uchar
 #define SGENTYPE char
+#define SCALAR 1
 #include BODY
+#undef SCALAR
 #undef GENTYPE
 #undef UGENTYPE
 #undef SGENTYPE
@@ -101,7 +105,9 @@
 #define GENTYPE short
 #define UGENTYPE ushort
 #define SGENTYPE short
+#define SCALAR 1
 #include BODY
+#undef SCALAR
 #undef GENTYPE
 #undef UGENTYPE
 #undef SGENTYPE
@@ -149,7 +155,9 @@
 #define GENTYPE ushort
 #define UGENTYPE ushort
 #define SGENTYPE short
+#define SCALAR 1
 #include BODY
+#undef SCALAR
 #undef GENTYPE
 #undef UGENTYPE
 #undef SGENTYPE
@@ -200,7 +208,9 @@
 #define GENTYPE int
 #define UGENTYPE uint
 #define SGENTYPE int
+#define SCALAR 1
 #include BODY
+#undef SCALAR
 #undef GENTYPE
 #undef UGENTYPE
 #undef SGENTYPE
@@ -248,7 +258,9 @@
 #define GENTYPE uint
 #define UGENTYPE uint
 #define SGENTYPE int
+#define SCALAR 1
 #include BODY
+#undef SCALAR
 #undef GENTYPE
 #undef UGENTYPE
 #undef SGENTYPE
@@ -299,7 +311,9 @@
 #define GENTYPE long
 #define UGENTYPE ulong
 #define SGENTYPE long
+#define SCALAR 1
 #include BODY
+#undef SCALAR
 #undef GENTYPE
 #undef UGENTYPE
 #undef SGENTYPE
@@ -347,7 +361,9 @@
 #define GENTYPE ulong
 #define UGENTYPE ulong
 #define SGENTYPE long
+#define SCALAR 1
 #include BODY
+#undef SCALAR
 #undef GENTYPE
 #undef UGENTYPE
 #undef SGENTYPE
diff --git a/generic/lib/integer/rotate.inc b/generic/lib/integer/rotate.inc
index e83dd51..7792a97 100644
--- a/generic/lib/integer/rotate.inc
+++ b/generic/lib/integer/rotate.inc
@@ -11,25 +11,25 @@ _CLC_OVERLOAD _CLC_DEF GENTYPE rotate(GENTYPE x, GENTYPE n){
     //Try to avoid extra work if someone's spinning the value through multiple
     //full rotations
     n = n % (GENTYPE)GENSIZE;
-    
-    //Determine if we're doing a right or left shift on each component
-    //The actual shift algorithm is based on a rotate right
-    //e.g. a rotate of int by 5 bits becomes rotate right by 26 bits
-    //     and a rotate of int by -4 bits becomes rotate right by 4
-    GENTYPE amt = (n > (GENTYPE)0 ? (GENTYPE)GENSIZE - n : (GENTYPE)0 - n );
-    
-    //Calculate the bits that will wrap
-    GENTYPE mask = ( (GENTYPE)1 << amt ) - (GENTYPE)1;
-    GENTYPE wrapped_bits = x & mask;
-    
-    //Shift the input value right and then AND a mask that eliminates
-    //sign-extension interference
-    //if the rotate amount is 0, just use ~0 for a mask
-    GENTYPE se_mask = !amt ? ~((GENTYPE)0) : 
-        ( ( (GENTYPE)1 << ((GENTYPE)GENSIZE - amt) ) - (GENTYPE)1 );
-    GENTYPE unwrapped_bits = x >> amt;
-    unwrapped_bits &= se_mask;
-    
-    //Finally shift the input right after moving the wrapped bits into position
-    return unwrapped_bits | (wrapped_bits << ( (GENTYPE)GENSIZE - amt ) );
+
+#ifdef SCALAR
+    if (n > 0){
+        return (x << n) | (((UGENTYPE)x) >> (GENSIZE - n));
+    } else if (n == 0){
+        return x;
+    } else {
+        return ( (((UGENTYPE)x) >> -n) | (x << (GENSIZE + n)) );
+    }
+#else
+    UGENTYPE x_1 = __builtin_astype(x, UGENTYPE);
+
+    UGENTYPE amt;
+    amt = (n < (GENTYPE)0 ? __builtin_astype((GENTYPE)0-n, UGENTYPE) : (UGENTYPE)0);
+    x_1 = (x_1 >> amt) | (x_1 << ((UGENTYPE)GENSIZE - amt));
+
+    amt = (n < (GENTYPE)0 ? (UGENTYPE)0 : __builtin_astype(n, UGENTYPE));
+    x_1 = (x_1 << amt) | (x_1 >> ((UGENTYPE)GENSIZE - amt));
+
+    return __builtin_astype(x_1, GENTYPE);
+#endif
 }
\ No newline at end of file