[libclc] Move mad to the CLC library (#123607)

All targets build `__clc_mad` -- even SPIR-V targets -- since it
compiles to the optimal `llvm.fmuladd` intrinsic. There is no change to
the bytecode generated for non-SPIR-V targets.

The `mix` builtin, which is implemented as a wrapper around `mad`, is
left as an OpenCL-layer wrapper of `__clc_mad`. I don't know if it's
worth having a specific CLC version of `mix`.

The changes to the other CLC files/functions are moving uses of `mad` to
`__clc_mad`, and reformatting. There is an additional instance of
`trunc` becoming `__clc_trunc`, which was missed before.

Author: frasercrmck

libclc/clc/include/clc/clcmacro.h

@@ -184,6 +184,33 @@
     return BUILTIN(x);                                                         \
   }
 
+#define _CLC_DEFINE_TERNARY_BUILTIN(RET_TYPE, FUNCTION, BUILTIN, ARG1_TYPE,    \
+                                    ARG2_TYPE, ARG3_TYPE)                      \
+  _CLC_DEF _CLC_OVERLOAD RET_TYPE FUNCTION(ARG1_TYPE x, ARG2_TYPE y,           \
+                                           ARG3_TYPE z) {                      \
+    return BUILTIN(x, y, z);                                                   \
+  }                                                                            \
+  _CLC_DEF _CLC_OVERLOAD RET_TYPE##2 FUNCTION(ARG1_TYPE##2 x, ARG2_TYPE##2 y,  \
+                                              ARG3_TYPE##2 z) {                \
+    return BUILTIN(x, y, z);                                                   \
+  }                                                                            \
+  _CLC_DEF _CLC_OVERLOAD RET_TYPE##3 FUNCTION(ARG1_TYPE##3 x, ARG2_TYPE##3 y,  \
+                                              ARG3_TYPE##3 z) {                \
+    return BUILTIN(x, y, z);                                                   \
+  }                                                                            \
+  _CLC_DEF _CLC_OVERLOAD RET_TYPE##4 FUNCTION(ARG1_TYPE##4 x, ARG2_TYPE##4 y,  \
+                                              ARG3_TYPE##4 z) {                \
+    return BUILTIN(x, y, z);                                                   \
+  }                                                                            \
+  _CLC_DEF _CLC_OVERLOAD RET_TYPE##8 FUNCTION(ARG1_TYPE##8 x, ARG2_TYPE##8 y,  \
+                                              ARG3_TYPE##8 z) {                \
+    return BUILTIN(x, y, z);                                                   \
+  }                                                                            \
+  _CLC_DEF _CLC_OVERLOAD RET_TYPE##16 FUNCTION(                                \
+      ARG1_TYPE##16 x, ARG2_TYPE##16 y, ARG3_TYPE##16 z) {                     \
+    return BUILTIN(x, y, z);                                                   \
+  }
+
 #ifdef cl_khr_fp16
 
 #pragma OPENCL EXTENSION cl_khr_fp16 : enable

libclc/clc/include/clc/math/clc_mad.h

@@ -0,0 +1,12 @@
+#ifndef __CLC_MATH_CLC_MAD_H__
+#define __CLC_MATH_CLC_MAD_H__
+
+#define __CLC_BODY <clc/math/ternary_decl.inc>
+#define __CLC_FUNCTION __clc_mad
+
+#include <clc/math/gentype.inc>
+
+#undef __CLC_BODY
+#undef __CLC_FUNCTION
+
+#endif // __CLC_MATH_CLC_MAD_H__

libclc/clc/include/clc/math/ternary_decl.inc

@@ -0,0 +1,3 @@
+_CLC_OVERLOAD _CLC_DECL __CLC_GENTYPE __CLC_FUNCTION(__CLC_GENTYPE a,
+                                                     __CLC_GENTYPE b,
+                                                     __CLC_GENTYPE c);

libclc/clc/lib/clspv/SOURCES

@@ -1,6 +1,7 @@
 ../generic/math/clc_ceil.cl
 ../generic/math/clc_fabs.cl
 ../generic/math/clc_floor.cl
+../generic/math/clc_mad.cl
 ../generic/math/clc_rint.cl
 ../generic/math/clc_trunc.cl
 ../generic/shared/clc_clamp.cl

libclc/clc/lib/generic/SOURCES

@@ -7,6 +7,7 @@ integer/clc_abs_diff.cl
 math/clc_ceil.cl
 math/clc_fabs.cl
 math/clc_floor.cl
+math/clc_mad.cl
 math/clc_rint.cl
 math/clc_trunc.cl
 relational/clc_all.cl

libclc/clc/lib/generic/math/clc_mad.cl

@@ -0,0 +1,4 @@
+#include <clc/internal/clc.h>
+
+#define __CLC_BODY <clc_mad.inc>
+#include <clc/math/gentype.inc>

libclc/clc/lib/generic/math/clc_mad.inc

@@ -0,0 +1,5 @@
+_CLC_OVERLOAD _CLC_DEF __CLC_GENTYPE __clc_mad(__CLC_GENTYPE a, __CLC_GENTYPE b,
+                                               __CLC_GENTYPE c) {
+#pragma OPENCL FP_CONTRACT ON
+  return a * b + c;
+}

libclc/clc/lib/spirv/SOURCES

@@ -5,6 +5,7 @@
 ../generic/math/clc_ceil.cl
 ../generic/math/clc_fabs.cl
 ../generic/math/clc_floor.cl
+../generic/math/clc_mad.cl
 ../generic/math/clc_rint.cl
 ../generic/math/clc_trunc.cl
 ../generic/shared/clc_clamp.cl

libclc/clc/lib/spirv64/SOURCES

@@ -5,6 +5,7 @@
 ../generic/math/clc_ceil.cl
 ../generic/math/clc_fabs.cl
 ../generic/math/clc_floor.cl
+../generic/math/clc_mad.cl
 ../generic/math/clc_rint.cl
 ../generic/math/clc_trunc.cl
 ../generic/shared/clc_clamp.cl