[OpenCL] Add cl_khr_integer_dot_product

[svenvh]

Summary

Add the builtins defined by Section 42 "Integer dot product" in the OpenCL Extension Specification.

See https://www.khronos.org/registry/OpenCL/specs/3.0-unified/html/OpenCL_Ext.html#cl_khr_integer_dot_product

[Anastasia]

LGTM! Thanks

[stuart]

LGTM, too.

clang/lib/Headers/opencl-c-base.h

 #define cl_khr_subgroup_extended_types 1
 #define cl_khr_subgroup_non_uniform_vote 1
 #define cl_khr_subgroup_ballot 1
 #define cl_khr_subgroup_non_uniform_arithmetic 1
 #define cl_khr_subgroup_shuffle 1
 #define cl_khr_subgroup_shuffle_relative 1
 #define cl_khr_subgroup_clustered_reduce 1
 #define cl_khr_extended_bit_ops 1
+#define cl_khr_integer_dot_product 1
+#define __opencl_c_integer_dot_product_input_4x8bit 1
+#define __opencl_c_integer_dot_product_input_4x8bit_packed 1
+
 #endif // defined(__SPIR__)
 #endif // (defined(__OPENCL_CPP_VERSION__) || __OPENCL_C_VERSION__ >= 200)
 
 // Define feature macros for OpenCL C 2.0
 #if (defined(__OPENCL_CPP_VERSION__) || __OPENCL_C_VERSION__ == 200)
 #define __opencl_c_pipes 1
 #define __opencl_c_generic_address_space 1
 #define __opencl_c_work_group_collective_functions 1

 
 #define kernel_exec(X, typen) __kernel \
   __attribute__((work_group_size_hint(X, 1, 1))) \
   __attribute__((vec_type_hint(typen)))
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_1_2)
 // OpenCL v1.2 s6.12.13, v2.0 s6.13.13 - printf
 
 int printf(__constant const char* st, ...) __attribute__((format(printf, 1, 2)));
 #endif
 
 #ifdef cl_intel_device_side_avc_motion_estimation
 
 #define CLK_AVC_ME_MAJOR_16x16_INTEL 0x0
 #define CLK_AVC_ME_MAJOR_16x8_INTEL 0x1
 #define CLK_AVC_ME_MAJOR_8x16_INTEL 0x2
 #define CLK_AVC_ME_MAJOR_8x8_INTEL 0x3

clang/lib/Headers/opencl-c.h

 half __ovld fract(half x, half *iptr);
 half2 __ovld fract(half2 x, half2 *iptr);
 half3 __ovld fract(half3 x, half3 *iptr);
 half4 __ovld fract(half4 x, half4 *iptr);
 half8 __ovld fract(half8 x, half8 *iptr);
 half16 __ovld fract(half16 x, half16 *iptr);
 #endif //cl_khr_fp16
 #else
 float __ovld fract(float x, __global float *iptr);
 float2 __ovld fract(float2 x, __global float2 *iptr);
 float3 __ovld fract(float3 x, __global float3 *iptr);
 float4 __ovld fract(float4 x, __global float4 *iptr);
 float8 __ovld fract(float8 x, __global float8 *iptr);
 float16 __ovld fract(float16 x, __global float16 *iptr);
 float __ovld fract(float x, __local float *iptr);
 float2 __ovld fract(float2 x, __local float2 *iptr);
 float3 __ovld fract(float3 x, __local float3 *iptr);
 float4 __ovld fract(float4 x, __local float4 *iptr);
 float8 __ovld fract(float8 x, __local float8 *iptr);
 float16 __ovld fract(float16 x, __local float16 *iptr);
 float __ovld fract(float x, __private float *iptr);
 float2 __ovld fract(float2 x, __private float2 *iptr);
 float3 __ovld fract(float3 x, __private float3 *iptr);
 float4 __ovld fract(float4 x, __private float4 *iptr);
 float8 __ovld fract(float8 x, __private float8 *iptr);
 float16 __ovld fract(float16 x, __private float16 *iptr);

 short16 __ovld __cnfn bit_reverse(short16);
 ushort16 __ovld __cnfn bit_reverse(ushort16);
 int16 __ovld __cnfn bit_reverse(int16);
 uint16 __ovld __cnfn bit_reverse(uint16);
 long16 __ovld __cnfn bit_reverse(long16);
 ulong16 __ovld __cnfn bit_reverse(ulong16);
 #endif // cl_khr_extended_bit_ops
 
+#if defined(__opencl_c_integer_dot_product_input_4x8bit)
+uint __ovld __cnfn dot(uchar4, uchar4);
+int __ovld __cnfn dot(char4, char4);
+int __ovld __cnfn dot(uchar4, char4);
+int __ovld __cnfn dot(char4, uchar4);
+
+uint __ovld __cnfn dot_acc_sat(uchar4, uchar4, uint);
+int __ovld __cnfn dot_acc_sat(char4, char4, int);
+int __ovld __cnfn dot_acc_sat(uchar4, char4, int);
+int __ovld __cnfn dot_acc_sat(char4, uchar4, int);
+#endif // __opencl_c_integer_dot_product_input_4x8bit
+
+#if defined(__opencl_c_integer_dot_product_input_4x8bit_packed)
+uint __ovld __cnfn dot_4x8packed_uu_uint(uint, uint);
+int __ovld __cnfn dot_4x8packed_ss_int(uint, uint);
+int __ovld __cnfn dot_4x8packed_us_int(uint, uint);
+int __ovld __cnfn dot_4x8packed_su_int(uint, uint);
+
+uint __ovld __cnfn dot_acc_sat_4x8packed_uu_uint(uint, uint, uint);
+int __ovld __cnfn dot_acc_sat_4x8packed_ss_int(uint, uint, int);
+int __ovld __cnfn dot_acc_sat_4x8packed_us_int(uint, uint, int);
+int __ovld __cnfn dot_acc_sat_4x8packed_su_int(uint, uint, int);
+#endif // __opencl_c_integer_dot_product_input_4x8bit_packed
+
 #if defined(cl_intel_subgroups)
 // Intel-Specific Sub Group Functions
 float   __ovld __conv intel_sub_group_shuffle( float  x, uint c );
 float2  __ovld __conv intel_sub_group_shuffle( float2 x, uint c );
 float3  __ovld __conv intel_sub_group_shuffle( float3 x, uint c );
 float4  __ovld __conv intel_sub_group_shuffle( float4 x, uint c );
 float8  __ovld __conv intel_sub_group_shuffle( float8 x, uint c );
 float16 __ovld __conv intel_sub_group_shuffle( float16 x, uint c );

clang/lib/Sema/OpenCLBuiltins.td

let Extension = FuncExtKhrExtendedBitOps in {
   def : Builtin<"bitfield_insert", [AIGenTypeN, AIGenTypeN, AIGenTypeN, UInt, UInt], Attr.Const>;
   def : Builtin<"bitfield_extract_signed", [SGenTypeN, SGenTypeN, UInt, UInt], Attr.Const>;
   def : Builtin<"bitfield_extract_signed", [SGenTypeN, UGenTypeN, UInt, UInt], Attr.Const>;
   def : Builtin<"bitfield_extract_unsigned", [UGenTypeN, SGenTypeN, UInt, UInt], Attr.Const>;
   def : Builtin<"bitfield_extract_unsigned", [UGenTypeN, UGenTypeN, UInt, UInt], Attr.Const>;
   def : Builtin<"bit_reverse", [AIGenTypeN, AIGenTypeN], Attr.Const>;
 }
 
+// Section 42.3 - cl_khr_integer_dot_product
+let Extension = FunctionExtension<"__opencl_c_integer_dot_product_input_4x8bit"> in {
+  def : Builtin<"dot", [UInt, VectorType<UChar, 4>, VectorType<UChar, 4>], Attr.Const>;
+  def : Builtin<"dot", [Int, VectorType<Char, 4>, VectorType<Char, 4>], Attr.Const>;
+  def : Builtin<"dot", [Int, VectorType<UChar, 4>, VectorType<Char, 4>], Attr.Const>;
+  def : Builtin<"dot", [Int, VectorType<Char, 4>, VectorType<UChar, 4>], Attr.Const>;
+
+  def : Builtin<"dot_acc_sat", [UInt, VectorType<UChar, 4>, VectorType<UChar, 4>, UInt], Attr.Const>;
+  def : Builtin<"dot_acc_sat", [Int, VectorType<Char, 4>, VectorType<Char, 4>, Int], Attr.Const>;
+  def : Builtin<"dot_acc_sat", [Int, VectorType<UChar, 4>, VectorType<Char, 4>, Int], Attr.Const>;
+  def : Builtin<"dot_acc_sat", [Int, VectorType<Char, 4>, VectorType<UChar, 4>, Int], Attr.Const>;
+}
+
+let Extension = FunctionExtension<"__opencl_c_integer_dot_product_input_4x8bit_packed"> in {
+  def : Builtin<"dot_4x8packed_uu_uint", [UInt, UInt, UInt], Attr.Const>;
+  def : Builtin<"dot_4x8packed_ss_int", [Int, UInt, UInt], Attr.Const>;
+  def : Builtin<"dot_4x8packed_us_int", [Int, UInt, UInt], Attr.Const>;
+  def : Builtin<"dot_4x8packed_su_int", [Int, UInt, UInt], Attr.Const>;
+
+  def : Builtin<"dot_acc_sat_4x8packed_uu_uint", [UInt, UInt, UInt, UInt], Attr.Const>;
+  def : Builtin<"dot_acc_sat_4x8packed_ss_int", [Int, UInt, UInt, Int], Attr.Const>;
+  def : Builtin<"dot_acc_sat_4x8packed_us_int", [Int, UInt, UInt, Int], Attr.Const>;
+  def : Builtin<"dot_acc_sat_4x8packed_su_int", [Int, UInt, UInt, Int], Attr.Const>;
+}
+
 //--------------------------------------------------------------------
 // Arm extensions.
 let Extension = ArmIntegerDotProductInt8 in {
   foreach name = ["arm_dot"] in {
     def : Builtin<name, [UInt, VectorType<UChar, 4>, VectorType<UChar, 4>]>;
     def : Builtin<name, [Int, VectorType<Char, 4>, VectorType<Char, 4>]>;
   }
 }

clang/test/Headers/opencl-c-header.cl

 #error "Incorrectly defined cl_khr_subgroup_shuffle_relative"
 #endif
 #if cl_khr_subgroup_clustered_reduce != 1
 #error "Incorrectly defined cl_khr_subgroup_clustered_reduce"
 #endif
 #if cl_khr_extended_bit_ops != 1
 #error "Incorrectly defined cl_khr_extended_bit_ops"
 #endif
+#if cl_khr_integer_dot_product != 1
+#error "Incorrectly defined cl_khr_integer_dot_product"
+#endif
+#if __opencl_c_integer_dot_product_input_4x8bit != 1
+#error "Incorrectly defined __opencl_c_integer_dot_product_input_4x8bit"
+#endif
+#if __opencl_c_integer_dot_product_input_4x8bit_packed != 1
+#error "Incorrectly defined __opencl_c_integer_dot_product_input_4x8bit_packed"
+#endif
 
 #else
 
 #ifdef cl_khr_subgroup_extended_types
 #error "Incorrect cl_khr_subgroup_extended_types define"
 #endif
 #ifdef cl_khr_subgroup_non_uniform_vote
 #error "Incorrect cl_khr_subgroup_non_uniform_vote define"

 #error "Incorrect cl_khr_subgroup_shuffle_relative define"
 #endif
 #ifdef cl_khr_subgroup_clustered_reduce
 #error "Incorrect cl_khr_subgroup_clustered_reduce define"
 #endif
 #ifdef cl_khr_extended_bit_ops
 #error "Incorrect cl_khr_extended_bit_ops define"
 #endif
+#ifdef cl_khr_integer_dot_product
+#error "Incorrect cl_khr_integer_dot_product define"
+#endif
+#ifdef __opencl_c_integer_dot_product_input_4x8bit
+#error "Incorrect __opencl_c_integer_dot_product_input_4x8bit define"
+#endif
+#ifdef __opencl_c_integer_dot_product_input_4x8bit_packed
+#error "Incorrect __opencl_c_integer_dot_product_input_4x8bit_packed define"
+#endif
 
 #endif //(defined(__OPENCL_CPP_VERSION__) || __OPENCL_C_VERSION__ >= 200)
 
 // OpenCL C features.
 #if (__OPENCL_C_VERSION__ == 300)
 
 #if __opencl_c_atomic_scope_all_devices != 1
 #error "Incorrectly defined feature macro __opencl_c_atomic_scope_all_devices"