[OpenCL] Add sub-group builtin functions
Add the sub-group builtin functions from the OpenCL Extension
specification. This patch excludes the sub_group_barrier builtins
that take argument types not yet handled by the
`-fdeclare-opencl-builtins` machinery.
Co-authored-by: Pierre Gondois <pierre.gondois@arm.com>
diff --git a/clang/lib/Sema/OpenCLBuiltins.td b/clang/lib/Sema/OpenCLBuiltins.td
index 40798b7..745363a 100644
--- a/clang/lib/Sema/OpenCLBuiltins.td
+++ b/clang/lib/Sema/OpenCLBuiltins.td
@@ -1273,15 +1273,6 @@
// TODO: ndrange functions
-// OpenCL v2.0 s9.17.3: Additions to section 6.13.1: Work-Item Functions
-let MinVersion = CL20 in {
- let Extension = FuncExtKhrSubgroups in {
- def get_sub_group_size : Builtin<"get_sub_group_size", [UInt]>;
- def get_max_sub_group_size : Builtin<"get_max_sub_group_size", [UInt]>;
- def get_num_sub_groups : Builtin<"get_num_sub_groups", [UInt]>;
- }
-}
-
//--------------------------------------------------------------------
// End of the builtin functions defined in the OpenCL C specification.
// Builtin functions defined in the OpenCL C Extension are below.
@@ -1457,6 +1448,41 @@
}
//--------------------------------------------------------------------
+// OpenCL Extension v2.0 s28 - Subgroups
+// --- Table 28.2.1 ---
+let Extension = FuncExtKhrSubgroups in {
+ foreach name = ["get_sub_group_size", "get_max_sub_group_size",
+ "get_num_sub_groups", "get_sub_group_id",
+ "get_sub_group_local_id"] in {
+ def : Builtin<name, [UInt]>;
+ }
+ let MinVersion = CL20 in {
+ foreach name = ["get_enqueued_num_sub_groups"] in {
+ def : Builtin<name, [UInt]>;
+ }
+ }
+}
+
+// --- Table 28.2.2 ---
+// TODO: sub_group_barrier
+
+// --- Table 28.2.4 ---
+let Extension = FuncExtKhrSubgroups in {
+ foreach name = ["sub_group_all", "sub_group_any"] in {
+ def : Builtin<name, [Int, Int], Attr.Convergent>;
+ }
+ foreach name = ["sub_group_broadcast"] in {
+ def : Builtin<name, [IntLongFloatGenType1, IntLongFloatGenType1, UInt], Attr.Convergent>;
+ }
+ foreach name = ["sub_group_reduce_", "sub_group_scan_exclusive_",
+ "sub_group_scan_inclusive_"] in {
+ foreach op = ["add", "min", "max"] in {
+ def : Builtin<name # op, [IntLongFloatGenType1, IntLongFloatGenType1], Attr.Convergent>;
+ }
+ }
+}
+
+//--------------------------------------------------------------------
// Arm extensions.
let Extension = ArmIntegerDotProductInt8 in {
foreach name = ["arm_dot"] in {
diff --git a/clang/test/SemaOpenCL/fdeclare-opencl-builtins.cl b/clang/test/SemaOpenCL/fdeclare-opencl-builtins.cl
index e593b21..d1dcdfe 100644
--- a/clang/test/SemaOpenCL/fdeclare-opencl-builtins.cl
+++ b/clang/test/SemaOpenCL/fdeclare-opencl-builtins.cl
@@ -140,14 +140,11 @@
kernel void basic_subgroup(global uint *out) {
out[0] = get_sub_group_size();
-#if !defined(__OPENCL_CPP_VERSION__) && __OPENCL_C_VERSION__ < CL_VERSION_2_0
-// expected-error@-2{{implicit declaration of function 'get_sub_group_size' is invalid in OpenCL}}
-// expected-error@-3{{implicit conversion changes signedness: 'int' to 'uint' (aka 'unsigned int')}}
-#elif defined(__OPENCL_CPP_VERSION__)
-// expected-error@-5{{no matching function for call to 'get_sub_group_size'}}
-// expected-note@-6{{candidate unavailable as it requires OpenCL extension 'cl_khr_subgroups' to be enabled}}
+#if defined(__OPENCL_CPP_VERSION__)
+ // expected-error@-2{{no matching function for call to 'get_sub_group_size'}}
+ // expected-note@-3{{candidate unavailable as it requires OpenCL extension 'cl_khr_subgroups' to be enabled}}
#else
-// expected-error@-8{{use of declaration 'get_sub_group_size' requires cl_khr_subgroups extension to be enabled}}
+ // expected-error@-5{{use of declaration 'get_sub_group_size' requires cl_khr_subgroups extension to be enabled}}
#endif
}