RISC-V: Support RVV VFREDOSUM.VS rounding mode intrinsic API

This patch would like to support the rounding mode API for the
VFREDOSUM.VS as the below samples.

* __riscv_vfredosum_vs_f32m1_f32m1_rm
* __riscv_vfredosum_vs_f32m1_f32m1_rm_m

Signed-off-by: Pan Li <pan2.li@intel.com>
gcc/ChangeLog:

	* config/riscv/riscv-vector-builtins-bases.cc
	(vfredosum_frm_obj): New declaration.
	(BASE): Ditto.
	* config/riscv/riscv-vector-builtins-bases.h: Ditto.
	* config/riscv/riscv-vector-builtins-functions.def
	(vfredosum_frm): New intrinsic function def.

gcc/testsuite/ChangeLog:

	* gcc.target/riscv/rvv/base/float-point-redosum.c: New test.

diff --git a/gcc/config/riscv/riscv-vector-builtins-bases.cc b/gcc/config/riscv/riscv-vector-builtins-bases.cc
index 65f1d9c8ff7d0945829eb8e745b8bb98bcc5b221..ef2991359dad6c6be2791b4974c566b3095a2af5 100644 (file)
--- a/gcc/config/riscv/riscv-vector-builtins-bases.cc
+++ b/gcc/config/riscv/riscv-vector-builtins-bases.cc
@@ -2539,6 +2539,7 @@ static CONSTEXPR const widen_reducop<UNSPEC_WREDUC_USUM> vwredsumu_obj;
 static CONSTEXPR const freducop<UNSPEC_UNORDERED> vfredusum_obj;
 static CONSTEXPR const freducop<UNSPEC_UNORDERED, HAS_FRM> vfredusum_frm_obj;
 static CONSTEXPR const freducop<UNSPEC_ORDERED> vfredosum_obj;
+static CONSTEXPR const freducop<UNSPEC_ORDERED, HAS_FRM> vfredosum_frm_obj;
 static CONSTEXPR const reducop<SMAX> vfredmax_obj;
 static CONSTEXPR const reducop<SMIN> vfredmin_obj;
 static CONSTEXPR const widen_freducop<UNSPEC_UNORDERED> vfwredusum_obj;
@@ -2797,6 +2798,7 @@ BASE (vwredsumu)
 BASE (vfredusum)
 BASE (vfredusum_frm)
 BASE (vfredosum)
+BASE (vfredosum_frm)
 BASE (vfredmax)
 BASE (vfredmin)
 BASE (vfwredosum)

diff --git a/gcc/config/riscv/riscv-vector-builtins-bases.h b/gcc/config/riscv/riscv-vector-builtins-bases.h
index fd1a84f3e68de99a11992e93b6228b281f8add7c..da8412b66df972b38fe345d1a1a9e4206f167a46 100644 (file)
--- a/gcc/config/riscv/riscv-vector-builtins-bases.h
+++ b/gcc/config/riscv/riscv-vector-builtins-bases.h
@@ -241,6 +241,7 @@ extern const function_base *const vwredsumu;
 extern const function_base *const vfredusum;
 extern const function_base *const vfredusum_frm;
 extern const function_base *const vfredosum;
+extern const function_base *const vfredosum_frm;
 extern const function_base *const vfredmax;
 extern const function_base *const vfredmin;
 extern const function_base *const vfwredosum;

diff --git a/gcc/config/riscv/riscv-vector-builtins-functions.def b/gcc/config/riscv/riscv-vector-builtins-functions.def
index 90a83c02d52f7d91341a71746c3fb2cd6d255a03..80e65bfb14bd54d692c3605eef0adb329e6767d9 100644 (file)
--- a/gcc/config/riscv/riscv-vector-builtins-functions.def
+++ b/gcc/config/riscv/riscv-vector-builtins-functions.def
@@ -501,6 +501,7 @@ DEF_RVV_FUNCTION (vfredmax, reduc_alu, no_mu_preds, f_vs_ops)
 DEF_RVV_FUNCTION (vfredmin, reduc_alu, no_mu_preds, f_vs_ops)
 
 DEF_RVV_FUNCTION (vfredusum_frm, reduc_alu_frm, no_mu_preds, f_vs_ops)
+DEF_RVV_FUNCTION (vfredosum_frm, reduc_alu_frm, no_mu_preds, f_vs_ops)
 
 // 14.4. Vector Widening Floating-Point Reduction Instructions
 DEF_RVV_FUNCTION (vfwredosum, reduc_alu, no_mu_preds, wf_vs_ops)

diff --git a/gcc/testsuite/gcc.target/riscv/rvv/base/float-point-redosum.c b/gcc/testsuite/gcc.target/riscv/rvv/base/float-point-redosum.c
new file mode 100644 (file)
index 0000000..2e6a3c2
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/base/float-point-redosum.c
@@ -0,0 +1,33 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gcv -mabi=lp64 -O3 -Wno-psabi" } */
+
+#include "riscv_vector.h"
+
+vfloat32m1_t
+test_riscv_vfredosum_vs_f32m1_f32m1_rm (vfloat32m1_t op1, vfloat32m1_t op2,
+                                       size_t vl) {
+  return __riscv_vfredosum_vs_f32m1_f32m1_rm (op1, op2, 0, vl);
+}
+
+vfloat32m1_t
+test_vfredosum_vs_f32m1_f32m1_rm_m (vbool32_t mask, vfloat32m1_t op1,
+                                   vfloat32m1_t op2, size_t vl) {
+  return __riscv_vfredosum_vs_f32m1_f32m1_rm_m (mask, op1, op2, 1, vl);
+}
+
+vfloat32m1_t
+test_riscv_vfredosum_vs_f32m1_f32m1 (vfloat32m1_t op1, vfloat32m1_t op2,
+                                    size_t vl) {
+  return __riscv_vfredosum_vs_f32m1_f32m1 (op1, op2, vl);
+}
+
+vfloat32m1_t
+test_vfredosum_vs_f32m1_f32m1_m (vbool32_t mask, vfloat32m1_t op1,
+                                vfloat32m1_t op2, size_t vl) {
+  return __riscv_vfredosum_vs_f32m1_f32m1_m (mask, op1, op2, vl);
+}
+
+/* { dg-final { scan-assembler-times {vfredosum\.vs\s+v[0-9]+,\s*v[0-9]+,\s*v[0-9]+} 4 } } */
+/* { dg-final { scan-assembler-times {frrm\s+[axs][0-9]+} 2 } } */
+/* { dg-final { scan-assembler-times {fsrm\s+[axs][0-9]+} 2 } } */
+/* { dg-final { scan-assembler-times {fsrmi\s+[01234]} 2 } } */