Re: [PATCH v1] RISC-V: Support RVV FP16 MISC vget/vset intrinsic API

["juzhe.zhong@rivai.ai" <juzhe.zhong@rivai.ai>]

[-- Attachment #1: Type: text/plain, Size: 7653 bytes --]

LGTM



juzhe.zhong@rivai.ai
 
From: pan2.li
Date: 2023-06-12 15:40
To: gcc-patches
CC: juzhe.zhong; rdapp.gcc; jeffreyalaw; pan2.li; yanzhang.wang; kito.cheng
Subject: [PATCH v1] RISC-V: Support RVV FP16 MISC vget/vset intrinsic API
From: Pan Li <pan2.li@intel.com>
 
This patch support the intrinsic API of FP16 ZVFHMIN vget/vset. From
the user's perspective, it is reasonable to do some get/set operations
for the vfloat16*_t types when only ZVFHMIN is enabled.
 
Signed-off-by: Pan Li <pan2.li@intel.com>
 
gcc/ChangeLog:
 
* config/riscv/riscv-vector-builtins-types.def
(vfloat16m1_t): Add type to lmul1 ops.
(vfloat16m2_t): Likewise.
(vfloat16m4_t): Likewise.
 
gcc/testsuite/ChangeLog:
 
* gcc.target/riscv/rvv/base/zvfh-over-zvfhmin.c: Add new test cases.
* gcc.target/riscv/rvv/base/zvfhmin-intrinsic.c: Likewise.
---
.../riscv/riscv-vector-builtins-types.def     |  3 ++
.../riscv/rvv/base/zvfh-over-zvfhmin.c        | 15 +++++++--
.../riscv/rvv/base/zvfhmin-intrinsic.c        | 32 ++++++++++++++-----
3 files changed, 40 insertions(+), 10 deletions(-)
 
diff --git a/gcc/config/riscv/riscv-vector-builtins-types.def b/gcc/config/riscv/riscv-vector-builtins-types.def
index db8e61fea6a..4926bd8a2d2 100644
--- a/gcc/config/riscv/riscv-vector-builtins-types.def
+++ b/gcc/config/riscv/riscv-vector-builtins-types.def
@@ -1091,6 +1091,7 @@ DEF_RVV_LMUL1_OPS (vuint8m1_t, 0)
DEF_RVV_LMUL1_OPS (vuint16m1_t, 0)
DEF_RVV_LMUL1_OPS (vuint32m1_t, 0)
DEF_RVV_LMUL1_OPS (vuint64m1_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_LMUL1_OPS (vfloat16m1_t, RVV_REQUIRE_ELEN_FP_16)
DEF_RVV_LMUL1_OPS (vfloat32m1_t, RVV_REQUIRE_ELEN_FP_32)
DEF_RVV_LMUL1_OPS (vfloat64m1_t, RVV_REQUIRE_ELEN_FP_64)
@@ -1102,6 +1103,7 @@ DEF_RVV_LMUL2_OPS (vuint8m2_t, 0)
DEF_RVV_LMUL2_OPS (vuint16m2_t, 0)
DEF_RVV_LMUL2_OPS (vuint32m2_t, 0)
DEF_RVV_LMUL2_OPS (vuint64m2_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_LMUL2_OPS (vfloat16m2_t, RVV_REQUIRE_ELEN_FP_16)
DEF_RVV_LMUL2_OPS (vfloat32m2_t, RVV_REQUIRE_ELEN_FP_32)
DEF_RVV_LMUL2_OPS (vfloat64m2_t, RVV_REQUIRE_ELEN_FP_64)
@@ -1113,6 +1115,7 @@ DEF_RVV_LMUL4_OPS (vuint8m4_t, 0)
DEF_RVV_LMUL4_OPS (vuint16m4_t, 0)
DEF_RVV_LMUL4_OPS (vuint32m4_t, 0)
DEF_RVV_LMUL4_OPS (vuint64m4_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_LMUL4_OPS (vfloat16m4_t, RVV_REQUIRE_ELEN_FP_16)
DEF_RVV_LMUL4_OPS (vfloat32m4_t, RVV_REQUIRE_ELEN_FP_32)
DEF_RVV_LMUL4_OPS (vfloat64m4_t, RVV_REQUIRE_ELEN_FP_64)
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/base/zvfh-over-zvfhmin.c b/gcc/testsuite/gcc.target/riscv/rvv/base/zvfh-over-zvfhmin.c
index c3ed4191a36..1d82cc8de2d 100644
--- a/gcc/testsuite/gcc.target/riscv/rvv/base/zvfh-over-zvfhmin.c
+++ b/gcc/testsuite/gcc.target/riscv/rvv/base/zvfh-over-zvfhmin.c
@@ -61,6 +61,14 @@ vfloat16m8_t test_vundefined_f16m8() {
   return __riscv_vundefined_f16m8();
}
+vfloat16m2_t test_vset_v_f16m1_f16m2(vfloat16m2_t dest, size_t index, vfloat16m1_t val) {
+  return __riscv_vset_v_f16m1_f16m2(dest, 0, val);
+}
+
+vfloat16m4_t test_vget_v_f16m8_f16m4(vfloat16m8_t src, size_t index) {
+  return __riscv_vget_v_f16m8_f16m4(src, 0);
+}
+
/* { dg-final { scan-assembler-times {vsetvli\s+zero,\s*[a-x0-9]+,\s*e16,\s*mf4,\s*t[au],\s*m[au]} 3 } } */
/* { dg-final { scan-assembler-times {vsetvli\s+zero,\s*[a-x0-9]+,\s*e16,\s*m4,\s*t[au],\s*m[au]} 2 } } */
/* { dg-final { scan-assembler-times {vsetvli\s+zero,\s*[a-x0-9]+,\s*e16,\s*m8,\s*t[au],\s*m[au]} 1 } } */
@@ -71,7 +79,10 @@ vfloat16m8_t test_vundefined_f16m8() {
/* { dg-final { scan-assembler-times {vfncvt\.f\.f\.w\s+v[0-9]+,\s*v[0-9]+} 2 } } */
/* { dg-final { scan-assembler-times {vle16\.v\s+v[0-9]+,\s*0\([0-9ax]+\)} 7 } } */
/* { dg-final { scan-assembler-times {vse16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 6 } } */
-/* { dg-final { scan-assembler-times {vl4re16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 1 } } */
+/* { dg-final { scan-assembler-times {vl1re16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 1 } } */
+/* { dg-final { scan-assembler-times {vl2re16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 1 } } */
+/* { dg-final { scan-assembler-times {vl4re16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 3 } } */
/* { dg-final { scan-assembler-times {vl8re16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 1 } } */
-/* { dg-final { scan-assembler-times {vs4r\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 1 } } */
+/* { dg-final { scan-assembler-times {vs2r\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 1 } } */
+/* { dg-final { scan-assembler-times {vs4r\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 3 } } */
/* { dg-final { scan-assembler-times {vs8r\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 5 } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/base/zvfhmin-intrinsic.c b/gcc/testsuite/gcc.target/riscv/rvv/base/zvfhmin-intrinsic.c
index 8d39a2ed4c2..1026b3f82f1 100644
--- a/gcc/testsuite/gcc.target/riscv/rvv/base/zvfhmin-intrinsic.c
+++ b/gcc/testsuite/gcc.target/riscv/rvv/base/zvfhmin-intrinsic.c
@@ -165,6 +165,22 @@ vfloat16m8_t test_vundefined_f16m8() {
   return __riscv_vundefined_f16m8();
}
+vfloat16m2_t test_vset_v_f16m1_f16m2(vfloat16m2_t dest, size_t index, vfloat16m1_t val) {
+  return __riscv_vset_v_f16m1_f16m2(dest, 0, val);
+}
+
+vfloat16m8_t test_vset_v_f16m4_f16m8(vfloat16m8_t dest, size_t index, vfloat16m4_t val) {
+  return __riscv_vset_v_f16m4_f16m8(dest, 0, val);
+}
+
+vfloat16m1_t test_vget_v_f16m2_f16m1(vfloat16m2_t src, size_t index) {
+  return __riscv_vget_v_f16m2_f16m1(src, 0);
+}
+
+vfloat16m4_t test_vget_v_f16m8_f16m4(vfloat16m8_t src, size_t index) {
+  return __riscv_vget_v_f16m8_f16m4(src, 0);
+}
+
/* { dg-final { scan-assembler-times {vsetvli\s+zero,\s*[a-x0-9]+,\s*e16,\s*mf4,\s*t[au],\s*m[au]} 3 } } */
/* { dg-final { scan-assembler-times {vsetvli\s+zero,\s*[a-x0-9]+,\s*e16,\s*mf2,\s*t[au],\s*m[au]} 2 } } */
/* { dg-final { scan-assembler-times {vsetvli\s+zero,\s*[a-x0-9]+,\s*e16,\s*m1,\s*t[au],\s*m[au]} 2 } } */
@@ -180,11 +196,11 @@ vfloat16m8_t test_vundefined_f16m8() {
/* { dg-final { scan-assembler-times {vfncvt\.f\.f\.w\s+v[0-9]+,\s*v[0-9]+} 5 } } */
/* { dg-final { scan-assembler-times {vle16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 20 } } */
/* { dg-final { scan-assembler-times {vse16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 15 } } */
-/* { dg-final { scan-assembler-times {vl1re16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 3 } } */
-/* { dg-final { scan-assembler-times {vl2re16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 3 } } */
-/* { dg-final { scan-assembler-times {vl8re16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 4 } } */
-/* { dg-final { scan-assembler-times {vl4re16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 3 } } */
-/* { dg-final { scan-assembler-times {vs1r\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 4 } } */
-/* { dg-final { scan-assembler-times {vs2r\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 4 } } */
-/* { dg-final { scan-assembler-times {vs4r\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 4 } } */
-/* { dg-final { scan-assembler-times {vs8r\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 12 } } */
+/* { dg-final { scan-assembler-times {vl1re16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 5 } } */
+/* { dg-final { scan-assembler-times {vl2re16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 4 } } */
+/* { dg-final { scan-assembler-times {vl8re16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 5 } } */
+/* { dg-final { scan-assembler-times {vl4re16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 5 } } */
+/* { dg-final { scan-assembler-times {vs1r\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 5 } } */
+/* { dg-final { scan-assembler-times {vs2r\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 5 } } */
+/* { dg-final { scan-assembler-times {vs4r\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 5 } } */
+/* { dg-final { scan-assembler-times {vs8r\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 13 } } */
-- 
2.34.1