2018-01-11 Tamar Christina <tamar.christina@arm.com>
authorTamar Christina <tnfchris@gcc.gnu.org>
Thu, 11 Jan 2018 11:56:22 +0000 (11:56 +0000)
committerTamar Christina <tnfchris@gcc.gnu.org>
Thu, 11 Jan 2018 11:56:22 +0000 (11:56 +0000)
* config/aarch64/aarch64.h
(AARCH64_FL_FOR_ARCH8_4): Add  AARCH64_FL_DOTPROD.

gcc/testsuite/
2018-01-11  Tamar Christina  <tamar.christina@arm.com>

* gcc.target/aarch64/advsimd-intrinsics/vdot-compile-2.c: New.

From-SVN: r256527

gcc/ChangeLog
gcc/config/aarch64/aarch64.h
gcc/testsuite/ChangeLog
gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vdot-compile-2.c [new file with mode: 0644]

index d34418e82eeab2aa89f8dcc7be6abd9590e1534b..48e2def073f89f6d695e66fa8e1b40936e168ccb 100644 (file)
@@ -1,4 +1,9 @@
-2017-01-11  Sudakshina Das  <sudi.das@arm.com>
+2018-01-11  Tamar Christina  <tamar.christina@arm.com>
+
+       * config/aarch64/aarch64.h
+       (AARCH64_FL_FOR_ARCH8_4): Add  AARCH64_FL_DOTPROD.
+
+2018-01-11  Sudakshina Das  <sudi.das@arm.com>
 
        PR target/82096
        * expmed.c (emit_store_flag_force): Swap if const op0
index 82412e8676b88c34f6b93e597ba89b5e73ab2afd..ed7a8e5e7e8e55d24f237916963073dcd5799dfe 100644 (file)
@@ -173,7 +173,8 @@ extern unsigned aarch64_architecture_version;
 #define AARCH64_FL_FOR_ARCH8_3                 \
   (AARCH64_FL_FOR_ARCH8_2 | AARCH64_FL_V8_3)
 #define AARCH64_FL_FOR_ARCH8_4                 \
-  (AARCH64_FL_FOR_ARCH8_3 | AARCH64_FL_V8_4 | AARCH64_FL_F16FML)
+  (AARCH64_FL_FOR_ARCH8_3 | AARCH64_FL_V8_4 | AARCH64_FL_F16FML \
+   | AARCH64_FL_DOTPROD)
 
 /* Macros to test ISA flags.  */
 
index 3b49b86a30b28715097837321ea163483d752105..d54bc9f89ea5557c91ead86b2618d01dc55b6489 100644 (file)
@@ -1,4 +1,8 @@
-2017-01-11  Sudakshina Das  <sudi.das@arm.com>
+2018-01-11  Tamar Christina  <tamar.christina@arm.com>
+
+       * gcc.target/aarch64/advsimd-intrinsics/vdot-compile-2.c: New.
+
+2018-01-11  Sudakshina Das  <sudi.das@arm.com>
 
        PR target/82096
        * gcc.c-torture/compile/pr82096.c: New test.
diff --git a/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vdot-compile-2.c b/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vdot-compile-2.c
new file mode 100644 (file)
index 0000000..7d8d641
--- /dev/null
@@ -0,0 +1,73 @@
+/* { dg-do compile { target { aarch64*-*-* } } } */
+/* { dg-additional-options "-O3 -march=armv8.4-a" } */
+
+#include <arm_neon.h>
+
+/* Unsigned Dot Product instructions.  */
+
+uint32x2_t ufoo (uint32x2_t r, uint8x8_t x, uint8x8_t y)
+{
+  return vdot_u32 (r, x, y);
+}
+
+uint32x4_t ufooq (uint32x4_t r, uint8x16_t x, uint8x16_t y)
+{
+  return vdotq_u32 (r, x, y);
+}
+
+uint32x2_t ufoo_lane (uint32x2_t r, uint8x8_t x, uint8x8_t y)
+{
+  return vdot_lane_u32 (r, x, y, 0);
+}
+
+uint32x2_t ufoo_laneq (uint32x2_t r, uint8x8_t x, uint8x16_t y)
+{
+  return vdot_laneq_u32 (r, x, y, 0);
+}
+
+uint32x4_t ufooq_lane (uint32x4_t r, uint8x16_t x, uint8x8_t y)
+{
+  return vdotq_lane_u32 (r, x, y, 0);
+}
+
+uint32x4_t ufooq_laneq (uint32x4_t r, uint8x16_t x, uint8x16_t y)
+{
+  return vdotq_laneq_u32 (r, x, y, 0);
+}
+
+/* Signed Dot Product instructions.  */
+
+int32x2_t sfoo (int32x2_t r, int8x8_t x, int8x8_t y)
+{
+  return vdot_s32 (r, x, y);
+}
+
+int32x4_t sfooq (int32x4_t r, int8x16_t x, int8x16_t y)
+{
+  return vdotq_s32 (r, x, y);
+}
+
+int32x2_t sfoo_lane (int32x2_t r, int8x8_t x, int8x8_t y)
+{
+  return vdot_lane_s32 (r, x, y, 0);
+}
+
+int32x2_t sfoo_laneq (int32x2_t r, int8x8_t x, int8x16_t y)
+{
+  return vdot_laneq_s32 (r, x, y, 0);
+}
+
+int32x4_t sfooq_lane (int32x4_t r, int8x16_t x, int8x8_t y)
+{
+  return vdotq_lane_s32 (r, x, y, 0);
+}
+
+int32x4_t sfooq_laneq (int32x4_t r, int8x16_t x, int8x16_t y)
+{
+  return vdotq_laneq_s32 (r, x, y, 0);
+}
+
+/* { dg-final { scan-assembler-times {[us]dot\tv[0-9]+\.2s, v[0-9]+\.8b, v[0-9]+\.8b} 2 } } */
+/* { dg-final { scan-assembler-times {[us]dot\tv[0-9]+\.2s, v[0-9]+\.8b, v[0-9]+\.4b\[[0-9]+\]}  4 } } */
+/* { dg-final { scan-assembler-times {[us]dot\tv[0-9]+\.4s, v[0-9]+\.16b, v[0-9]+\.16b}  2 } } */
+/* { dg-final { scan-assembler-times {[us]dot\tv[0-9]+\.4s, v[0-9]+\.16b, v[0-9]+\.4b\[[0-9]+\]}  4 } } */