aarch64: ACLE I8MM multiply-accumulate intrinsics

This patch adds intrinsics for 8-bit integer matrix multiply-accumulate
operations including vmmlaq_s32, vmmlaq_u32, and vusmmlaq_s32.

gcc/ChangeLog:

2020-02-07  Dennis Zhang  <dennis.zhang@arm.com>

	* config/aarch64/aarch64-simd-builtins.def (simd_smmla): New entry.
	(simd_ummla, simd_usmmla): Likewise.
	* config/aarch64/aarch64-simd.md (aarch64_simd_<sur>mmlav16qi): New.
	* config/aarch64/arm_neon.h (vmmlaq_s32, vmmlaq_u32): New.
	(vusmmlaq_s32): New.

gcc/testsuite/ChangeLog:

2020-02-07  Dennis Zhang  <dennis.zhang@arm.com>

	* gcc.target/aarch64/simd/vmmla.c: New test.

6 files changed, 79 insertions, 0 deletions

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index d5d29f5..d255bd9 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,11 @@
+2020-02-07  Dennis Zhang  <dennis.zhang@arm.com>
+
+	* config/aarch64/aarch64-simd-builtins.def (simd_smmla): New entry.
+	(simd_ummla, simd_usmmla): Likewise.
+	* config/aarch64/aarch64-simd.md (aarch64_simd_<sur>mmlav16qi): New.
+	* config/aarch64/arm_neon.h (vmmlaq_s32, vmmlaq_u32): New.
+	(vusmmlaq_s32): New.
+
 2020-02-07  Richard Biener  <rguenther@suse.de>
 
 	PR middle-end/93519
diff --git a/gcc/config/aarch64/aarch64-simd-builtins.def b/gcc/config/aarch64/aarch64-simd-builtins.def
index 02b2154..fe3c7f1 100644
--- a/gcc/config/aarch64/aarch64-simd-builtins.def
+++ b/gcc/config/aarch64/aarch64-simd-builtins.def
@@ -703,3 +703,8 @@
   VAR1 (QUADOP_LANE, bfmlalt_lane, 0, v4sf)
   VAR1 (QUADOP_LANE, bfmlalb_lane_q, 0, v4sf)
   VAR1 (QUADOP_LANE, bfmlalt_lane_q, 0, v4sf)
+
+  /* Implemented by aarch64_simd_<sur>mmlav16qi.  */
+  VAR1 (TERNOP, simd_smmla, 0, v16qi)
+  VAR1 (TERNOPU, simd_ummla, 0, v16qi)
+  VAR1 (TERNOP_SSUS, simd_usmmla, 0, v16qi)
diff --git a/gcc/config/aarch64/aarch64-simd.md b/gcc/config/aarch64/aarch64-simd.md
index f2b440c..c8e1012 100644
--- a/gcc/config/aarch64/aarch64-simd.md
+++ b/gcc/config/aarch64/aarch64-simd.md
@@ -7173,3 +7173,15 @@
 }
   [(set_attr "type" "neon_fp_mla_s_scalar_q")]
 )
+
+;; 8-bit integer matrix multiply-accumulate
+(define_insn "aarch64_simd_<sur>mmlav16qi"
+  [(set (match_operand:V4SI 0 "register_operand" "=w")
+	(plus:V4SI
+	 (unspec:V4SI [(match_operand:V16QI 2 "register_operand" "w")
+		       (match_operand:V16QI 3 "register_operand" "w")] MATMUL)
+	 (match_operand:V4SI 1 "register_operand" "0")))]
+  "TARGET_I8MM"
+  "<sur>mmla\\t%0.4s, %2.16b, %3.16b"
+  [(set_attr "type" "neon_mla_s_q")]
+)
diff --git a/gcc/config/aarch64/arm_neon.h b/gcc/config/aarch64/arm_neon.h
index db845a3..a6bcdf1 100644
--- a/gcc/config/aarch64/arm_neon.h
+++ b/gcc/config/aarch64/arm_neon.h
@@ -34797,6 +34797,29 @@ vsudotq_laneq_s32 (int32x4_t __r, int8x16_t __a, uint8x16_t __b,
   return __builtin_aarch64_sudot_laneqv16qi_sssus (__r, __a, __b, __index);
 }
 
+/* Matrix Multiply-Accumulate.  */
+
+__extension__ extern __inline int32x4_t
+__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
+vmmlaq_s32 (int32x4_t __r, int8x16_t __a, int8x16_t __b)
+{
+  return __builtin_aarch64_simd_smmlav16qi (__r, __a, __b);
+}
+
+__extension__ extern __inline uint32x4_t
+__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
+vmmlaq_u32 (uint32x4_t __r, uint8x16_t __a, uint8x16_t __b)
+{
+  return __builtin_aarch64_simd_ummlav16qi_uuuu (__r, __a, __b);
+}
+
+__extension__ extern __inline int32x4_t
+__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
+vusmmlaq_s32 (int32x4_t __r, uint8x16_t __a, int8x16_t __b)
+{
+  return __builtin_aarch64_simd_usmmlav16qi_ssus (__r, __a, __b);
+}
+
 #pragma GCC pop_options
 
 #undef __aarch64_vget_lane_any
diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog
index 69f7223..7ec36cc 100644
--- a/gcc/testsuite/ChangeLog
+++ b/gcc/testsuite/ChangeLog
@@ -1,3 +1,7 @@
+2020-02-07  Dennis Zhang  <dennis.zhang@arm.com>
+
+	* gcc.target/aarch64/simd/vmmla.c: New test.
+
 2020-02-07  Richard Biener  <rguenther@suse.de>
 
 	PR middle-end/93519
diff --git a/gcc/testsuite/gcc.target/aarch64/simd/vmmla.c b/gcc/testsuite/gcc.target/aarch64/simd/vmmla.c
new file mode 100644
index 0000000..5eec2b5
--- /dev/null
+++ b/gcc/testsuite/gcc.target/aarch64/simd/vmmla.c
@@ -0,0 +1,27 @@
+/* { dg-do assemble} */
+/* { dg-require-effective-target arm_v8_2a_i8mm_ok } */
+/* { dg-additional-options "-march=armv8.2-a+i8mm" } */
+
+#include "arm_neon.h"
+
+int32x4_t
+test_vmmlaq_s32 (int32x4_t r, int8x16_t a, int8x16_t b)
+{
+  return vmmlaq_s32 (r, a, b);
+}
+
+uint32x4_t
+test_vmmlaq_u32 (uint32x4_t r, uint8x16_t a, uint8x16_t b)
+{
+  return vmmlaq_u32 (r, a, b);
+}
+
+int32x4_t
+test_vusmmlaq_s32 (int32x4_t r, uint8x16_t a, int8x16_t b)
+{
+  return vusmmlaq_s32 (r, a, b);
+}
+
+/* { dg-final { scan-assembler-times {\tsmmla\tv[0-9]+.4s, v[0-9]+.16b, v[0-9]+.16b} 1 } } */
+/* { dg-final { scan-assembler-times {\tummla\tv[0-9]+.4s, v[0-9]+.16b, v[0-9]+.16b} 1 } } */
+/* { dg-final { scan-assembler-times {\tusmmla\tv[0-9]+.4s, v[0-9]+.16b, v[0-9]+.16b} 1 } } */