ARM64: add patterns for scalar sqdmlal & sqdmlsl.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@205207 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/ARM64/ARM64InstrInfo.td b/lib/Target/ARM64/ARM64InstrInfo.td
index 8416fad9344ff68a4239938f10a2b6594aeaaaa7..da4d7261eab5374f8137900ce0262c22e0a576ec 100644 (file)
--- a/lib/Target/ARM64/ARM64InstrInfo.td
+++ b/lib/Target/ARM64/ARM64InstrInfo.td
@@ -2525,6 +2525,15 @@ defm SQDMULL  : SIMDThreeScalarMixedHS<0, 0b11010, "sqdmull",
 defm SQDMLAL  : SIMDThreeScalarMixedTiedHS<0, 0b10010, "sqdmlal">;
 defm SQDMLSL  : SIMDThreeScalarMixedTiedHS<0, 0b10110, "sqdmlsl">;
 
+def : Pat<(i64 (int_arm64_neon_sqadd (i64 FPR64:$Rd),
+                   (i64 (int_arm64_neon_sqdmulls_scalar (i32 FPR32:$Rn),
+                                                        (i32 FPR32:$Rm))))),
+          (SQDMLALi32 FPR64:$Rd, FPR32:$Rn, FPR32:$Rm)>;
+def : Pat<(i64 (int_arm64_neon_sqsub (i64 FPR64:$Rd),
+                   (i64 (int_arm64_neon_sqdmulls_scalar (i32 FPR32:$Rn),
+                                                        (i32 FPR32:$Rm))))),
+          (SQDMLSLi32 FPR64:$Rd, FPR32:$Rn, FPR32:$Rm)>;
+
 //===----------------------------------------------------------------------===//
 // Advanced SIMD two scalar instructions.
 //===----------------------------------------------------------------------===//

diff --git a/test/CodeGen/ARM64/vmul.ll b/test/CodeGen/ARM64/vmul.ll
index aeaea98f936abd0818eb3ebe87b7749c60a72692..433c09d37f811a90e877a72841da78308aa07dd7 100644 (file)
--- a/test/CodeGen/ARM64/vmul.ll
+++ b/test/CodeGen/ARM64/vmul.ll
@@ -1967,3 +1967,19 @@ define <1 x double> @test_fdiv_v1f64(<1 x double> %L, <1 x double> %R) nounwind
   %prod = fdiv <1 x double> %L, %R
   ret <1 x double> %prod
 }
+
+define i64 @sqdmlal_d(i32 %A, i32 %B, i64 %C) nounwind {
+;CHECK-LABEL: sqdmlal_d:
+;CHECK: sqdmlal
+  %tmp4 = call i64 @llvm.arm64.neon.sqdmulls.scalar(i32 %A, i32 %B)
+  %tmp5 = call i64 @llvm.arm64.neon.sqadd.i64(i64 %C, i64 %tmp4)
+  ret i64 %tmp5
+}
+
+define i64 @sqdmlsl_d(i32 %A, i32 %B, i64 %C) nounwind {
+;CHECK-LABEL: sqdmlsl_d:
+;CHECK: sqdmlsl
+  %tmp4 = call i64 @llvm.arm64.neon.sqdmulls.scalar(i32 %A, i32 %B)
+  %tmp5 = call i64 @llvm.arm64.neon.sqsub.i64(i64 %C, i64 %tmp4)
+  ret i64 %tmp5
+}