[cfe-commits] r112504 - /cfe/trunk/lib/CodeGen/CGBuiltin.cpp

[Bob Wilson]

Author: bwilson
Date: Mon Aug 30 14:57:13 2010
New Revision: 112504

URL: http://llvm.org/viewvc/llvm-project?rev=112504&view=rev
Log:
Translate NEON vmovn builtin to a vector truncation instead of using an llvm
intrinsic.

Modified:
    cfe/trunk/lib/CodeGen/CGBuiltin.cpp

Modified: cfe/trunk/lib/CodeGen/CGBuiltin.cpp
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CGBuiltin.cpp?rev=112504&r1=112503&r2=112504&view=diff
==============================================================================
--- cfe/trunk/lib/CodeGen/CGBuiltin.cpp (original)
+++ cfe/trunk/lib/CodeGen/CGBuiltin.cpp Mon Aug 30 14:57:13 2010
@@ -1420,8 +1420,7 @@
       return Builder.CreateZExt(Ops[0], Ty, "vmovl");
     return Builder.CreateSExt(Ops[0], Ty, "vmovl");
   case ARM::BI__builtin_neon_vmovn_v:
-    return EmitNeonCall(CGM.getIntrinsic(Intrinsic::arm_neon_vmovn, &Ty, 1),
-                        Ops, "vmovn");
+    return Builder.CreateTrunc(Ops[0], Ty, "vmovn");
   case ARM::BI__builtin_neon_vmull_lane_v:
     splat = true;
   case ARM::BI__builtin_neon_vmull_v: