re PR tree-optimization/55524 (If fnma exists but not fms, convert_mult_to_fma should...

gcc:
2013-04-10  Joern Rennecke <joern.rennecke@embecosm.com>

        PR tree-optimization/55524
        * tree-ssa-math-opts.c
        (convert_mult_to_fma): Don't use an fms construct
        when we don't have an fms operation, but fnma, and it looks
        likely that we'll be able to use the latter.

gcc/testsuite:
2013-04-10  Joern Rennecke <joern.rennecke@embecosm.com>

        PR tree-optimization/55524
        * gcc.target/epiphany/fnma-1.c: New test.

From-SVN: r197668

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index b8a67893a0a91668d0fb65433fa12e7057a98775..98c6462794d401b6da0a7f6f12e8fd9306b1777b 100644 (file)
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,11 @@
+2013-04-10  Joern Rennecke <joern.rennecke@embecosm.com>
+
+       PR tree-optimization/55524
+       * tree-ssa-math-opts.c
+       (convert_mult_to_fma): Don't use an fms construct
+       when we don't have an fms operation, but fnma, and it looks
+       likely that we'll be able to use the latter.
+
 2013-04-10  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
 
        * cif-code.def (OVERWRITABLE): Correct the comment for overwritable

diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog
index 13585fc07584321bb8c3cbead4fe6e588cb2300a..aacd0094a0951eedfb174e26d31b5e6cc3a71073 100644 (file)
--- a/gcc/testsuite/ChangeLog
+++ b/gcc/testsuite/ChangeLog
@@ -1,3 +1,8 @@
+2013-04-10  Joern Rennecke <joern.rennecke@embecosm.com>
+
+       PR tree-optimization/55524
+       * gcc.target/epiphany/fnma-1.c: New test.
+
 2013-04-10  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
 
        * gcc.dg/tree-ssa/inline-11.c: New test

diff --git a/gcc/testsuite/gcc.target/epiphany/fnma-1.c b/gcc/testsuite/gcc.target/epiphany/fnma-1.c
new file mode 100644 (file)
index 0000000..3155079
--- /dev/null
+++ b/gcc/testsuite/gcc.target/epiphany/fnma-1.c
@@ -0,0 +1,9 @@
+/* { dg-do compile } */
+/* { dg-options "-O2" } */
+/* { dg-final { scan-assembler-times "fmsub\[ \ta-zA-Z0-9\]*," 1 } } */
+
+float
+f (float ar, float ai, float br, float bi)
+{
+  return ar * br - ai * bi;
+}

diff --git a/gcc/tree-ssa-math-opts.c b/gcc/tree-ssa-math-opts.c
index 2140ced495b8cf4f5b2971b31b083c62430070b2..e7e09f6a9decb8a9e00e702df7551219d47395ee 100644 (file)
--- a/gcc/tree-ssa-math-opts.c
+++ b/gcc/tree-ssa-math-opts.c
@@ -2570,6 +2570,24 @@ convert_mult_to_fma (gimple mul_stmt, tree op1, tree op2)
          return false;
        }
 
+      /* If the subtrahend (gimple_assign_rhs2 (use_stmt)) is computed
+        by a MULT_EXPR that we'll visit later, we might be able to
+        get a more profitable match with fnma.
+        OTOH, if we don't, a negate / fma pair has likely lower latency
+        that a mult / subtract pair.  */
+      if (use_code == MINUS_EXPR && !negate_p
+         && gimple_assign_rhs1 (use_stmt) == result
+         && optab_handler (fms_optab, TYPE_MODE (type)) == CODE_FOR_nothing
+         && optab_handler (fnma_optab, TYPE_MODE (type)) != CODE_FOR_nothing)
+       {
+         tree rhs2 = gimple_assign_rhs2 (use_stmt);
+         gimple stmt2 = SSA_NAME_DEF_STMT (rhs2);
+
+         if (has_single_use (rhs2)
+             && gimple_assign_rhs_code (stmt2) == MULT_EXPR)
+           return false;
+       }
+
       /* We can't handle a * b + a * b.  */
       if (gimple_assign_rhs1 (use_stmt) == gimple_assign_rhs2 (use_stmt))
        return false;