From patchwork Fri Feb  8 18:28:17 2013
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Subject: [i386] : Enable standard vector-mode FMA patterns also for
 !TARGET_SSE_MATH
Date: Fri, 08 Feb 2013 08:28:17 -0000
From: Uros Bizjak <ubizjak@gmail.com>
X-Patchwork-Id: 219244
Message-Id: <CAFULd4Z5LAbvuC=fX5VOHQkxqDEYsHRXYnHhujx62O2MBki7VA@mail.gmail.com>
To: gcc-patches@gcc.gnu.org
Cc: Richard Biener <rguenther@suse.de>

Hello!

As pointed out by Richi in PR56253 [1], we should enable standard
vector-mode FMA patterns also for !TARGET_SSE_MATH, otherwise various
transformations (convert_mult_to_fma) can't convert to a vectorized
FMA.

2013-02-08  Uros Bizjak  <ubizjak@gmail.com>

	* config/i386/sse.md (FMAMODEM): New mode iterator.
	(fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
	mode iterator. Do not use TARGET_SSE_MATH in insn constraint.

Tested on x86_64-pc-linux-gnu {,-m32}, committed to mainline SVN. Will
be committed to 4.7 branch.

[1] http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56253

Uros.


Index: config/i386/sse.md
===================================================================
--- config/i386/sse.md	(revision 195894)
+++ config/i386/sse.md	(working copy)
@@ -2037,42 +2037,46 @@
 ;;
 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 
-(define_mode_iterator FMAMODE [SF DF V4SF V2DF V8SF V4DF])
+;; The standard names for scalar FMA are only available with SSE math enabled.
+(define_mode_iterator FMAMODEM [(SF "TARGET_SSE_MATH")
+				(DF "TARGET_SSE_MATH")
+				V4SF V2DF V8SF V4DF])
 
-;; The standard names for fma is only available with SSE math enabled.
 (define_expand "fma<mode>4"
-  [(set (match_operand:FMAMODE 0 "register_operand")
-	(fma:FMAMODE
-	  (match_operand:FMAMODE 1 "nonimmediate_operand")
-	  (match_operand:FMAMODE 2 "nonimmediate_operand")
-	  (match_operand:FMAMODE 3 "nonimmediate_operand")))]
-  "(TARGET_FMA || TARGET_FMA4) && TARGET_SSE_MATH")
+  [(set (match_operand:FMAMODEM 0 "register_operand")
+	(fma:FMAMODEM
+	  (match_operand:FMAMODEM 1 "nonimmediate_operand")
+	  (match_operand:FMAMODEM 2 "nonimmediate_operand")
+	  (match_operand:FMAMODEM 3 "nonimmediate_operand")))]
+  "TARGET_FMA || TARGET_FMA4")
 
 (define_expand "fms<mode>4"
-  [(set (match_operand:FMAMODE 0 "register_operand")
-	(fma:FMAMODE
-	  (match_operand:FMAMODE 1 "nonimmediate_operand")
-	  (match_operand:FMAMODE 2 "nonimmediate_operand")
-	  (neg:FMAMODE (match_operand:FMAMODE 3 "nonimmediate_operand"))))]
-  "(TARGET_FMA || TARGET_FMA4) && TARGET_SSE_MATH")
+  [(set (match_operand:FMAMODEM 0 "register_operand")
+	(fma:FMAMODEM
+	  (match_operand:FMAMODEM 1 "nonimmediate_operand")
+	  (match_operand:FMAMODEM 2 "nonimmediate_operand")
+	  (neg:FMAMODEM (match_operand:FMAMODEM 3 "nonimmediate_operand"))))]
+  "TARGET_FMA || TARGET_FMA4")
 
 (define_expand "fnma<mode>4"
-  [(set (match_operand:FMAMODE 0 "register_operand")
-	(fma:FMAMODE
-	  (neg:FMAMODE (match_operand:FMAMODE 1 "nonimmediate_operand"))
-	  (match_operand:FMAMODE 2 "nonimmediate_operand")
-	  (match_operand:FMAMODE 3 "nonimmediate_operand")))]
-  "(TARGET_FMA || TARGET_FMA4) && TARGET_SSE_MATH")
+  [(set (match_operand:FMAMODEM 0 "register_operand")
+	(fma:FMAMODEM
+	  (neg:FMAMODEM (match_operand:FMAMODEM 1 "nonimmediate_operand"))
+	  (match_operand:FMAMODEM 2 "nonimmediate_operand")
+	  (match_operand:FMAMODEM 3 "nonimmediate_operand")))]
+  "TARGET_FMA || TARGET_FMA4")
 
 (define_expand "fnms<mode>4"
-  [(set (match_operand:FMAMODE 0 "register_operand")
-	(fma:FMAMODE
-	  (neg:FMAMODE (match_operand:FMAMODE 1 "nonimmediate_operand"))
-	  (match_operand:FMAMODE 2 "nonimmediate_operand")
-	  (neg:FMAMODE (match_operand:FMAMODE 3 "nonimmediate_operand"))))]
-  "(TARGET_FMA || TARGET_FMA4) && TARGET_SSE_MATH")
+  [(set (match_operand:FMAMODEM 0 "register_operand")
+	(fma:FMAMODEM
+	  (neg:FMAMODEM (match_operand:FMAMODEM 1 "nonimmediate_operand"))
+	  (match_operand:FMAMODEM 2 "nonimmediate_operand")
+	  (neg:FMAMODEM (match_operand:FMAMODEM 3 "nonimmediate_operand"))))]
+  "TARGET_FMA || TARGET_FMA4")
 
-;; The builtin for intrinsics is not constrained by SSE math enabled.
+;; The builtins for intrinsics are not constrained by SSE math enabled.
+(define_mode_iterator FMAMODE [SF DF V4SF V2DF V8SF V4DF])
+
 (define_expand "fma4i_fmadd_<mode>"
   [(set (match_operand:FMAMODE 0 "register_operand")
 	(fma:FMAMODE