From: Xi Ruoyao <xry111@xry111.site>
Date: Tue, 17 Oct 2023 13:55:05 +0000 (+0800)
Subject: LoongArch: Use fcmp.caf.s instead of movgr2cf for zeroing a fcc
X-Git-Tag: basepoints/gcc-15~5427
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=b588dcb77e96d77777eb5647cba9e8f454e314dc;p=thirdparty%2Fgcc.git

LoongArch: Use fcmp.caf.s instead of movgr2cf for zeroing a fcc

During the review of an LLVM change [1], on LA464 we found that zeroing
an fcc with fcmp.caf.s is much faster than a movgr2cf from $r0.

[1]: https://github.com/llvm/llvm-project/pull/69300

gcc/ChangeLog:

	* config/loongarch/loongarch.md (movfcc): Use fcmp.caf.s for
	zeroing a fcc.
---

diff --git a/gcc/config/loongarch/loongarch.md b/gcc/config/loongarch/loongarch.md
index 3286b0c56ae4..9f5a75390829 100644
--- a/gcc/config/loongarch/loongarch.md
+++ b/gcc/config/loongarch/loongarch.md
@@ -2150,7 +2150,7 @@
   [(set (match_operand:FCC 0 "register_operand" "=z")
 	(const_int 0))]
   ""
-  "movgr2cf\t%0,$r0")
+  "fcmp.caf.s\t%0,$f0,$f0")
 
 ;; Conditional move instructions.