From: Kyrylo Tkachov <kyrylo.tkachov@arm.com>
Date: Tue, 18 Nov 2014 16:26:02 +0000 (+0000)
Subject: [ARM/AArch64] Improve modeled latency between FP operations and FP->GP register moves
X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=58bd290e5b19164a956e08ffadc3684640c61e39;p=gcc.git

[ARM/AArch64] Improve modeled latency between FP operations and FP->GP register moves

	* config/arm/cortex-a15-neon.md (cortex_a15_vfp_to_from_gp):
	Split into...
	(cortex_a15_gp_to_vfp): ...This.
	(cortex_a15_fp_to_gp): ...And this.
	Define and comment bypass from vfp operations to fp->gp moves.

From-SVN: r217725
---

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index a15197eefc6..4a825c17310 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,11 @@
+2014-11-18  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
+
+	* config/arm/cortex-a15-neon.md (cortex_a15_vfp_to_from_gp):
+	Split into...
+	(cortex_a15_gp_to_vfp): ...This.
+	(cortex_a15_fp_to_gp): ...And this.
+	Define and comment bypass from vfp operations to fp->gp moves.
+
 2014-11-18  Martin Liska  <mliska@suse.cz>
 
 	* var-tracking.c (vt_find_locations): New fibonacci_node is used.
diff --git a/gcc/config/arm/cortex-a15-neon.md b/gcc/config/arm/cortex-a15-neon.md
index 02d4a530b2b..bc09cd6a4d7 100644
--- a/gcc/config/arm/cortex-a15-neon.md
+++ b/gcc/config/arm/cortex-a15-neon.md
@@ -655,10 +655,20 @@
        (eq_attr "type" "fmov"))
   "ca15_issue1,ca15_cx_perm")
 
-(define_insn_reservation "cortex_a15_vfp_to_from_gp" 5
+(define_insn_reservation "cortex_a15_gp_to_vfp" 5
   (and (eq_attr "tune" "cortexa15")
-       (eq_attr "type" "f_mcr, f_mcrr, f_mrc, f_mrrc"))
-  "ca15_issue1,ca15_ls1+ca15_ls2")
+       (eq_attr "type" "f_mcr, f_mcrr"))
+  "ca15_issue1,ca15_ls")
+
+(define_insn_reservation "cortex_a15_mov_vfp_to_gp" 5
+  (and (eq_attr "tune" "cortexa15")
+       (eq_attr "type" "f_mrc, f_mrrc"))
+  "ca15_issue1,ca15_ls")
+
+;; Moves from floating point registers to general purpose registers
+;; induce additional latency.
+(define_bypass 10 "cortex_a15_vfp*, cortex_a15_neon*, cortex_a15_gp_to_vfp" "cortex_a15_mov_vfp_to_gp")
+
 
 (define_insn_reservation "cortex_a15_vfp_ariths" 7
   (and (eq_attr "tune" "cortexa15")