Makefile.in (OBJS): Add graphite-optimize-isl.o.

2012-07-04  Tobias Grosser <tobias@grosser.es>
	Michael Matz  <matz@suse.de>

	* Makefile.in (OBJS): Add graphite-optimize-isl.o.
	(graphite-optimize-isl.o): Add dependencies.
	* common.opt (floop-nest-optimize): New flag.
	* doc/invoke.texi (floop-nest-optimize): Document.
	* graphite-dependences.c (compute_deps): Export.
	* graphite-poly.h (compute_deps): Declare.
	* graphite-optimize-isl.c: New file.
	* graphite-poly.c (apply_poly_transforms): Run the loop
	nest optimizer.
	* tree-ssa-loop.c (gate_graphite_transforms): Enable graphite
	if -floop-nest-optimize is enabled.

Co-Authored-By: Michael Matz <matz@suse.de>

From-SVN: r189249

gcc/ChangeLog

+2012-07-04  Tobias Grosser <tobias@grosser.es>
+	Michael Matz  <matz@suse.de>
+
+	* Makefile.in (OBJS): Add graphite-optimize-isl.o.
+	(graphite-optimize-isl.o): Add dependencies.
+	* common.opt (floop-nest-optimize): New flag.
+	* doc/invoke.texi (floop-nest-optimize): Document.
+	* graphite-dependences.c (compute_deps): Export.
+	* graphite-poly.h (compute_deps): Declare.
+	* graphite-optimize-isl.c: New file.
+	* graphite-poly.c (apply_poly_transforms): Run the loop
+	nest optimizer.
+	* tree-ssa-loop.c (gate_graphite_transforms): Enable graphite
+	if -floop-nest-optimize is enabled.
+
 2012-07-03  Oleg Endo  <olegendo@gcc.gnu.org>
 
 	* config/sh/predicates.md (logical_and_operand): New predicate.

gcc/Makefile.in

@@ -1248,6 +1248,7 @@ OBJS = \
 	graphite-clast-to-gimple.o \
 	graphite-dependences.o \
 	graphite-interchange.o \
+	graphite-optimize-isl.o \
 	graphite-poly.o \
 	graphite-scop-detection.o \
 	graphite-sese-to-poly.o \
@@ -2560,6 +2561,9 @@ graphite-sese-to-poly.o : graphite-sese-to-poly.c $(CONFIG_H) \
    $(SYSTEM_H) coretypes.h $(TREE_FLOW_H) $(TREE_DUMP_H) $(CFGLOOP_H) \
    $(TREE_DATA_REF_H) domwalk.h sese.h graphite-poly.h \
    graphite-sese-to-poly.h
+graphite-optimize-isl.o : graphite-optimize-isl.c $(CONFIG_H) $(SYSTEM_H) \
+    coretypes.h $(TREE_FLOW_H) $(CFGLOOP_H) $(TREE_DATA_REF_H) $(SCEV_H) \
+    $(TREE_DUMP_H) sese.h graphite-poly.h
 tree-vect-loop.o: tree-vect-loop.c $(CONFIG_H) $(SYSTEM_H) coretypes.h \
    $(TM_H) $(GGC_H) $(TREE_H) $(BASIC_BLOCK_H) $(TREE_FLOW_H) \
    $(TREE_DUMP_H) $(CFGLOOP_H) $(EXPR_H) $(RECOG_H) $(OPTABS_H) \

gcc/common.opt

@@ -1219,6 +1219,10 @@ floop-flatten
 Common Ignore
 Does nothing. Preserved for backward compatibility.
 
+floop-nest-optimize
+Common Report Var(flag_loop_optimize_isl) Optimization
+Enable the ISL based loop nest optimizer
+
 fstrict-volatile-bitfields
 Common Report Var(flag_strict_volatile_bitfields) Init(-1)
 Force bitfield accesses to match their type width

gcc/doc/invoke.texi

@@ -373,7 +373,7 @@ Objective-C and Objective-C++ Dialects}.
 -fira-loop-pressure -fno-ira-share-save-slots @gol
 -fno-ira-share-spill-slots -fira-verbose=@var{n} @gol
 -fivopts -fkeep-inline-functions -fkeep-static-consts @gol
--floop-block -floop-interchange -floop-strip-mine @gol
+-floop-block -floop-interchange -floop-strip-mine -floop-nest-optimize @gol
 -floop-parallelize-all -flto -flto-compression-level @gol
 -flto-partition=@var{alg} -flto-report -fmerge-all-constants @gol
 -fmerge-constants -fmodulo-sched -fmodulo-sched-allow-regmoves @gol
@@ -7367,6 +7367,13 @@ GIMPLE -> GRAPHITE -> GIMPLE transformation.  Some minimal optimizations
 are also performed by the code generator CLooG, like index splitting and
 dead code elimination in loops.
 
+@item -floop-nest-optimize
+@opindex floop-nest-optimize
+Enable the ISL based loop nest optimizer.  This is a generic loop nest
+optimizer based on the Pluto optimization algorithms.  It calculates a loop
+structure optimized for data-locality and parallelism.  This option
+is experimental.
+
 @item -floop-parallelize-all
 @opindex floop-parallelize-all
 Use the Graphite data dependence analysis to identify loops that can

gcc/graphite-dependences.c

@@ -443,7 +443,7 @@ subtract_commutative_associative_deps (scop_p scop,
 /* Compute the original data dependences in SCOP for all the reads and
    writes in PBBS.  */
 
-static void
+void
 compute_deps (scop_p scop, VEC (poly_bb_p, heap) *pbbs,
 	      isl_union_map **must_raw,
 	      isl_union_map **may_raw,

gcc/graphite-poly.c

@@ -250,6 +250,11 @@ apply_poly_transforms (scop_p scop)
 	transform_done |= scop_do_interchange (scop);
     }
 
+  /* This pass needs to be run at the final stage, as it does not
+     update the lst.  */
+  if (flag_loop_optimize_isl)
+    transform_done |= optimize_isl (scop);
+
   return transform_done;
 }
 

gcc/graphite-poly.h

@@ -409,6 +409,7 @@ extern int scop_do_interchange (scop_p);
 extern int scop_do_strip_mine (scop_p, int);
 extern bool scop_do_block (scop_p);
 extern bool flatten_all_loops (scop_p);
+extern bool optimize_isl(scop_p);
 extern void pbb_number_of_iterations_at_time (poly_bb_p, graphite_dim_t, mpz_t);
 extern void debug_gmp_value (mpz_t);
 
@@ -1549,4 +1550,20 @@ isl_map *reverse_loop_at_level (poly_bb_p, int);
 isl_union_map *reverse_loop_for_pbbs (scop_p, VEC (poly_bb_p, heap) *, int);
 __isl_give isl_union_map *extend_schedule (__isl_take isl_union_map *);
 
+
+void
+compute_deps (scop_p scop, VEC (poly_bb_p, heap) *pbbs,
+	      isl_union_map **must_raw,
+	      isl_union_map **may_raw,
+	      isl_union_map **must_raw_no_source,
+	      isl_union_map **may_raw_no_source,
+	      isl_union_map **must_war,
+	      isl_union_map **may_war,
+	      isl_union_map **must_war_no_source,
+	      isl_union_map **may_war_no_source,
+	      isl_union_map **must_waw,
+	      isl_union_map **may_waw,
+	      isl_union_map **must_waw_no_source,
+	      isl_union_map **may_waw_no_source);
+
 #endif

gcc/tree-ssa-loop.c

@@ -265,7 +265,8 @@ gate_graphite_transforms (void)
       || flag_loop_interchange
       || flag_loop_strip_mine
       || flag_graphite_identity
-      || flag_loop_parallelize_all)
+      || flag_loop_parallelize_all
+      || flag_loop_optimize_isl)
     flag_graphite = 1;
 
   return flag_graphite != 0;