Implement target hook for loop unrolling

2010-04-08  Christian Borntraeger  <borntraeger@de.ibm.com>
            Wolfgang Gellerich  <gellerich@de.ibm.com>

        Implement target hook for loop unrolling
        * target.h (loop_unroll_adjust): Add a new target hook function.
        * target-def.h (TARGET_LOOP_UNROLL_ADJUST): Likewise.
        * doc/tm.texi (TARGET_LOOP_UNROLL_ADJUST): Document it.
        * config/s390/s390.c (TARGET_LOOP_UNROLL_ADJUST): Define it.
        (s390_loop_unroll_adjust): Implement the new target hook for s390.
        * loop-unroll.c (decide_unroll_runtime_iterations): Call loop unroll target hook
        (decide_unroll_stupid): Likewise.


Co-Authored-By: Wolfgang Gellerich <gellerich@de.ibm.com>

From-SVN: r158132

6 files changed, 96 insertions, 0 deletions

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index 0d9eb6d..ecf1f26 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,15 @@
+2010-04-08  Christian Borntraeger  <borntraeger@de.ibm.com>
+            Wolfgang Gellerich  <gellerich@de.ibm.com>
+
+        Implement target hook for loop unrolling
+        * target.h (loop_unroll_adjust): Add a new target hook function.
+        * target-def.h (TARGET_LOOP_UNROLL_ADJUST): Likewise.
+        * doc/tm.texi (TARGET_LOOP_UNROLL_ADJUST): Document it.
+        * config/s390/s390.c (TARGET_LOOP_UNROLL_ADJUST): Define it.
+        (s390_loop_unroll_adjust): Implement the new target hook for s390.
+        * loop-unroll.c (decide_unroll_runtime_iterations): Call loop unroll target hook
+        (decide_unroll_stupid): Likewise.
+
 2010-04-08  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
 
 	PR target/43643
diff --git a/gcc/config/s390/s390.c b/gcc/config/s390/s390.c
index e6cc4ba..fea70fb 100644
--- a/gcc/config/s390/s390.c
+++ b/gcc/config/s390/s390.c
@@ -53,6 +53,7 @@ along with GCC; see the file COPYING3.  If not see
 #include "gimple.h"
 #include "df.h"
 #include "params.h"
+#include "cfgloop.h"
 
 
 /* Define the specific costs for a given cpu.  */
@@ -10245,6 +10246,62 @@ s390_sched_init (FILE *file ATTRIBUTE_UNUSED,
   last_scheduled_insn = NULL_RTX;
 }
 
+/* This function checks the whole of insn X for memory references. The
+   function always returns zero because the framework it is called
+   from would stop recursively analyzing the insn upon a return value
+   other than zero. The real result of this function is updating
+   counter variable MEM_COUNT.  */
+static int
+check_dpu (rtx *x, unsigned *mem_count)
+{
+  if (*x != NULL_RTX && MEM_P (*x))
+    (*mem_count)++;
+  return 0;
+}
+
+/* This target hook implementation for TARGET_LOOP_UNROLL_ADJUST calculates
+   a new number struct loop *loop should be unrolled if tuned for the z10
+   cpu. The loop is analyzed for memory accesses by calling check_dpu for
+   each rtx of the loop. Depending on the loop_depth and the amount of
+   memory accesses a new number <=nunroll is returned to improve the
+   behaviour of the hardware prefetch unit.  */
+static unsigned
+s390_loop_unroll_adjust (unsigned nunroll, struct loop *loop)
+{
+  basic_block *bbs;
+  rtx insn;
+  unsigned i;
+  unsigned mem_count = 0;
+
+  /* Only z10 needs special handling.  */
+  if (s390_tune != PROCESSOR_2097_Z10)
+    return nunroll;
+
+  /* Count the number of memory references within the loop body.  */
+  bbs = get_loop_body (loop);
+  for (i = 0; i < loop->num_nodes; i++)
+    {
+      for (insn = BB_HEAD (bbs[i]); insn != BB_END (bbs[i]); insn = NEXT_INSN (insn))
+	if (INSN_P (insn) && INSN_CODE (insn) != -1)
+            for_each_rtx (&insn, (rtx_function) check_dpu, &mem_count);
+    }
+  free (bbs);
+
+  /* Prevent division by zero, and we do not need to adjust nunroll in this case.  */
+  if (mem_count == 0)
+    return nunroll;
+
+  switch (loop_depth(loop))
+    {
+    case 1:
+      return MIN (nunroll, 28 / mem_count);
+    case 2:
+      return MIN (nunroll, 22 / mem_count);
+    default:
+      return MIN (nunroll, 16 / mem_count);
+    }
+}
+
 /* Initialize GCC target structure.  */
 
 #undef  TARGET_ASM_ALIGNED_HI_OP
@@ -10373,6 +10430,9 @@ s390_sched_init (FILE *file ATTRIBUTE_UNUSED,
 #undef TARGET_CAN_ELIMINATE
 #define TARGET_CAN_ELIMINATE s390_can_eliminate
 
+#undef TARGET_LOOP_UNROLL_ADJUST
+#define TARGET_LOOP_UNROLL_ADJUST s390_loop_unroll_adjust
+
 #undef TARGET_ASM_TRAMPOLINE_TEMPLATE
 #define TARGET_ASM_TRAMPOLINE_TEMPLATE s390_asm_trampoline_template
 #undef TARGET_TRAMPOLINE_INIT
diff --git a/gcc/doc/tm.texi b/gcc/doc/tm.texi
index 012edd7..222117d 100644
--- a/gcc/doc/tm.texi
+++ b/gcc/doc/tm.texi
@@ -10881,6 +10881,15 @@ This target hook is required only when the target has several different
 modes and they have different conditional execution capability, such as ARM.
 @end deftypefn
 
+@deftypefn {Target Hook} unsigned TARGET_LOOP_UNROLL_ADJUST (unsigned @var{nunroll}, struct loop *@var{loop})
+This target hook returns a new value for the number of times @var{loop}
+should be unrolled. The parameter @var{nunroll} is the number of times
+the loop is to be unrolled. The parameter @var{loop} is a pointer to
+the loop, which is going to be checked for unrolling. This target hook
+is required only when the target has special constraints like maximum
+number of memory accesses.
+@end deftypefn
+
 @defmac POWI_MAX_MULTS
 If defined, this macro is interpreted as a signed integer C expression
 that specifies the maximum number of floating point multiplications
diff --git a/gcc/loop-unroll.c b/gcc/loop-unroll.c
index 231d12d..8ea449d 100644
--- a/gcc/loop-unroll.c
+++ b/gcc/loop-unroll.c
@@ -33,6 +33,7 @@ along with GCC; see the file COPYING3.  If not see
 #include "expr.h"
 #include "hashtab.h"
 #include "recog.h"
+#include "target.h"
 
 /* This pass performs loop unrolling and peeling.  We only perform these
    optimizations on innermost loops (with single exception) because
@@ -826,6 +827,9 @@ decide_unroll_runtime_iterations (struct loop *loop, int flags)
   if (nunroll > (unsigned) PARAM_VALUE (PARAM_MAX_UNROLL_TIMES))
     nunroll = PARAM_VALUE (PARAM_MAX_UNROLL_TIMES);
 
+  if (targetm.loop_unroll_adjust)
+    nunroll = targetm.loop_unroll_adjust (nunroll, loop);
+
   /* Skip big loops.  */
   if (nunroll <= 1)
     {
@@ -1366,6 +1370,9 @@ decide_unroll_stupid (struct loop *loop, int flags)
   if (nunroll > (unsigned) PARAM_VALUE (PARAM_MAX_UNROLL_TIMES))
     nunroll = PARAM_VALUE (PARAM_MAX_UNROLL_TIMES);
 
+  if (targetm.loop_unroll_adjust)
+    nunroll = targetm.loop_unroll_adjust (nunroll, loop);
+
   /* Skip big loops.  */
   if (nunroll <= 1)
     {
diff --git a/gcc/target-def.h b/gcc/target-def.h
index a3c0b2c..d056665 100644
--- a/gcc/target-def.h
+++ b/gcc/target-def.h
@@ -545,6 +545,7 @@
   default_branch_target_register_class
 #define TARGET_BRANCH_TARGET_REGISTER_CALLEE_SAVED hook_bool_bool_false
 #define TARGET_HAVE_CONDITIONAL_EXECUTION default_have_conditional_execution
+#define TARGET_LOOP_UNROLL_ADJUST NULL
 #define TARGET_CANNOT_FORCE_CONST_MEM hook_bool_rtx_false
 #define TARGET_CANNOT_COPY_INSN_P NULL
 #define TARGET_COMMUTATIVE_P hook_bool_const_rtx_commutative_p
@@ -947,6 +948,7 @@
   TARGET_BRANCH_TARGET_REGISTER_CLASS,		\
   TARGET_BRANCH_TARGET_REGISTER_CALLEE_SAVED,	\
   TARGET_HAVE_CONDITIONAL_EXECUTION,		\
+  TARGET_LOOP_UNROLL_ADJUST,			\
   TARGET_CANNOT_FORCE_CONST_MEM,		\
   TARGET_CANNOT_COPY_INSN_P,			\
   TARGET_COMMUTATIVE_P,				\
diff --git a/gcc/target.h b/gcc/target.h
index 41ae3e3..e4e82a3 100644
--- a/gcc/target.h
+++ b/gcc/target.h
@@ -97,6 +97,9 @@ struct _dep;
 /* This is defined in ddg.h .  */
 struct ddg;
 
+/* This is defined in cfgloop.h .  */
+struct loop;
+
 /* Assembler instructions for creating various kinds of integer object.  */
 
 struct asm_int_op
@@ -637,6 +640,9 @@ struct gcc_target
   /* Return true if the target supports conditional execution.  */
   bool (* have_conditional_execution) (void);
 
+  /* Return a new value for loop unroll size.  */
+  unsigned (* loop_unroll_adjust) (unsigned nunroll, struct loop *loop);
+
   /* True if the constant X cannot be placed in the constant pool.  */
   bool (* cannot_force_const_mem) (rtx);