i386-protos.h (ix86_memory_move_cost): Move offline.

	* i386-protos.h (ix86_memory_move_cost): Move offline.
	* i386.c (ix86_register_move_cost): Compute properly cost of
	SSE, MMX and i387 instructions.
	(*_cost): Add costs of SSE/MMX moves.
	(ix86_memory_move_cost): Move offline from ....; Likewise.
	* i386.h (MEMORY_MOVE_COST): .... here;
	(struct processor costs): Add new fields to represent costs
	of SSE/MMX moves.

From-SVN: r39689

4 files changed, 200 insertions, 28 deletions

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index 732e24a..f7bb9a0 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,14 @@
+Wed Feb 14 11:12:38 CET 2001  Jan Hubicka  <jh@suse.cz>
+
+	* i386-protos.h (ix86_memory_move_cost): Move offline.
+	* i386.c (ix86_register_move_cost): Compute properly cost of
+	SSE, MMX and i387 instructions.
+	(*_cost): Add costs of SSE/MMX moves.
+	(ix86_memory_move_cost): Move offline from ....; Likewise.
+	* i386.h (MEMORY_MOVE_COST): .... here;
+	(struct processor costs): Add new fields to represent costs
+	of SSE/MMX moves.
+
 Wed Feb 14 10:08:26 CET 2001  Jan Hubicka  <jh@suse.cz>
 
 	* regclass.c (init_reg_sets_1): Reinstall the optimization of
diff --git a/gcc/config/i386/i386-protos.h b/gcc/config/i386/i386-protos.h
index 708bb0f..0e0a520 100644
--- a/gcc/config/i386/i386-protos.h
+++ b/gcc/config/i386/i386-protos.h
@@ -139,6 +139,8 @@ extern int ix86_secondary_memory_needed PARAMS ((enum reg_class,
 						 enum machine_mode, int));
 extern enum reg_class ix86_preferred_reload_class PARAMS ((rtx,
 							   enum reg_class));
+extern int ix86_memory_move_cost PARAMS ((enum machine_mode, enum reg_class,
+					  int));
 
 #ifdef TREE_CODE
 extern void init_cumulative_args PARAMS ((CUMULATIVE_ARGS *, tree, rtx));
diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index 0ef4acc..222dbf7 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -65,7 +65,18 @@ struct processor_costs i386_cost = {	/* 386 specific costs */
   2,					/* cost of reg,reg fld/fst */
   {8, 8, 8},				/* cost of loading fp registers
 					   in SFmode, DFmode and XFmode */
-  {8, 8, 8}				/* cost of loading integer registers */
+  {8, 8, 8},				/* cost of loading integer registers */
+  2,					/* cost of moving MMX register */
+  {4, 8},				/* cost of loading MMX registers
+					   in SImode and DImode */
+  {4, 8},				/* cost of storing MMX registers
+					   in SImode and DImode */
+  2,					/* cost of moving SSE register */
+  {4, 8, 16},				/* cost of loading SSE registers
+					   in SImode, DImode and TImode */
+  {4, 8, 16},				/* cost of storing SSE registers
+					   in SImode, DImode and TImode */
+  3,					/* MMX or SSE register to integer */
 };
 
 struct processor_costs i486_cost = {	/* 486 specific costs */
@@ -86,7 +97,18 @@ struct processor_costs i486_cost = {	/* 486 specific costs */
   2,					/* cost of reg,reg fld/fst */
   {8, 8, 8},				/* cost of loading fp registers
 					   in SFmode, DFmode and XFmode */
-  {8, 8, 8}				/* cost of loading integer registers */
+  {8, 8, 8},				/* cost of loading integer registers */
+  2,					/* cost of moving MMX register */
+  {4, 8},				/* cost of loading MMX registers
+					   in SImode and DImode */
+  {4, 8},				/* cost of storing MMX registers
+					   in SImode and DImode */
+  2,					/* cost of moving SSE register */
+  {4, 8, 16},				/* cost of loading SSE registers
+					   in SImode, DImode and TImode */
+  {4, 8, 16},				/* cost of storing SSE registers
+					   in SImode, DImode and TImode */
+  3					/* MMX or SSE register to integer */
 };
 
 struct processor_costs pentium_cost = {
@@ -107,7 +129,18 @@ struct processor_costs pentium_cost = {
   2,					/* cost of reg,reg fld/fst */
   {2, 2, 6},				/* cost of loading fp registers
 					   in SFmode, DFmode and XFmode */
-  {4, 4, 6}				/* cost of loading integer registers */
+  {4, 4, 6},				/* cost of loading integer registers */
+  8,					/* cost of moving MMX register */
+  {8, 8},				/* cost of loading MMX registers
+					   in SImode and DImode */
+  {8, 8},				/* cost of storing MMX registers
+					   in SImode and DImode */
+  2,					/* cost of moving SSE register */
+  {4, 8, 16},				/* cost of loading SSE registers
+					   in SImode, DImode and TImode */
+  {4, 8, 16},				/* cost of storing SSE registers
+					   in SImode, DImode and TImode */
+  3					/* MMX or SSE register to integer */
 };
 
 struct processor_costs pentiumpro_cost = {
@@ -128,7 +161,18 @@ struct processor_costs pentiumpro_cost = {
   2,					/* cost of reg,reg fld/fst */
   {2, 2, 6},				/* cost of loading fp registers
 					   in SFmode, DFmode and XFmode */
-  {4, 4, 6}				/* cost of loading integer registers */
+  {4, 4, 6},				/* cost of loading integer registers */
+  2,					/* cost of moving MMX register */
+  {2, 2},				/* cost of loading MMX registers
+					   in SImode and DImode */
+  {2, 2},				/* cost of storing MMX registers
+					   in SImode and DImode */
+  2,					/* cost of moving SSE register */
+  {2, 2, 8},				/* cost of loading SSE registers
+					   in SImode, DImode and TImode */
+  {2, 2, 8},				/* cost of storing SSE registers
+					   in SImode, DImode and TImode */
+  3					/* MMX or SSE register to integer */
 };
 
 struct processor_costs k6_cost = {
@@ -149,7 +193,18 @@ struct processor_costs k6_cost = {
   4,					/* cost of reg,reg fld/fst */
   {6, 6, 6},				/* cost of loading fp registers
 					   in SFmode, DFmode and XFmode */
-  {4, 4, 4}				/* cost of loading integer registers */
+  {4, 4, 4},				/* cost of loading integer registers */
+  2,					/* cost of moving MMX register */
+  {2, 2},				/* cost of loading MMX registers
+					   in SImode and DImode */
+  {2, 2},				/* cost of storing MMX registers
+					   in SImode and DImode */
+  2,					/* cost of moving SSE register */
+  {2, 2, 8},				/* cost of loading SSE registers
+					   in SImode, DImode and TImode */
+  {2, 2, 8},				/* cost of storing SSE registers
+					   in SImode, DImode and TImode */
+  6					/* MMX or SSE register to integer */
 };
 
 struct processor_costs athlon_cost = {
@@ -170,7 +225,18 @@ struct processor_costs athlon_cost = {
   4,					/* cost of reg,reg fld/fst */
   {6, 6, 20},				/* cost of loading fp registers
 					   in SFmode, DFmode and XFmode */
-  {4, 4, 16}				/* cost of loading integer registers */
+  {4, 4, 16},				/* cost of loading integer registers */
+  2,					/* cost of moving MMX register */
+  {2, 2},				/* cost of loading MMX registers
+					   in SImode and DImode */
+  {2, 2},				/* cost of storing MMX registers
+					   in SImode and DImode */
+  2,					/* cost of moving SSE register */
+  {2, 2, 8},				/* cost of loading SSE registers
+					   in SImode, DImode and TImode */
+  {2, 2, 8},				/* cost of storing SSE registers
+					   in SImode, DImode and TImode */
+  6					/* MMX or SSE register to integer */
 };
 
 struct processor_costs *ix86_cost = &pentium_cost;
@@ -8854,7 +8920,13 @@ ix86_register_move_cost (mode, class1, class2)
      ??? We should make this cost CPU specific.  */
   if (MMX_CLASS_P (class1) != MMX_CLASS_P (class2)
       || SSE_CLASS_P (class1) != SSE_CLASS_P (class2))
-    return 3;
+    return ix86_cost->mmxsse_to_integer;
+  if (MAYBE_FLOAT_CLASS_P (class1))
+    return ix86_cost->fp_move;
+  if (MAYBE_SSE_CLASS_P (class1))
+    return ix86_cost->sse_move;
+  if (MAYBE_MMX_CLASS_P (class1))
+    return ix86_cost->mmx_move;
   return 2;
 }
 
@@ -8887,3 +8959,96 @@ ix86_hard_regno_mode_ok (regno, mode)
     return 1;
   return reload_in_progress || reload_completed || !TARGET_PARTIAL_REG_STALL;
 }
+
+/* Return the cost of moving data of mode M between a
+   register and memory.  A value of 2 is the default; this cost is
+   relative to those in `REGISTER_MOVE_COST'.
+
+   If moving between registers and memory is more expensive than
+   between two registers, you should define this macro to express the
+   relative cost.  
+ 
+   Model also increased moving costs of QImode registers in non
+   Q_REGS classes.
+ */
+int
+ix86_memory_move_cost (mode, class, in)
+     enum machine_mode mode;
+     enum reg_class class;
+     int in;
+{
+  if (FLOAT_CLASS_P (class))
+    {
+      int index;
+      switch (mode)
+	{
+	  case SFmode:
+	    index = 0;
+	    break;
+	  case DFmode:
+	    index = 1;
+	    break;
+	  case XFmode:
+	  case TFmode:
+	    index = 2;
+	    break;
+	  default:
+	    return 100;
+	}
+      return in ? ix86_cost->fp_load [index] : ix86_cost->fp_store [index];
+    }
+  if (SSE_CLASS_P (class))
+    {
+      int index;
+      switch (GET_MODE_SIZE (mode))
+	{
+	  case 4:
+	    index = 0;
+	    break;
+	  case 8:
+	    index = 1;
+	    break;
+	  case 16:
+	    index = 2;
+	    break;
+	  default:
+	    return 100;
+	}
+      return in ? ix86_cost->sse_load [index] : ix86_cost->sse_store [index];
+    }
+  if (MMX_CLASS_P (class))
+    {
+      int index;
+      switch (GET_MODE_SIZE (mode))
+	{
+	  case 4:
+	    index = 0;
+	    break;
+	  case 8:
+	    index = 1;
+	    break;
+	  default:
+	    return 100;
+	}
+      return in ? ix86_cost->mmx_load [index] : ix86_cost->mmx_store [index];
+    }
+  switch (GET_MODE_SIZE (mode))
+    {
+      case 1:
+	if (in)
+	  return (Q_CLASS_P (class) ? ix86_cost->int_load[0]
+		  : ix86_cost->movzbl_load);
+	else
+	  return (Q_CLASS_P (class) ? ix86_cost->int_store[0]
+		  : ix86_cost->int_store[0] + 4);
+	break;
+      case 2:
+	return in ? ix86_cost->int_load[1] : ix86_cost->int_store[1];
+      default:
+	/* Compute number of 32bit moves needed.  TFmode is moved as XFmode.  */
+	if (mode == TFmode)
+	  mode = XFmode;
+	return ((in ? ix86_cost->int_load[1] : ix86_cost->int_store[1])
+		* (int) GET_MODE_SIZE (mode) / 4);
+    }
+}
diff --git a/gcc/config/i386/i386.h b/gcc/config/i386/i386.h
index 1dd722c..d97d988 100644
--- a/gcc/config/i386/i386.h
+++ b/gcc/config/i386/i386.h
@@ -72,6 +72,18 @@ struct processor_costs {
 				   in SFmode, DFmode and XFmode */
   int fp_store[3];		/* cost of storing FP register
 				   in SFmode, DFmode and XFmode */
+  int mmx_move;			/* cost of moving MMX register.  */
+  int mmx_load[2];		/* cost of loading MMX register
+				   in SImode and DImode */
+  int mmx_store[2];		/* cost of storing MMX register
+				   in SImode and DImode */
+  int sse_move;			/* cost of moving SSE register.  */
+  int sse_load[3];		/* cost of loading SSE register
+				   in SImode, DImode and TImode*/
+  int sse_store[3];		/* cost of storing SSE register
+				   in SImode, DImode and TImode*/
+  int mmxsse_to_integer;	/* cost of moving mmxsse register to
+				   integer and vice versa.  */
 };
 
 extern struct processor_costs *ix86_cost;
@@ -2395,28 +2407,10 @@ while (0)
 
    If moving between registers and memory is more expensive than
    between two registers, you should define this macro to express the
-   relative cost.  
- 
-   Model also increased moving costs of QImode registers in non
-   Q_REGS classes.
- */
+   relative cost.  */
 
-#define MEMORY_MOVE_COST(MODE,CLASS,IN)					\
-  (FLOAT_CLASS_P (CLASS)						\
-   ? (GET_MODE_SIZE (MODE)==4						\
-      ? (IN ? ix86_cost->fp_load[0] : ix86_cost->fp_store[0])		\
-      : (GET_MODE_SIZE (MODE)==8					\
-	 ? (IN ? ix86_cost->fp_load[1] : ix86_cost->fp_store[1])	\
-	 : (IN ? ix86_cost->fp_load[2] : ix86_cost->fp_store[2])))	\
-   : (GET_MODE_SIZE (MODE)==1						\
-      ? (IN ? (Q_CLASS_P (CLASS) ? ix86_cost->int_load[0]		\
-				 : ix86_cost->movzbl_load)		\
-	    : (Q_CLASS_P (CLASS) ? ix86_cost->int_store[0]		\
-				 : ix86_cost->int_store[0] + 4))	\
-      : (GET_MODE_SIZE (MODE)==2					\
-	 ? (IN ? ix86_cost->int_load[1] : ix86_cost->int_store[1])	\
-	 : ((IN ? ix86_cost->int_load[2] : ix86_cost->int_store[2])	\
-	    * (int) GET_MODE_SIZE (MODE) / 4))))
+#define MEMORY_MOVE_COST(MODE,CLASS,IN)	\
+  ix86_memory_move_cost (MODE, CLASS, IN)
 
 /* A C expression for the cost of a branch instruction.  A value of 1
    is the default; other values are interpreted relative to that.  */