i386.c (pentium4_cost): New.

	* i386.c (pentium4_cost): New.
	(m_PENT4): New macro.
	(x86_push_memory, x86_movx,x86_cmove, x86_deep_branch, x86_use_sahf
	x86_sub_esp_4, x86_sub_esp_8, x86_add_esp_4, x86_add_esp_8
	x86_integer_DFmode_moves, x86_partial_reg_dependency,
	x86_memory_mismatch_stall): Add Pentium4
	(x86_use_q_reg, x86_use_any_reg): Kill.
	(override_options): Add pentium4.
	(incdec_operand): Return 0 for pentium4.
	(ix86_issue_rate): Add PROCESSOR_PENTIUM4 and PROCESSOR_ATHLON.
	* i386.h (x86_use_q_reg, x86_use_any_reg): Kill.
	(TARGET_PENTIUM4): Define.
	(enum processor_type): Add PROCESSOR_PENTIUM4.
	(CPP_CPU_DEFAULT_SPEC): Add pentium4 support.
	* i386.md (attribute "cpu"): Add pentium4.
	* invoke.texi (march): Add pentium4.

From-SVN: r40134

5 files changed, 86 insertions, 23 deletions

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index 8ed7bbe..4db5d52 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,22 @@
+Wed Feb 28 19:31:42 CET 2001  Jan Hubicka  <jh@suse.cz>
+
+	* i386.c (pentium4_cost): New.
+	(m_PENT4): New macro.
+	(x86_push_memory, x86_movx,x86_cmove, x86_deep_branch, x86_use_sahf
+	x86_sub_esp_4, x86_sub_esp_8, x86_add_esp_4, x86_add_esp_8
+	x86_integer_DFmode_moves, x86_partial_reg_dependency,
+	x86_memory_mismatch_stall): Add Pentium4
+	(x86_use_q_reg, x86_use_any_reg): Kill.
+	(override_options): Add pentium4.
+	(incdec_operand): Return 0 for pentium4.
+	(ix86_issue_rate): Add PROCESSOR_PENTIUM4 and PROCESSOR_ATHLON.
+	* i386.h (x86_use_q_reg, x86_use_any_reg): Kill.
+	(TARGET_PENTIUM4): Define.
+	(enum processor_type): Add PROCESSOR_PENTIUM4.
+	(CPP_CPU_DEFAULT_SPEC): Add pentium4 support.
+	* i386.md (attribute "cpu"): Add pentium4.
+	* invoke.texi (march): Add pentium4.
+
 Wed Feb 28 19:28:06 CET 2001  Jan Hubicka  <jh@suse.cz>
 
 	* i386.md (sse_mov?fcc*): New patterns and splitters.
diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index da323ab..d287d1a 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -239,6 +239,38 @@ struct processor_costs athlon_cost = {
   6					/* MMX or SSE register to integer */
 };
 
+struct processor_costs pentium4_cost = {
+  1,					/* cost of an add instruction */
+  1,					/* cost of a lea instruction */
+  8,					/* variable shift costs */
+  8,					/* constant shift costs */
+  30,					/* cost of starting a multiply */
+  0,					/* cost of multiply per each bit set */
+  112,					/* cost of a divide/mod */
+  16,					/* "large" insn */
+  6,					/* MOVE_RATIO */
+  2,					/* cost for loading QImode using movzbl */
+  {4, 5, 4},				/* cost of loading integer registers
+					   in QImode, HImode and SImode.
+					   Relative to reg-reg move (2).  */
+  {2, 3, 2},				/* cost of storing integer registers */
+  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 */
+  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 */
+  12,					/* cost of moving SSE register */
+  {12, 12, 12},				/* cost of loading SSE registers
+					   in SImode, DImode and TImode */
+  {2, 2, 8},				/* cost of storing SSE registers
+					   in SImode, DImode and TImode */
+  10,					/* MMX or SSE register to integer */
+};
+
 struct processor_costs *ix86_cost = &pentium_cost;
 
 /* Processor feature/optimization bitmasks.  */
@@ -248,19 +280,18 @@ struct processor_costs *ix86_cost = &pentium_cost;
 #define m_PPRO (1<<PROCESSOR_PENTIUMPRO)
 #define m_K6  (1<<PROCESSOR_K6)
 #define m_ATHLON  (1<<PROCESSOR_ATHLON)
+#define m_PENT4  (1<<PROCESSOR_PENTIUM4)
 
 const int x86_use_leave = m_386 | m_K6 | m_ATHLON;
-const int x86_push_memory = m_386 | m_K6 | m_ATHLON;
+const int x86_push_memory = m_386 | m_K6 | m_ATHLON | m_PENT4;
 const int x86_zero_extend_with_and = m_486 | m_PENT;
-const int x86_movx = m_ATHLON | m_PPRO /* m_386 | m_K6 */;
+const int x86_movx = m_ATHLON | m_PPRO | m_PENT4 /* m_386 | m_K6 */;
 const int x86_double_with_add = ~m_386;
 const int x86_use_bit_test = m_386;
 const int x86_unroll_strlen = m_486 | m_PENT | m_PPRO | m_ATHLON | m_K6;
-const int x86_use_q_reg = m_PENT | m_PPRO | m_K6;
-const int x86_use_any_reg = m_486;
-const int x86_cmove = m_PPRO | m_ATHLON;
-const int x86_deep_branch = m_PPRO | m_K6 | m_ATHLON;
-const int x86_use_sahf = m_PPRO | m_K6;
+const int x86_cmove = m_PPRO | m_ATHLON | m_PENT4;
+const int x86_deep_branch = m_PPRO | m_K6 | m_ATHLON | m_PENT4;
+const int x86_use_sahf = m_PPRO | m_K6 | m_PENT4;
 const int x86_partial_reg_stall = m_PPRO;
 const int x86_use_loop = m_K6;
 const int x86_use_fiop = ~(m_PPRO | m_ATHLON | m_PENT);
@@ -270,18 +301,18 @@ const int x86_read_modify_write = ~m_PENT;
 const int x86_read_modify = ~(m_PENT | m_PPRO);
 const int x86_split_long_moves = m_PPRO;
 const int x86_promote_QImode = m_K6 | m_PENT | m_386 | m_486;
-const int x86_single_stringop = m_386;
+const int x86_single_stringop = m_386 | m_PENT4;
 const int x86_qimode_math = ~(0);
 const int x86_promote_qi_regs = 0;
 const int x86_himode_math = ~(m_PPRO);
 const int x86_promote_hi_regs = m_PPRO;
-const int x86_sub_esp_4 = m_ATHLON | m_PPRO;
-const int x86_sub_esp_8 = m_ATHLON | m_PPRO | m_386 | m_486;
-const int x86_add_esp_4 = m_ATHLON | m_K6;
-const int x86_add_esp_8 = m_ATHLON | m_PPRO | m_K6 | m_386 | m_486;
-const int x86_integer_DFmode_moves = ~m_ATHLON;
-const int x86_partial_reg_dependency = m_ATHLON;
-const int x86_memory_mismatch_stall = m_ATHLON;
+const int x86_sub_esp_4 = m_ATHLON | m_PPRO | m_PENT4;
+const int x86_sub_esp_8 = m_ATHLON | m_PPRO | m_386 | m_486 | m_PENT4;
+const int x86_add_esp_4 = m_ATHLON | m_K6 | m_PENT4;
+const int x86_add_esp_8 = m_ATHLON | m_PPRO | m_K6 | m_386 | m_486 | m_PENT4;
+const int x86_integer_DFmode_moves = ~(m_ATHLON | m_PENT4);
+const int x86_partial_reg_dependency = m_ATHLON | m_PENT4;
+const int x86_memory_mismatch_stall = m_ATHLON | m_PENT4;
 
 #define AT_BP(mode) (gen_rtx_MEM ((mode), hard_frame_pointer_rtx))
 
@@ -577,7 +608,8 @@ override_options ()
       {&pentium_cost, 0, 0, -4, -4, -4, 1},
       {&pentiumpro_cost, 0, 0, 4, -4, 4, 1},
       {&k6_cost, 0, 0, -5, -5, 4, 1},
-      {&athlon_cost, 0, 0, 4, -4, 4, 1}
+      {&athlon_cost, 0, 0, 4, -4, 4, 1},
+      {&pentium4_cost, 0, 0, 2, 2, 2, 1}
     };
 
   static struct pta
@@ -595,6 +627,7 @@ override_options ()
       {"pentiumpro", PROCESSOR_PENTIUMPRO},
       {"k6", PROCESSOR_K6},
       {"athlon", PROCESSOR_ATHLON},
+      {"pentium4", PROCESSOR_PENTIUM4},
     };
 
   int const pta_size = sizeof (processor_alias_table) / sizeof (struct pta);
@@ -1202,6 +1235,10 @@ incdec_operand (op, mode)
      register rtx op;
      enum machine_mode mode;
 {
+  /* On Pentium4, the inc and dec operations causes extra dependancy on flag
+     registers, since carry flag is not set.  */
+  if (TARGET_PENTIUM4 && !optimize_size)
+    return 0;
   if (op == const1_rtx || op == constm1_rtx)
     return 1;
   if (GET_CODE (op) != CONST_INT)
@@ -6900,6 +6937,8 @@ ix86_issue_rate ()
       return 2;
 
     case PROCESSOR_PENTIUMPRO:
+    case PROCESSOR_PENTIUM4:
+    case PROCESSOR_ATHLON:
       return 3;
 
     default:
diff --git a/gcc/config/i386/i386.h b/gcc/config/i386/i386.h
index 1086757..b7a74af 100644
--- a/gcc/config/i386/i386.h
+++ b/gcc/config/i386/i386.h
@@ -184,11 +184,12 @@ extern int target_flags;
 #define TARGET_PENTIUMPRO (ix86_cpu == PROCESSOR_PENTIUMPRO)
 #define TARGET_K6 (ix86_cpu == PROCESSOR_K6)
 #define TARGET_ATHLON (ix86_cpu == PROCESSOR_ATHLON)
+#define TARGET_PENTIUM4 (ix86_cpu == PROCESSOR_PENTIUM4)
 
 #define CPUMASK (1 << ix86_cpu)
 extern const int x86_use_leave, x86_push_memory, x86_zero_extend_with_and;
 extern const int x86_use_bit_test, x86_cmove, x86_deep_branch;
-extern const int x86_unroll_strlen, x86_use_q_reg, x86_use_any_reg;
+extern const int x86_unroll_strlen;
 extern const int x86_double_with_add, x86_partial_reg_stall, x86_movx;
 extern const int x86_use_loop, x86_use_fiop, x86_use_mov0;
 extern const int x86_use_cltd, x86_read_modify_write;
@@ -204,8 +205,6 @@ extern const int x86_partial_reg_dependency, x86_memory_mismatch_stall;
 #define TARGET_ZERO_EXTEND_WITH_AND (x86_zero_extend_with_and & CPUMASK)
 #define TARGET_USE_BIT_TEST (x86_use_bit_test & CPUMASK)
 #define TARGET_UNROLL_STRLEN (x86_unroll_strlen & CPUMASK)
-#define TARGET_USE_Q_REG (x86_use_q_reg & CPUMASK)
-#define TARGET_USE_ANY_REG (x86_use_any_reg & CPUMASK)
 /* For sane SSE instruction set generation we need fcomi instruction.  It is
    safe to enable all CMOVE instructions.  */
 #define TARGET_CMOVE ((x86_cmove & (1 << ix86_arch)) || TARGET_SSE)
@@ -345,6 +344,7 @@ enum processor_type
   PROCESSOR_PENTIUMPRO,
   PROCESSOR_K6,
   PROCESSOR_ATHLON,
+  PROCESSOR_PENTIUM4,
   PROCESSOR_max
 };
 
@@ -431,6 +431,9 @@ extern int ix86_arch;
 #if TARGET_CPU_DEFAULT == 5
 #define CPP_CPU_DEFAULT_SPEC "-D__tune_athlon__"
 #endif
+#if TARGET_CPU_DEFAULT == 6
+#define CPP_CPU_DEFAULT_SPEC "-D__tune_pentium4__"
+#endif
 #ifndef CPP_CPU_DEFAULT_SPEC
 #define CPP_CPU_DEFAULT_SPEC "-D__tune_i386__"
 #endif
@@ -449,12 +452,14 @@ extern int ix86_arch;
   %{!mcpu*:-D__tune_i686__ -D__tune_pentiumpro__ }}\
 %{march=k6:-D__k6 -D__k6__ %{!mcpu*:-D__tune_k6__ }}\
 %{march=athlon:-D__athlon -D__athlon__ %{!mcpu*:-D__tune_athlon__ }}\
+%{mpentium4=pentium4:-D__pentium4 -D__pentium4__ %{!mcpu*:-D__tune_pentium4__ }}\
 %{m386|mcpu=i386:-D__tune_i386__ }\
 %{m486|mcpu=i486:-D__tune_i486__ }\
 %{mpentium|mcpu=pentium|mcpu=i586:-D__tune_i586__ -D__tune_pentium__ }\
 %{mpentiumpro|mcpu=pentiumpro|mcpu=i686:-D__tune_i686__ -D__tune_pentiumpro__ }\
 %{mcpu=k6:-D__tune_k6__ }\
 %{mcpu=athlon:-D__tune_athlon__ }\
+%{mcpu=pentium4:-D__tune_pentium4__ }\
 %{!march*:%{!mcpu*:%{!m386:%{!m486:%{!mpentium*:%(cpp_cpu_default)}}}}}"
 #endif
 
diff --git a/gcc/config/i386/i386.md b/gcc/config/i386/i386.md
index e6257a7..aa66d8e 100644
--- a/gcc/config/i386/i386.md
+++ b/gcc/config/i386/i386.md
@@ -97,7 +97,7 @@
 
 ;; Processor type.  This attribute must exactly match the processor_type
 ;; enumeration in i386.h.
-(define_attr "cpu" "i386,i486,pentium,pentiumpro,k6,athlon"
+(define_attr "cpu" "i386,i486,pentium,pentiumpro,k6,athlon,pentium4"
   (const (symbol_ref "ix86_cpu")))
 
 ;; A basic instruction type.  Refinements due to arguments to be
diff --git a/gcc/invoke.texi b/gcc/invoke.texi
index 9804d0d..c95e774 100644
--- a/gcc/invoke.texi
+++ b/gcc/invoke.texi
@@ -6534,14 +6534,14 @@ These @samp{-m} options are defined for the i386 family of computers:
 Assume the defaults for the machine type @var{cpu type} when scheduling
 instructions.  The choices for @var{cpu type} are @samp{i386},
 @samp{i486}, @samp{i586}, @samp{i686}, @samp{pentium},
-@samp{pentiumpro}, @samp{k6}, and @samp{athlon}
+@samp{pentiumpro}, @samp{pentium4}, @samp{k6}, and @samp{athlon}
 
 While picking a specific @var{cpu type} will schedule things appropriately
 for that particular chip, the compiler will not generate any code that
 does not run on the i386 without the @samp{-march=@var{cpu type}} option
 being used.  @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
-is equivalent to @samp{pentiumpro}.  @samp{k6} is the AMD chip as
-opposed to the Intel ones.
+is equivalent to @samp{pentiumpro}.  @samp{k6} and @samp{athlon} sre the
+AMD chips as opposed to the Intel ones.
 
 @item -march=@var{cpu type}
 Generate instructions for the machine type @var{cpu type}.  The choices