i386.h (X86_TUNE_PROMOTE_HIMODE_IMUL): New tuning option.

        * config/i386/i386.h (X86_TUNE_PROMOTE_HIMODE_IMUL): New tuning
        option.
        (TARGET_TUNE_PROMOTE_HIMODE_IMUL): New define. Use new tuning option.
        * config/i386/i386.c (ix86_tune_features): Initialize new
        tuning option.
        * config/i386/predicates.md (promotable_binary_operator): Use
        TARGET_TUNE_PROMOTE_HIMODE_IMUL.

From-SVN: r123301

4 files changed, 45 insertions, 30 deletions

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index 5baf6af..f9c69a1 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,13 @@
+2007-03-28  Uros Bizjak  <ubizjak@gmail.com>
+
+	* config/i386/i386.h (X86_TUNE_PROMOTE_HIMODE_IMUL): New tuning
+	option.
+	(TARGET_TUNE_PROMOTE_HIMODE_IMUL): New define. Use new tuning option.
+	* config/i386/i386.c (ix86_tune_features): Initialize new
+	tuning option.
+	* config/i386/predicates.md (promotable_binary_operator): Use
+	TARGET_TUNE_PROMOTE_HIMODE_IMUL.
+
 2007-03-28  Grigory Zagorodnev <grigory_zagorodnev@linux.intel.com>
 
 	PR target/31380
@@ -85,22 +95,22 @@
 
 2007-03-26  Richard Henderson  <rth@redhat.com>
 
-        PR target/31361
-        * config/i386/i386.c (IX86_BUILTIN_PSLLDQ128, IX86_BUILTIN_PSLLW128,
-        IX86_BUILTIN_PSLLD128, IX86_BUILTIN_PSLLQ128, IX86_BUILTIN_PSRAW128,
-        IX86_BUILTIN_PSRAD128, IX86_BUILTIN_PSRLW128, IX86_BUILTIN_PSRLD128,
-        IX86_BUILTIN_PSRLQ128): New.
-        (ix86_init_mmx_sse_builtins): Add them.
-        (ix86_expand_builtin): Expand them.
-        * config/i386/sse.md (ashr<mode>3, lshr<mode>3, ashl<mode>3): Make
-        operand 2 be TImode.
-        * config/i386/emmintrin.h (_mm_slli_epi64, _mm_srai_epi16,
-        _mm_srai_epi32, _mm_srli_epi16, _mm_srli_epi32,
-        _mm_srli_epi64): Mark __B const.
-        (_mm_srli_si128, _mm_srli_si128): Fix disabled inline versions.
-        (_mm_sll_epi16, _mm_sll_epi32, _mm_sll_epi64, _mm_sra_epi16,
-        _mm_sra_epi32, _mm_srl_epi16, _mm_srl_epi32, _mm_srl_epi64): Use 
-        new two-vector shift builtins.
+	PR target/31361
+	* config/i386/i386.c (IX86_BUILTIN_PSLLDQ128, IX86_BUILTIN_PSLLW128,
+	IX86_BUILTIN_PSLLD128, IX86_BUILTIN_PSLLQ128, IX86_BUILTIN_PSRAW128,
+	IX86_BUILTIN_PSRAD128, IX86_BUILTIN_PSRLW128, IX86_BUILTIN_PSRLD128,
+	IX86_BUILTIN_PSRLQ128): New.
+	(ix86_init_mmx_sse_builtins): Add them.
+	(ix86_expand_builtin): Expand them.
+	* config/i386/sse.md (ashr<mode>3, lshr<mode>3, ashl<mode>3): Make
+	operand 2 be TImode.
+	* config/i386/emmintrin.h (_mm_slli_epi64, _mm_srai_epi16,
+	_mm_srai_epi32, _mm_srli_epi16, _mm_srli_epi32,
+	_mm_srli_epi64): Mark __B const.
+	(_mm_srli_si128, _mm_srli_si128): Fix disabled inline versions.
+	(_mm_sll_epi16, _mm_sll_epi32, _mm_sll_epi64, _mm_sra_epi16,
+	_mm_sra_epi32, _mm_srl_epi16, _mm_srl_epi32, _mm_srl_epi64): Use 
+	new two-vector shift builtins.
 
 2007-03-26  Ian Lance Taylor  <iant@google.com>
 
@@ -288,7 +298,7 @@
 	IX86_BUILTIN_PSRAW128, IX86_BUILTIN_PSRAD128, IX86_BUILTIN_PSRLW128,
 	IX86_BUILTIN_PSRLD128, IX86_BUILTIN_PSRLQ128): Remove.
 	(ix86_init_mmx_sse_builtins): Remove v8hi_ftype_v8hi_v2di and
-	v4si_ftype_v4si_v2di.   Remove __builtin_ia32_psllw128,
+	v4si_ftype_v4si_v2di.  Remove __builtin_ia32_psllw128,
 	__builtin_ia32_pslld128, __builtin_ia32_psllq128,
 	__builtin_ia32_psrlw128, __builtin_ia32_psrld128,
 	__builtin_ia32_psrlq128, __builtin_ia32_psraw128 and
diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index c5e08da..45876bd 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -1200,25 +1200,29 @@ unsigned int ix86_tune_features[X86_TUNE_LAST] = {
   /* X86_TUNE_AVOID_VECTOR_DECODE */
   m_K8 | m_GENERIC64,
 
-  /* X86_TUNE_SLOW_IMUL_IMM32_MEM (imul of 32-bit constant and memory is vector
-     path on AMD machines) */
+  /* X86_TUNE_PROMOTE_HIMODE_IMUL: Modern CPUs have same latency for HImode
+     and SImode multiply, but 386 and 486 do HImode multiply faster.  */
+  ~(m_386 | m_486),
+
+  /* X86_TUNE_SLOW_IMUL_IMM32_MEM: Imul of 32-bit constant and memory is
+     vector path on AMD machines.  */
   m_K8 | m_GENERIC64 | m_AMDFAM10,
 
-  /* X86_TUNE_SLOW_IMUL_IMM8 (imul of 8-bit constant is vector path on AMD
-     machines)  */
+  /* X86_TUNE_SLOW_IMUL_IMM8: Imul of 8-bit constant is vector path on AMD
+     machines.  */
   m_K8 | m_GENERIC64 | m_AMDFAM10,
 
-  /* X86_TUNE_MOVE_M1_VIA_OR (on pentiums, it is faster to load -1 via OR than
-     a MOV) */
+  /* X86_TUNE_MOVE_M1_VIA_OR: On pentiums, it is faster to load -1 via OR
+     than a MOV.  */
   m_PENT,
 
-  /* X86_TUNE_NOT_UNPAIRABLE (NOT is not pairable on Pentium, while XOR is, but
-     one byte longer).  */
+  /* X86_TUNE_NOT_UNPAIRABLE: NOT is not pairable on Pentium, while XOR is,
+     but one byte longer.  */
   m_PENT,
 
-  /* X86_TUNE_NOT_VECTORMODE (On AMD K6, NOT is vector decoded with memory
+  /* X86_TUNE_NOT_VECTORMODE: On AMD K6, NOT is vector decoded with memory
      operand that cannot be represented using a modRM byte.  The XOR
-     replacement is long decoded, so this split helps here as well).  */
+     replacement is long decoded, so this split helps here as well.  */
   m_K6,
 };
 
diff --git a/gcc/config/i386/i386.h b/gcc/config/i386/i386.h
index f0c6ab9..4c62ce5 100644
--- a/gcc/config/i386/i386.h
+++ b/gcc/config/i386/i386.h
@@ -235,6 +235,7 @@ enum ix86_tune_indices {
   X86_TUNE_EXT_80387_CONSTANTS,
   X86_TUNE_SHORTEN_X87_SSE,
   X86_TUNE_AVOID_VECTOR_DECODE,
+  X86_TUNE_PROMOTE_HIMODE_IMUL,
   X86_TUNE_SLOW_IMUL_IMM32_MEM,
   X86_TUNE_SLOW_IMUL_IMM8,
   X86_TUNE_MOVE_M1_VIA_OR,
@@ -312,6 +313,8 @@ extern unsigned int ix86_tune_features[X86_TUNE_LAST];
 #define TARGET_SHORTEN_X87_SSE	ix86_tune_features[X86_TUNE_SHORTEN_X87_SSE]
 #define TARGET_AVOID_VECTOR_DECODE \
 	ix86_tune_features[X86_TUNE_AVOID_VECTOR_DECODE]
+#define TARGET_TUNE_PROMOTE_HIMODE_IMUL \
+	ix86_tune_features[X86_TUNE_PROMOTE_HIMODE_IMUL]
 #define TARGET_SLOW_IMUL_IMM32_MEM \
 	ix86_tune_features[X86_TUNE_SLOW_IMUL_IMM32_MEM]
 #define TARGET_SLOW_IMUL_IMM8	ix86_tune_features[X86_TUNE_SLOW_IMUL_IMM8]
diff --git a/gcc/config/i386/predicates.md b/gcc/config/i386/predicates.md
index 86b6774..97deda6 100644
--- a/gcc/config/i386/predicates.md
+++ b/gcc/config/i386/predicates.md
@@ -963,12 +963,10 @@
 	       mod,udiv,umod,ashift,rotate,ashiftrt,lshiftrt,rotatert"))
 
 ;; Return 1 if OP is a binary operator that can be promoted to wider mode.
-;; Modern CPUs have same latency for HImode and SImode multiply,
-;; but 386 and 486 do HImode multiply faster.  */
 (define_predicate "promotable_binary_operator"
   (ior (match_code "plus,and,ior,xor,ashift")
        (and (match_code "mult")
-	    (match_test "ix86_tune > PROCESSOR_I486"))))
+	    (match_test "TARGET_TUNE_PROMOTE_HIMODE_IMUL"))))
 
 ;; To avoid problems when jump re-emits comparisons like testqi_ext_ccno_0,
 ;; re-recognize the operand to avoid a copy_to_mode_reg that will fail.