From patchwork Tue Oct 19 11:52:32 2010
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Subject: 0002-Fix-32-bit-alignment for Core i7
Date: Tue, 19 Oct 2010 01:52:32 -0000
From: Maxim Kuvyrkov <maxim@codesourcery.com>
X-Patchwork-Id: 68326
Message-Id: <4CBD8680.5010302@codesourcery.com>
To: gcc-patches <gcc-patches@gcc.gnu.org>
Cc: Richard Henderson <rth@redhat.com>,
 Bernd Schmidt <bernds@codesourcery.com>, "H.J. Lu" <hongjiu.lu@intel.com>

This patch adjusts alignment for Core i7 32-bit ABI.  This speeds up 
SPECfp by 0.4% and SPECint by 0.2% in 32-bit mode.

Tested by bootstrapping on i686-pc-linux-gnu.

OK to commit?

Thank you,


diff --git a/gcc/config/i386/i386-c.c b/gcc/config/i386/i386-c.c
index 1846efb..3b4409e 100644
--- a/gcc/config/i386/i386-c.c
+++ b/gcc/config/i386/i386-c.c
@@ -122,6 +122,11 @@ ix86_target_macros_internal (int isa_flag,
       def_or_undef (parse_in, "__core2");
       def_or_undef (parse_in, "__core2__");
       break;
+    case PROCESSOR_COREI7_32:
+    case PROCESSOR_COREI7_64:
+      def_or_undef (parse_in, "__corei7");
+      def_or_undef (parse_in, "__corei7__");
+      break;
     case PROCESSOR_ATOM:
       def_or_undef (parse_in, "__atom");
       def_or_undef (parse_in, "__atom__");
@@ -197,6 +202,10 @@ ix86_target_macros_internal (int isa_flag,
     case PROCESSOR_CORE2:
       def_or_undef (parse_in, "__tune_core2__");
       break;
+    case PROCESSOR_COREI7_32:
+    case PROCESSOR_COREI7_64:
+      def_or_undef (parse_in, "__tune_corei7__");
+      break;
     case PROCESSOR_ATOM:
       def_or_undef (parse_in, "__tune_atom__");
       break;
diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index d04c20e..00d37a1 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -1355,6 +1355,8 @@ const struct processor_costs *ix86_cost = &pentium_cost;
 #define m_PENT4  (1<<PROCESSOR_PENTIUM4)
 #define m_NOCONA  (1<<PROCESSOR_NOCONA)
 #define m_CORE2  (1<<PROCESSOR_CORE2)
+#define m_COREI7_32  (1<<PROCESSOR_COREI7_32)
+#define m_COREI7_64  (1<<PROCESSOR_COREI7_64)
 #define m_ATOM  (1<<PROCESSOR_ATOM)
 
 #define m_GEODE  (1<<PROCESSOR_GEODE)
@@ -1367,8 +1369,8 @@ const struct processor_costs *ix86_cost = &pentium_cost;
 #define m_BDVER1  (1<<PROCESSOR_BDVER1)
 #define m_AMD_MULTIPLE  (m_K8 | m_ATHLON | m_AMDFAM10 | m_BDVER1)
 
-#define m_GENERIC32 (1<<PROCESSOR_GENERIC32)
-#define m_GENERIC64 (1<<PROCESSOR_GENERIC64)
+#define m_GENERIC32 (1<<PROCESSOR_GENERIC32 | m_COREI7_32)
+#define m_GENERIC64 (1<<PROCESSOR_GENERIC64 | m_COREI7_64)
 
 /* Generic instruction choice should be common subset of supported CPUs
    (PPro/PENT4/NOCONA/CORE2/Athlon/K8).  */
@@ -2173,6 +2175,10 @@ static const struct ptt processor_target_table[PROCESSOR_max] =
   {&k8_cost, 16, 7, 16, 7, 16},
   {&nocona_cost, 0, 0, 0, 0, 0},
   {&core2_cost, 16, 10, 16, 10, 16},
+  /* Core i7 32-bit.  */
+  {&generic32_cost, 16, 10, 16, 10, 16},
+  /* Core i7 64-bit.  */
+  {&generic64_cost, 16, 10, 16, 10, 16},
   {&generic32_cost, 16, 7, 16, 7, 16},
   {&generic64_cost, 16, 10, 16, 10, 16},
   {&amdfam10_cost, 32, 24, 32, 7, 32},
diff --git a/gcc/config/i386/i386.h b/gcc/config/i386/i386.h
index 62f35ca..e0c7260 100644
--- a/gcc/config/i386/i386.h
+++ b/gcc/config/i386/i386.h
@@ -239,6 +239,9 @@ extern const struct processor_costs ix86_size_cost;
 #define TARGET_ATHLON_K8 (TARGET_K8 || TARGET_ATHLON)
 #define TARGET_NOCONA (ix86_tune == PROCESSOR_NOCONA)
 #define TARGET_CORE2 (ix86_tune == PROCESSOR_CORE2)
+#define TARGET_COREI7_32 (ix86_tune == PROCESSOR_COREI7_32)
+#define TARGET_COREI7_64 (ix86_tune == PROCESSOR_COREI7_64)
+#define TARGET_COREI7 (TARGET_COREI7_32 || TARGET_COREI7_64)
 #define TARGET_GENERIC32 (ix86_tune == PROCESSOR_GENERIC32)
 #define TARGET_GENERIC64 (ix86_tune == PROCESSOR_GENERIC64)
 #define TARGET_GENERIC (TARGET_GENERIC32 || TARGET_GENERIC64)
@@ -2057,8 +2060,7 @@ do {									\
 	"call " CRT_MKSTR(__USER_LABEL_PREFIX__) #FUNC "\n"	\
 	TEXT_SECTION_ASM_OP);
 
-/* Which processor to schedule for. The cpu attribute defines a list that
-   mirrors this list, so changes to i386.md must be made at the same time.  */
+/* Which processor to tune code generation for.  */
 
 enum processor_type
 {
@@ -2073,6 +2075,8 @@ enum processor_type
   PROCESSOR_K8,
   PROCESSOR_NOCONA,
   PROCESSOR_CORE2,
+  PROCESSOR_COREI7_32,
+  PROCESSOR_COREI7_64,
   PROCESSOR_GENERIC32,
   PROCESSOR_GENERIC64,
   PROCESSOR_AMDFAM10,
diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.texi
index 776fdd0..24a8e87 100644
--- a/gcc/doc/invoke.texi
+++ b/gcc/doc/invoke.texi
@@ -12051,6 +12051,9 @@ SSE2 and SSE3 instruction set support.
 @item core2
 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
 instruction set support.
+@item corei7
+Intel Core i7 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1
+and SSE4.2 instruction set support.
 @item atom
 Intel Atom CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
 instruction set support.