Support Intel AMX-TF32

gcc/ChangeLog:

	* common/config/i386/cpuinfo.h (get_available_features):
	Detect AMX-TF32.
	* common/config/i386/i386-common.cc (OPTION_MASK_ISA2_AMX_TF32_SET,
	OPTION_MASK_ISA2_AMX_TF32_UNSET): New.
	(ix86_handle_option): Handle -mamx-tf32.
	* common/config/i386/i386-cpuinfo.h (enum processor_features):
	Add FEATURE_AMX_TF32.
	* common/config/i386/i386-isas.h: Add ISA_NAME_TABLE_ENTRY for
	amx-tf32.
	* config.gcc: Add amxtf32intrin.h
	* config/i386/cpuid.h (bit_AMX_TF32): New.
	* config/i386/i386-c.cc (ix86_target_macros_internal): Handle amx-tf32.
	* config/i386/i386-isa.def (AMX_TF32): Add DEF_PTA(AMX_TF32).
	* config/i386/i386-options.cc (ix86_valid_target_attribute_inner_p):
	Handle amx-tf32.
	* config/i386/i386.opt: Add option -mamx-tf32.
	* config/i386/i386.opt.urls: Regenerated.
	* config/i386/immintrin.h: Include amxtf32intrin.h.
	* doc/extend.texi: Document amx-tf32.
	* doc/invoke.texi: Document -mamx-tf32.
	* doc/sourcebuild.texi: Document target amx-tf32.
	* config/i386/amxtf32intrin.h: New file.

gcc/testsuite/ChangeLog:

	* g++.dg/other/i386-2.C: Add -mamx-tf32.
	* g++.dg/other/i386-3.C: Ditto.
	* gcc.target/i386/amx-check.h: Add cpu check for AMX-TF32.
	* gcc.target/i386/funcspec-56.inc: Add new target attribute.
	* gcc.target/i386/sse-12.c: Add -mamx-tf32.
	* gcc.target/i386/sse-13.c: Ditto.
	* gcc.target/i386/sse-14.c: Ditto.
	* gcc.target/i386/sse-22.c: Add amx-tf32.
	* gcc.target/i386/sse-23.c: Ditto.
	* lib/target-supports.exp (check_effective_target_amx_tf32): New.
	* gcc.target/i386/amx-helper.h: New file for tf32 support.
	* gcc.target/i386/amxtf32-asmatt-1.c: New test.
	* gcc.target/i386/amxtf32-asmintel-1.c: Ditto.
	* gcc.target/i386/amxtf32-mmultf32ps-2.c: Ditto.

8 files changed, 63 insertions, 1 deletions

diff --git a/gcc/config/i386/amxtf32intrin.h b/gcc/config/i386/amxtf32intrin.h
new file mode 100644
index 0000000..450a33e
--- /dev/null
+++ b/gcc/config/i386/amxtf32intrin.h
@@ -0,0 +1,47 @@
+/* Copyright (C) 2024 Free Software Foundation, Inc.
+   This file is part of GCC.
+   GCC is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 3, or (at your option)
+   any later version.
+   GCC is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+   Under Section 7 of GPL version 3, you are granted additional
+   permissions described in the GCC Runtime Library Exception, version
+   3.1, as published by the Free Software Foundation.
+   You should have received a copy of the GNU General Public License and
+   a copy of the GCC Runtime Library Exception along with this program;
+   see the files COPYING3 and COPYING.RUNTIME respectively.  If not, see
+   <http://www.gnu.org/licenses/>.  */
+
+#if !defined _IMMINTRIN_H_INCLUDED
+#error "Never use <amxtf32intrin.h> directly; include <immintrin.h> instead."
+#endif
+
+#ifndef _AMXTF32INTRIN_H_INCLUDED
+#define _AMXTF32INTRIN_H_INCLUDED
+
+#if !defined(__AMX_TF32__)
+#pragma GCC push_options
+#pragma GCC target("amx-tf32")
+#define __DISABLE_AMX_TF32__
+#endif /* __AMX_TF32__ */
+
+#if defined(__x86_64__)
+#define _tile_mmultf32ps_internal(src1_dst,src2,src3)			\
+  __asm__ volatile\
+  ("{tmmultf32ps\t%%tmm"#src3", %%tmm"#src2", %%tmm"#src1_dst"|tmmultf32ps\t%%tmm"#src1_dst", %%tmm"#src2", %%tmm"#src3"}" ::)
+
+#define _tile_mmultf32ps(src1_dst,src2,src3)				\
+  _tile_mmultf32ps_internal (src1_dst, src2, src3)
+
+#endif
+
+#ifdef __DISABLE_AMX_TF32__
+#undef __DISABLE_AMX_TF32__
+#pragma GCC pop_options
+#endif /* __DISABLE_AMX_TF32__ */
+
+#endif /* _AMXTF32INTRIN_H_INCLUDED */
diff --git a/gcc/config/i386/cpuid.h b/gcc/config/i386/cpuid.h
index 2fc163b..7cd8319 100644
--- a/gcc/config/i386/cpuid.h
+++ b/gcc/config/i386/cpuid.h
@@ -164,6 +164,7 @@
 
 /* AMX sub leaf (%eax == 0x1e, %ecx == 1) */
 /* %eax */
+#define bit_AMX_TF32	(1 << 6)
 #define bit_AMX_AVX512  (1 << 7)
 
 /* AVX10 sub leaf (%eax == 0x24) */
diff --git a/gcc/config/i386/i386-c.cc b/gcc/config/i386/i386-c.cc
index 1c36beb..98cb676 100644
--- a/gcc/config/i386/i386-c.cc
+++ b/gcc/config/i386/i386-c.cc
@@ -743,6 +743,8 @@ ix86_target_macros_internal (HOST_WIDE_INT isa_flag,
     def_or_undef (parse_in, "__AVX10_2_512__");
   if (isa_flag2 & OPTION_MASK_ISA2_AMX_AVX512)
     def_or_undef (parse_in, "__AMX_AVX512__");
+  if (isa_flag2 & OPTION_MASK_ISA2_AMX_TF32)
+    def_or_undef (parse_in, "__AMX_TF32__");
   if (TARGET_IAMCU)
     {
       def_or_undef (parse_in, "__iamcu");
diff --git a/gcc/config/i386/i386-isa.def b/gcc/config/i386/i386-isa.def
index fcc3bc4..1b82a69 100644
--- a/gcc/config/i386/i386-isa.def
+++ b/gcc/config/i386/i386-isa.def
@@ -124,3 +124,4 @@ DEF_PTA(AVX10_1_512)
 DEF_PTA(AVX10_2_256)
 DEF_PTA(AVX10_2_512)
 DEF_PTA(AMX_AVX512)
+DEF_PTA(AMX_TF32)
diff --git a/gcc/config/i386/i386-options.cc b/gcc/config/i386/i386-options.cc
index 2f6646f..6f68e79 100644
--- a/gcc/config/i386/i386-options.cc
+++ b/gcc/config/i386/i386-options.cc
@@ -264,7 +264,8 @@ static struct ix86_target_opts isa2_opts[] =
   { "-mavx10.1-512",	OPTION_MASK_ISA2_AVX10_1_512 },
   { "-mavx10.2-256",	OPTION_MASK_ISA2_AVX10_2_256 },
   { "-mavx10.2-512",	OPTION_MASK_ISA2_AVX10_2_512 },
-  { "-mamx-avx512",	OPTION_MASK_ISA2_AMX_AVX512 }
+  { "-mamx-avx512",	OPTION_MASK_ISA2_AMX_AVX512 },
+  { "-mamx-tf32",	OPTION_MASK_ISA2_AMX_TF32 }
 };
 static struct ix86_target_opts isa_opts[] =
 {
@@ -1134,6 +1135,7 @@ ix86_valid_target_attribute_inner_p (tree fndecl, tree args, char *p_strings[],
     IX86_ATTR_ISA ("avx10.2-256", OPT_mavx10_2_256),
     IX86_ATTR_ISA ("avx10.2-512", OPT_mavx10_2_512),
     IX86_ATTR_ISA ("amx-avx512", OPT_mamx_avx512),
+    IX86_ATTR_ISA ("amx-tf32", OPT_mamx_tf32),
 
     /* enum options */
     IX86_ATTR_ENUM ("fpmath=",	OPT_mfpmath_),
diff --git a/gcc/config/i386/i386.opt b/gcc/config/i386/i386.opt
index 232daff..280ad77 100644
--- a/gcc/config/i386/i386.opt
+++ b/gcc/config/i386/i386.opt
@@ -1394,3 +1394,7 @@ mamx-avx512
 Target Mask(ISA2_AMX_AVX512) Var(ix86_isa_flags2) Save
 Support MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, AVX, AVX2, AVX10.1-512,
 AVX10.2-512 and AMX-AVX512 built-in functions and code generation.
+
+mamx-tf32
+Target Mask(ISA2_AMX_TF32) Var(ix86_isa_flags2) Save
+Support AMX-TF32 built-in functions and code generation.
diff --git a/gcc/config/i386/i386.opt.urls b/gcc/config/i386/i386.opt.urls
index 9f590f6..9921cda 100644
--- a/gcc/config/i386/i386.opt.urls
+++ b/gcc/config/i386/i386.opt.urls
@@ -616,3 +616,6 @@ UrlSuffix(gcc/x86-Options.html#index-mavx10_002e2)
 mamx-avx512
 UrlSuffix(gcc/x86-Options.html#index-mamx-avx512)
 
+mamx-tf32
+UrlSuffix(gcc/x86-Options.html#index-mamx-tf32)
+
diff --git a/gcc/config/i386/immintrin.h b/gcc/config/i386/immintrin.h
index 772af56..84b8f60 100644
--- a/gcc/config/i386/immintrin.h
+++ b/gcc/config/i386/immintrin.h
@@ -134,6 +134,8 @@
 
 #include <amxavx512intrin.h>
 
+#include <amxtf32intrin.h>
+
 #include <prfchwintrin.h>
 
 #include <keylockerintrin.h>