support AVX

COPYRIGHT

@@ -1,5 +1,5 @@
 
-Copyright (c) 2007-2010 MITSUNARI Shigeo
+Copyright (c) 2007-2011 MITSUNARI Shigeo
 All rights reserved.
 
 Redistribution and use in source and binary forms, with or without

readme.txt

@@ -1,5 +1,5 @@
 
-    C++用x86(IA-32), x64(AMD64, x86-64) JITアセンブラ Xbyak version 2.29
+    C++用x86(IA-32), x64(AMD64, x86-64) JITアセンブラ Xbyak version 2.99
 
 -----------------------------------------------------------------------------
 ◎概要
@@ -14,12 +14,12 @@
     xbyak.hをインクルードするだけですぐ利用することができます．
     C++の枠組み内で閉じているため，外部アセンブラは不要です．
     32bit/64bit両対応です．
-    対応ニーモニック:特権命令除くx86, MMX/MMX2/SSE/SSE2/SSE3/SSSE3/SSE4/FPU(一部)
+    対応ニーモニック:特権命令除くx86, MMX/MMX2/SSE/SSE2/SSE3/SSSE3/SSE4/FPU(一部)/AVX
 
 ・Windows Xp(32bit, 64bit), Vista/Linux(32bit, 64bit)/Intel Mac対応
-    Windows Xp上ではVC2005 Express Ed., VC2008
+    Windows Xp上ではVC2005 Express Ed., VC2008, VC2010,
     Windows Vista
-    Linux (kernel 2.4.32)上ではgcc 4.3.0, CentOS 5.1上ではgcc 4.1.2
+    Linux (kernel 2.4.32)上ではgcc 4.5.0, CentOS 5.1上ではgcc 4.1.2
     Intel Mac
     などで動作確認をしています．
 
@@ -199,6 +199,9 @@ sample/{echo,hello}.bfは http://www.kmonos.net/alang/etc/brainfuck.php から
 -----------------------------------------------------------------------------
 ◎履歴
 
+2011/02/04 ver 2.99 beta support AVX
+2010/12/08 ver 2.31 fix ptr [rip + 32bit offset], support rtdscp
+2010/10/19 ver 2.30 support pclmulqdq, aesdec, aesdeclast, aesenc, aesenclast, aesimc, aeskeygenassist
 2010/07/07 ver 2.29 fix call(<label>)
 2010/06/17 ver 2.28 move some member functions to public
 2010/06/01 ver 2.27 support encoding of mov(reg64, imm) like yasm(not nasm)

readme_e.txt

@@ -1,5 +1,5 @@
 
-    Xbyak 2.29 ; JIT assembler for x86(IA32), x64(AMD64, x86-64) by C++
+    Xbyak 2.99 ; JIT assembler for x86(IA32), x64(AMD64, x86-64) by C++
 
 -----------------------------------------------------------------------------
 <Abstract>
@@ -12,11 +12,11 @@ x64(AMD64, x86-64) mnemonic.
 header file only
     you can use Xbyak's functions at once if xbyak.h is included.
 
-MMX/MMX2/SSE/SSE2/SSE3/SSSE3/SSE4/FPU(partial) are available.
+MMX/MMX2/SSE/SSE2/SSE3/SSSE3/SSE4/FPU(partial)/AVX are available.
 
 Windows Xp(32bit, 64bit), Vista, Linux(32bit, 64bit), Intel Mac ready
-support Visual Studio C++ 2005 Express Ed., VC2008 Pro,
-mingw 3.4.2, icc 7.2, gcc 4.1.1, and so on.
+support Visual Studio C++ 2005 Express Ed., VC2008 Pro, VC2010,
+mingw 3.4.2, icc 7.2, gcc 4.5, and so on.
 
 Note: "-fno-operator-names" option is required on gcc to avoid analyzing
 "and", "or", etc. as operators.
@@ -148,6 +148,9 @@ http://www.opensource.org/licenses/bsd-license.php
 -----------------------------------------------------------------------------
 <History>
 
+2011/Feb/04 ver 2.99 beta support AVX
+2010/Dec/08 ver 2.31 fix ptr [rip + 32bit offset], support rdtscp
+2010/Oct/19 ver 2.30 support pclmulqdq, aesdec, aesdeclast, aesenc, aesenclast, aesimc, aeskeygenassist
 2010/Jun/07 ver 2.29 fix call(<label>)
 2010/Jun/17 ver 2.28 move some member functions to public
 2010/Jun/01 ver 2.27 support encoding of mov(reg64, imm) like yasm(not nasm)
@@ -187,5 +190,5 @@ http://www.opensource.org/licenses/bsd-license.php
 MITSUNARI Shigeo(herumi at nifty dot com)
 
 ---
-$Revision: 1.49 $
-$Date: 2010/07/07 01:00:04 $
+$Revision: 1.54 $
+$Date: 2011/02/04 03:27:59 $

sample/Makefile

@@ -14,12 +14,12 @@ endif
 ifeq ($(BIT),64)
 TARGET += test64 bf64 memfunc64 test_util64
 ifeq ($(BOOST_EXIT),1)
-TARGET += calc64
+TARGET += calc64 calc2_64
 endif
 endif
 
 ifeq ($(BOOST_EXIT),1)
-TARGET += calc
+TARGET += calc calc2
 endif
 
 all: $(TARGET)
@@ -38,6 +38,10 @@ calc:
 	g++ $(CFLAGS) calc.cpp -o $@ -m32
 calc64:
 	g++ $(CFLAGS) calc.cpp -o $@ -m64
+#calc2:
+#	g++ $(CFLAGS) calc2.cpp -o $@ -m32
+#calc2_64:
+#	g++ $(CFLAGS) calc2.cpp -o $@ -m64
 
 bf:
 	g++ $(CFLAGS) bf.cpp -o $@ -m32
@@ -67,10 +71,13 @@ test64: test0.cpp $(XBYAK_INC)
 quantize : quantize.cpp $(XBYAK_INC)
 calc : calc.cpp $(XBYAK_INC)
 calc64 : calc.cpp $(XBYAK_INC)
+calc2 : calc2.cpp $(XBYAK_INC)
+calc2_64 : calc2.cpp $(XBYAK_INC)
 bf : bf.cpp $(XBYAK_INC)
 bf64 : bf.cpp $(XBYAK_INC)
 memfunc : memfunc.cpp $(XBYAK_INC)
 memfunc64 : memfunc.cpp $(XBYAK_INC)
 toyvm : toyvm.cpp $(XBYAK_INC)
 test_util : test_util.cpp $(XBYAK_INC) ../xbyak/xbyak_util.h
+test_util2 : test_util.cpp $(XBYAK_INC) ../xbyak/xbyak_util.h
 

sample/bf.cpp

@@ -29,7 +29,7 @@ class Brainfuck : public Xbyak::CodeGenerator {
 		is.unget();
 		return count;
 	}
-	Brainfuck(std::istream& is) : CodeGenerator(10000)
+	Brainfuck(std::istream& is) : CodeGenerator(100000)
 	{
 		// void (*)(void* putchar, void* getchar, int *stack)
 		using namespace Xbyak;
@@ -160,7 +160,7 @@ class Brainfuck : public Xbyak::CodeGenerator {
 
 void dump(const Xbyak::uint8 *code, size_t size)
 {
-	puts("#include <stdio.h>\nstatic int stack[32768];\nstatic const unsigned char code[] = {");
+	puts("#include <stdio.h>\nstatic int stack[128 * 1024];\nstatic const unsigned char code[] = {");
 	for (size_t i = 0; i < size; i++) {
 		printf("0x%02x,", code[i]); if ((i % 16) == 15) putchar('\n');
 	}
@@ -183,23 +183,28 @@ void dump(const Xbyak::uint8 *code, size_t size)
 int main(int argc, char *argv[])
 {
 #ifdef XBYAK32
-	puts("32bit mode");
+	fprintf(stderr, "32bit mode\n");
 #else
-	puts("64bit mode");
+	fprintf(stderr, "64bit mode\n");
 #endif
 	if (argc == 1) {
 		fprintf(stderr, "bf filename.bf [0|1]\n");
 		return 1;
 	}
 	std::ifstream ifs(argv[1]);
 	int mode = argc == 3 ? atoi(argv[2]) : 0;
-	Brainfuck bf(ifs);
-	if (mode == 0) {
-		static int stack[32768];
-		((void (*)(void*, void*, int *))bf.getCode())((void*)putchar, (void*)getchar, stack);
-	} else {
-		dump(bf.getCode(), bf.getSize());
+	try {
+		Brainfuck bf(ifs);
+		if (mode == 0) {
+			static int stack[128 * 1024];
+			((void (*)(void*, void*, int *))bf.getCode())((void*)putchar, (void*)getchar, stack);
+		} else {
+			dump(bf.getCode(), bf.getSize());
+		}
+	} catch (Xbyak::Error err) {
+		printf("ERR:%s(%d)\n", Xbyak::ConvertErrorToString(err), err);
+	} catch (...) {
+		printf("unknown error\n");
 	}
-	return 0;
 }
 

sample/calc.cpp

@@ -1,12 +1,12 @@
 /*
 	@author herumi
-	@date $Date: 2010/04/15 06:52:07 $
+	@date $Date: 2010/11/17 02:58:03 $
 
-	tiny calculator 2
+	tiny calculator
 	This program generates a function to calc the value of
 	polynomial given by user in run-time.
-	use boost::sprit
-
+	use boost::spirit::classic
+	see calc2.cpp for new version of boost::spirit
 */
 #include <stdio.h>
 #include <sstream>
@@ -16,8 +16,6 @@
 	#pragma warning(disable : 4127) // for boost(constant condition)
 	#pragma warning(disable : 4512) // for boost
 #endif
-//#include <boost/spirit/iterator/file_iterator.hpp>
-//#include <boost/spirit/core.hpp>
 #include <boost/spirit/include/classic_file_iterator.hpp>
 #include <boost/spirit/include/classic_core.hpp>
 #include <boost/bind.hpp>

sample/test0.cpp

@@ -138,6 +138,5 @@ int main()
 	} catch (...) {
 		printf("unknown error\n");
 	}
-	return 0;
 }
 

sample/test_util.cpp

@@ -4,6 +4,15 @@
 
 #define NUM_OF_ARRAY(x) (sizeof(x) / sizeof(x[0]))
 
+struct PopCountTest : public Xbyak::CodeGenerator {
+	PopCountTest(int n)
+	{
+		mov(eax, n);
+		popcnt(eax, eax);
+		ret();
+	}
+};
+
 void putCPUinfo()
 {
 	using namespace Xbyak::util;
@@ -27,18 +36,34 @@ void putCPUinfo()
 		{ Cpu::tE3DN, "e3dn" },
 		{ Cpu::tSSE4a, "sse4a" },
 		{ Cpu::tSSE5, "sse5" },
+		{ Cpu::tAESNI, "aesni" },
+		{ Cpu::tRDTSCP, "rdtscp" },
+		{ Cpu::tOSXSACE, "osxsace(xgetvb)" },
+		{ Cpu::tPCLMULQDQ, "pclmulqdq" },
+		{ Cpu::tAVX, "avx" },
+		{ Cpu::tFMA, "fma" },
 	};
 	for (size_t i = 0; i < NUM_OF_ARRAY(tbl); i++) {
 		if (cpu.has(tbl[i].type)) printf(" %s", tbl[i].str);
 	}
 	printf("\n");
+	if (cpu.has(Cpu::tPOPCNT)) {
+		const int n = 0x12345678; // bitcount = 13
+		const int ok = 13;
+		int r = ((int (*)())(PopCountTest(n).getCode()))();
+		if (r == ok) {
+			puts("popcnt ok");
+		} else {
+			printf("popcnt ng %d %d\n", r, ok);
+		}
+	}
 }
 
 #ifdef XBYAK32
-struct EipTest : public Xbyak::util::EnableSetEip<Xbyak::CodeGenerator> {
+struct EipTest : public Xbyak::CodeGenerator {
 	EipTest()
 	{
-		setEipTo(eax);
+		Xbyak::util::setEipTo(this, eax);
 		ret();
 	}
 };

test/Makefile

@@ -32,6 +32,11 @@ test: normalize_prefix jmp
 	./test_address.sh 64
 	./jmp
 
+test_avx:
+	./test_avx.sh
+	./test_avx.sh Y
+	./test_avx.sh 64
+	./test_avx.sh Y64
 clean:
 	rm -rf *.o $(TARGET)