Merge branch 'feature-asm' into dev

CMakeLists.txt

@@ -7,6 +7,7 @@ option(WITH_SUMO      "CryptoNight-Heavy support" ON)
 option(WITH_HTTPD     "HTTP REST API" ON)
 option(WITH_DEBUG_LOG "Enable debug log output" OFF)
 option(WITH_TLS       "Enable OpenSSL support" ON)
+option(WITH_ASM       "Enable ASM PoW implementations" ON)
 option(BUILD_STATIC   "Build static binary" OFF)
 
 include (CheckIncludeFile)
@@ -20,6 +21,7 @@ set(HEADERS
     src/common/config/ConfigLoader.h
     src/common/config/ConfigWatcher.h
     src/common/Console.h
+    src/common/cpu/Cpu.h
     src/common/crypto/Algorithm.h
     src/common/crypto/keccak.h
     src/common/interfaces/IClientListener.h
@@ -27,6 +29,7 @@ set(HEADERS
     src/common/interfaces/IConfigCreator.h
     src/common/interfaces/IConsoleListener.h
     src/common/interfaces/IControllerListener.h
+    src/common/interfaces/ICpuInfo.h
     src/common/interfaces/ILogBackend.h
     src/common/interfaces/IStrategy.h
     src/common/interfaces/IStrategyListener.h
@@ -49,7 +52,6 @@ set(HEADERS
     src/common/xmrig.h
     src/core/ConfigLoader_platform.h
     src/core/Controller.h
-    src/Cpu.h
     src/interfaces/IJobResultListener.h
     src/interfaces/IThread.h
     src/interfaces/IWorker.h
@@ -135,7 +137,6 @@ if (WIN32)
         res/app.rc
         src/App_win.cpp
         src/common/Platform_win.cpp
-        src/Cpu_win.cpp
         src/Mem_win.cpp
         )
 
@@ -145,14 +146,12 @@ elseif (APPLE)
     set(SOURCES_OS
         src/App_unix.cpp
         src/common/Platform_mac.cpp
-        src/Cpu_mac.cpp
         src/Mem_unix.cpp
         )
 else()
     set(SOURCES_OS
         src/App_unix.cpp
         src/common/Platform_unix.cpp
-        src/Cpu_unix.cpp
         src/Mem_unix.cpp
         )
 
@@ -184,18 +183,20 @@ if (WITH_LIBCPUID)
 
     include_directories(src/3rdparty/libcpuid)
     set(CPUID_LIB cpuid)
-    set(SOURCES_CPUID src/Cpu.cpp)
+    set(SOURCES_CPUID src/core/cpu/AdvancedCpuInfo.h src/core/cpu/AdvancedCpuInfo.cpp src/core/cpu/Cpu.cpp)
 else()
     add_definitions(/DXMRIG_NO_LIBCPUID)
+    set(SOURCES_CPUID src/common/cpu/BasicCpuInfo.h src/common/cpu/Cpu.cpp)
 
     if (XMRIG_ARM)
-        set(SOURCES_CPUID src/Cpu_arm.cpp)
+        set(SOURCES_CPUID ${SOURCES_CPUID} src/common/cpu/BasicCpuInfo_arm.cpp)
     else()
-        set(SOURCES_CPUID src/Cpu_stub.cpp)
+        set(SOURCES_CPUID ${SOURCES_CPUID} src/common/cpu/BasicCpuInfo.cpp)
     endif()
 endif()
 
 include(cmake/OpenSSL.cmake)
+include(cmake/asm.cmake)
 
 CHECK_INCLUDE_FILE (syslog.h HAVE_SYSLOG_H)
 if (HAVE_SYSLOG_H)
@@ -254,5 +255,5 @@ if (WITH_DEBUG_LOG)
     add_definitions(/DAPP_DEBUG)
 endif()
 
-add_executable(${PROJECT_NAME} ${HEADERS} ${SOURCES} ${SOURCES_OS} ${SOURCES_CPUID} ${HEADERS_CRYPTO} ${SOURCES_CRYPTO} ${SOURCES_SYSLOG} ${HTTPD_SOURCES} ${TLS_SOURCES})
+add_executable(${PROJECT_NAME} ${HEADERS} ${SOURCES} ${SOURCES_OS} ${SOURCES_CPUID} ${HEADERS_CRYPTO} ${SOURCES_CRYPTO} ${SOURCES_SYSLOG} ${HTTPD_SOURCES} ${TLS_SOURCES} ${XMRIG_ASM_SOURCES})
-target_link_libraries(${PROJECT_NAME} ${OPENSSL_LIBRARIES} ${UV_LIBRARIES} ${MHD_LIBRARY} ${EXTRA_LIBS} ${CPUID_LIB})
+target_link_libraries(${PROJECT_NAME} ${XMRIG_ASM_LIBRARY} ${OPENSSL_LIBRARIES} ${UV_LIBRARIES} ${MHD_LIBRARY} ${EXTRA_LIBS} ${CPUID_LIB})

cmake/asm.cmake

@@ -0,0 +1,27 @@
+if (WITH_ASM AND NOT XMRIG_ARM AND CMAKE_SIZEOF_VOID_P EQUAL 8)
+    set(XMRIG_ASM_LIBRARY "xmrig-asm")
+
+    if (CMAKE_C_COMPILER_ID MATCHES MSVC)
+        enable_language(ASM_MASM)
+        set(XMRIG_ASM_FILE "src/crypto/asm/cnv2_main_loop.asm")
+        set_property(SOURCE ${XMRIG_ASM_FILE} PROPERTY ASM_MASM)
+    else()
+        enable_language(ASM)
+
+        if (WIN32 AND CMAKE_C_COMPILER_ID MATCHES GNU)
+            set(XMRIG_ASM_FILE "src/crypto/asm/cnv2_main_loop_win.S")
+        else()
+            set(XMRIG_ASM_FILE "src/crypto/asm/cnv2_main_loop.S")
+        endif()
+
+        set_property(SOURCE ${XMRIG_ASM_FILE} PROPERTY C)
+    endif()
+
+    add_library(${XMRIG_ASM_LIBRARY} STATIC ${XMRIG_ASM_FILE})
+    set(XMRIG_ASM_SOURCES src/crypto/Asm.h src/crypto/Asm.cpp)
+    set_property(TARGET ${XMRIG_ASM_LIBRARY} PROPERTY LINKER_LANGUAGE C)
+else()
+    set(XMRIG_ASM_SOURCES "")
+    set(XMRIG_ASM_LIBRARY "")
+    add_definitions(/DXMRIG_NO_ASM)
+endif()

src/3rdparty/libcpuid/asm-bits.h

@@ -1,53 +1,71 @@
-/*
+/*
- * Copyright 2008  Veselin Georgiev,
+ * Copyright 2008  Veselin Georgiev,
- * anrieffNOSPAM @ mgail_DOT.com (convert to gmail)
+ * anrieffNOSPAM @ mgail_DOT.com (convert to gmail)
- *
+ *
- * Redistribution and use in source and binary forms, with or without
+ * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
+ * modification, are permitted provided that the following conditions
- * are met:
+ * are met:
- *
+ *
- * 1. Redistributions of source code must retain the above copyright
+ * 1. Redistributions of source code must retain the above copyright
- *    notice, this list of conditions and the following disclaimer.
+ *    notice, this list of conditions and the following disclaimer.
- * 2. Redistributions in binary form must reproduce the above copyright
+ * 2. Redistributions in binary form must reproduce the above copyright
- *    notice, this list of conditions and the following disclaimer in the
+ *    notice, this list of conditions and the following disclaimer in the
- *    documentation and/or other materials provided with the distribution.
+ *    documentation and/or other materials provided with the distribution.
- *
+ *
- * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
- * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
+ * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
- * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+ * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
- * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
- * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
- * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
- * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- */
+ */
-#ifndef __ASM_BITS_H__
+#ifndef __ASM_BITS_H__
-#define __ASM_BITS_H__
+#define __ASM_BITS_H__
-#include "libcpuid.h"
+#include "libcpuid.h"
-
+
-/* Determine Compiler: */
+/* Determine Compiler: */
-#if defined(_MSC_VER)
+#if defined(_MSC_VER)
-#	define COMPILER_MICROSOFT
+#if !defined(COMPILER_MICROSOFT)
-#elif defined(__GNUC__)
+#	define COMPILER_MICROSOFT
-#	define COMPILER_GCC
+#endif
-#endif
+#elif defined(__GNUC__)
-
+#if !defined(COMPILER_GCC)
-/* Determine Platform */
+#	define COMPILER_GCC
-#if defined(__x86_64__) || defined(_M_AMD64)
+#endif
-#	define PLATFORM_X64
+#elif defined(__clang__)
-#elif defined(__i386__) || defined(_M_IX86)
+#if !defined(COMPILER_CLANG)
-#	define PLATFORM_X86
+#	define COMPILER_CLANG
-#endif
+#endif
-
+#endif
-/* Under Windows/AMD64 with MSVC, inline assembly isn't supported */
+
-#if (defined(COMPILER_GCC) && defined(PLATFORM_X64)) || defined(PLATFORM_X86)
+/* Determine Platform */
-#	define INLINE_ASM_SUPPORTED
+#if defined(__x86_64__) || defined(_M_AMD64)
-#endif
+#if !defined(PLATFORM_X64)
-
+#	define PLATFORM_X64
-int cpuid_exists_by_eflags(void);
+#endif
-void exec_cpuid(uint32_t *regs);
+#elif defined(__i386__) || defined(_M_IX86)
-void busy_sse_loop(int cycles);
+#if !defined(PLATFORM_X86)
-
+#	define PLATFORM_X86
-#endif /* __ASM_BITS_H__ */
+#endif
+#elif defined(__ARMEL__)
+#if !defined(PLATFORM_ARM)
+#	define PLATFORM_ARM
+#endif
+#endif
+
+/* Under Windows/AMD64 with MSVC, inline assembly isn't supported */
+#if (((defined(COMPILER_GCC) || defined(COMPILER_CLANG))) &&  \
+     (defined(PLATFORM_X64) || defined(PLATFORM_X86) || defined(PLATFORM_ARM))) || \
+	 (defined(COMPILER_MICROSOFT) && defined(PLATFORM_X86))
+#	define INLINE_ASM_SUPPORTED
+#endif
+
+int cpuid_exists_by_eflags(void);
+void exec_cpuid(uint32_t *regs);
+void busy_sse_loop(int cycles);
+
+#endif /* __ASM_BITS_H__ */

src/3rdparty/libcpuid/cpuid_main.c

@@ -221,42 +221,42 @@ static void load_features_common(struct cpu_raw_data_t* raw, struct cpu_id_t* da
 
 static cpu_vendor_t cpuid_vendor_identify(const uint32_t *raw_vendor, char *vendor_str)
 {
-    int i;
+	int i;
-    cpu_vendor_t vendor = VENDOR_UNKNOWN;
+	cpu_vendor_t vendor = VENDOR_UNKNOWN;
-    const struct { cpu_vendor_t vendor; char match[16]; }
+	const struct { cpu_vendor_t vendor; char match[16]; }
-    matchtable[NUM_CPU_VENDORS] = {
+	matchtable[NUM_CPU_VENDORS] = {
-        /* source: http://www.sandpile.org/ia32/cpuid.htm */
+		/* source: http://www.sandpile.org/ia32/cpuid.htm */
-        { VENDOR_INTEL		, "GenuineIntel" },
+		{ VENDOR_INTEL		, "GenuineIntel" },
-        { VENDOR_AMD		, "AuthenticAMD" },
+		{ VENDOR_AMD		, "AuthenticAMD" },
-        { VENDOR_CYRIX		, "CyrixInstead" },
+		{ VENDOR_CYRIX		, "CyrixInstead" },
-        { VENDOR_NEXGEN		, "NexGenDriven" },
+		{ VENDOR_NEXGEN		, "NexGenDriven" },
-        { VENDOR_TRANSMETA	, "GenuineTMx86" },
+		{ VENDOR_TRANSMETA	, "GenuineTMx86" },
-        { VENDOR_UMC		, "UMC UMC UMC " },
+		{ VENDOR_UMC		, "UMC UMC UMC " },
-        { VENDOR_CENTAUR	, "CentaurHauls" },
+		{ VENDOR_CENTAUR	, "CentaurHauls" },
-        { VENDOR_RISE		, "RiseRiseRise" },
+		{ VENDOR_RISE		, "RiseRiseRise" },
-        { VENDOR_SIS		, "SiS SiS SiS " },
+		{ VENDOR_SIS		, "SiS SiS SiS " },
-        { VENDOR_NSC		, "Geode by NSC" },
+		{ VENDOR_NSC		, "Geode by NSC" },
-    };
+	};
 
-    memcpy(vendor_str + 0, &raw_vendor[1], 4);
+	memcpy(vendor_str + 0, &raw_vendor[1], 4);
-    memcpy(vendor_str + 4, &raw_vendor[3], 4);
+	memcpy(vendor_str + 4, &raw_vendor[3], 4);
-    memcpy(vendor_str + 8, &raw_vendor[2], 4);
+	memcpy(vendor_str + 8, &raw_vendor[2], 4);
-    vendor_str[12] = 0;
+	vendor_str[12] = 0;
 
-    /* Determine vendor: */
+	/* Determine vendor: */
-    for (i = 0; i < NUM_CPU_VENDORS; i++)
+	for (i = 0; i < NUM_CPU_VENDORS; i++)
-        if (!strcmp(vendor_str, matchtable[i].match)) {
+		if (!strcmp(vendor_str, matchtable[i].match)) {
-            vendor = matchtable[i].vendor;
+			vendor = matchtable[i].vendor;
-            break;
+			break;
-        }
+		}
-    return vendor;
+	return vendor;
 }
 
 static int cpuid_basic_identify(struct cpu_raw_data_t* raw, struct cpu_id_t* data)
 {
 	int i, j, basic, xmodel, xfamily, ext;
 	char brandstr[64] = {0};
-    data->vendor = cpuid_vendor_identify(raw->basic_cpuid[0], data->vendor_str);
+	data->vendor = cpuid_vendor_identify(raw->basic_cpuid[0], data->vendor_str);
 
 	if (data->vendor == VENDOR_UNKNOWN)
 		return set_error(ERR_CPU_UNKN);

src/3rdparty/libcpuid/libcpuid.h

@@ -60,7 +60,7 @@
  */
 
 /** @mainpage A simple libcpuid introduction
- *
+ * 
  * LibCPUID provides CPU identification and access to the CPUID and RDTSC
  * instructions on the x86.
  * <p>
@@ -82,6 +82,7 @@
  */
 
 /** @defgroup libcpuid LibCPUID
+ * @brief LibCPUID provides CPU identification
  @{ */
 
 /* Include some integer type specifications: */
@@ -535,23 +536,23 @@ typedef enum {
  * @brief Describes common library error codes
  */
 typedef enum {
-	ERR_OK       =  0,	/*!< "No error" */
+	ERR_OK       =  0,	/*!< No error */
-	ERR_NO_CPUID = -1,	/*!< "CPUID instruction is not supported" */
+	ERR_NO_CPUID = -1,	/*!< CPUID instruction is not supported */
-	ERR_NO_RDTSC = -2,	/*!< "RDTSC instruction is not supported" */
+	ERR_NO_RDTSC = -2,	/*!< RDTSC instruction is not supported */
-	ERR_NO_MEM   = -3,	/*!< "Memory allocation failed" */
+	ERR_NO_MEM   = -3,	/*!< Memory allocation failed */
-	ERR_OPEN     = -4,	/*!< "File open operation failed" */
+	ERR_OPEN     = -4,	/*!< File open operation failed */
-	ERR_BADFMT   = -5,	/*!< "Bad file format" */
+	ERR_BADFMT   = -5,	/*!< Bad file format */
-	ERR_NOT_IMP  = -6,	/*!< "Not implemented" */
+	ERR_NOT_IMP  = -6,	/*!< Not implemented */
-	ERR_CPU_UNKN = -7,	/*!< "Unsupported processor" */
+	ERR_CPU_UNKN = -7,	/*!< Unsupported processor */
-	ERR_NO_RDMSR = -8,	/*!< "RDMSR instruction is not supported" */
+	ERR_NO_RDMSR = -8,	/*!< RDMSR instruction is not supported */
-	ERR_NO_DRIVER= -9,	/*!< "RDMSR driver error (generic)" */
+	ERR_NO_DRIVER= -9,	/*!< RDMSR driver error (generic) */
-	ERR_NO_PERMS = -10,	/*!< "No permissions to install RDMSR driver" */
+	ERR_NO_PERMS = -10,	/*!< No permissions to install RDMSR driver */
-	ERR_EXTRACT  = -11,	/*!< "Cannot extract RDMSR driver (read only media?)" */
+	ERR_EXTRACT  = -11,	/*!< Cannot extract RDMSR driver (read only media?) */
-	ERR_HANDLE   = -12,	/*!< "Bad handle" */
+	ERR_HANDLE   = -12,	/*!< Bad handle */
-	ERR_INVMSR   = -13,	/*!< "Invalid MSR" */
+	ERR_INVMSR   = -13,	/*!< Invalid MSR */
-	ERR_INVCNB   = -14,	/*!< "Invalid core number" */
+	ERR_INVCNB   = -14,	/*!< Invalid core number */
-	ERR_HANDLE_R = -15,	/*!< "Error on handle read" */
+	ERR_HANDLE_R = -15,	/*!< Error on handle read */
-	ERR_INVRANGE = -16,	/*!< "Invalid given range" */
+	ERR_INVRANGE = -16,	/*!< Invalid given range */
 } cpu_error_t;
 
 /**
@@ -668,7 +669,7 @@ struct cpu_epc_t cpuid_get_epc(int index, const struct cpu_raw_data_t* raw);
 const char* cpuid_lib_version(void);
 
 #ifdef __cplusplus
-}; /* extern "C" */
+} /* extern "C" */
 #endif
 
 

src/3rdparty/libcpuid/libcpuid_internal.h

@@ -75,8 +75,9 @@ enum _intel_bits_t {
 	_3                      = LBIT( 14 ),
 	_5                      = LBIT( 15 ),
 	_7                      = LBIT( 16 ),
-	XEON_                   = LBIT( 17 ),
+	_9                      = LBIT( 17 ),
-	ATOM_                   = LBIT( 18 ),
+	XEON_                   = LBIT( 18 ),
+	ATOM_                   = LBIT( 19 ),
 };
 typedef enum _intel_bits_t intel_bits_t;
 

src/3rdparty/libcpuid/libcpuid_types.h

@@ -32,6 +32,32 @@
 #ifndef __LIBCPUID_TYPES_H__
 #define __LIBCPUID_TYPES_H__
 
-#include <stdint.h>
+#if !defined(_MSC_VER) || _MSC_VER >= 1600
+#  include <stdint.h>
+#else
+/* we have to provide our own: */
+#  if !defined(__int32_t_defined)
+typedef int int32_t;
+#  endif
+
+#  if !defined(__uint32_t_defined)
+typedef unsigned uint32_t;
+#  endif
+
+typedef signed char		int8_t;
+typedef unsigned char		uint8_t;
+typedef signed short		int16_t;
+typedef unsigned short		uint16_t;
+#if (defined _MSC_VER) && (_MSC_VER <= 1300)
+	/* MSVC 6.0: no long longs ... */
+	typedef signed __int64		int64_t;
+	typedef unsigned __int64	uint64_t;
+#else
+	/* all other sane compilers: */
+	typedef signed long long   int64_t;
+	typedef unsigned long long uint64_t;
+#endif
+
+#endif
 
 #endif /* __LIBCPUID_TYPES_H__ */

src/3rdparty/libcpuid/recog_amd.c

@@ -49,6 +49,10 @@ enum _amd_model_codes_t {
 	_1400,
 	_1500,
 	_1600,
+	_1900,
+	_2400,
+	_2500,
+	_2700,
 };
 
 static void load_amd_features(struct cpu_raw_data_t* raw, struct cpu_id_t* data)

src/3rdparty/libcpuid/recog_intel.c

@@ -376,7 +376,7 @@ static intel_code_and_bits_t get_brand_code_and_bits(struct cpu_id_t* data)
 			bits |= bit_matchtable[i].bit;
 	}
 	
-	if ((i = match_pattern(bs, "Core(TM) [im][357]")) != 0) {
+	if ((i = match_pattern(bs, "Core(TM) [im][3579]")) != 0) {
 		bits |= CORE_;
 		i--;
 		switch (bs[i + 9]) {
@@ -387,6 +387,7 @@ static intel_code_and_bits_t get_brand_code_and_bits(struct cpu_id_t* data)
 			case '3': bits |= _3; break;
 			case '5': bits |= _5; break;
 			case '7': bits |= _7; break;
+			case '9': bits |= _9; break;
 		}
 	}
 	for (i = 0; i < COUNT_OF(matchtable); i++)

src/App.cpp

@@ -29,11 +29,11 @@
 #include "api/Api.h"
 #include "App.h"
 #include "common/Console.h"
+#include "common/cpu/Cpu.h"
 #include "common/log/Log.h"
 #include "common/Platform.h"
 #include "core/Config.h"
 #include "core/Controller.h"
-#include "Cpu.h"
 #include "crypto/CryptoNight.h"
 #include "Mem.h"
 #include "net/Network.h"

src/Cpu.h

@@ -1,66 +0,0 @@
-/* XMRig
- * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
- * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
- * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
- * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
- * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
- * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
- * Copyright 2016-2018 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
- *
- *   This program 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 of the License, or
- *   (at your option) any later version.
- *
- *   This program 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.
- *
- *   You should have received a copy of the GNU General Public License
- *   along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#ifndef __CPU_H__
-#define __CPU_H__
-
-
-#include <stdint.h>
-
-
-class Cpu
-{
-public:
-    enum Flags {
-        X86_64 = 1,
-        AES    = 2,
-        BMI2   = 4
-    };
-
-    static size_t optimalThreadsCount(size_t size, int maxCpuUsage);
-    static void init();
-
-    static inline bool hasAES()       { return (m_flags & AES) != 0; }
-    static inline bool isX64()        { return (m_flags & X86_64) != 0; }
-    static inline const char *brand() { return m_brand; }
-    static inline int cores()         { return m_totalCores; }
-    static inline int l2()            { return m_l2_cache; }
-    static inline int l3()            { return m_l3_cache; }
-    static inline int sockets()       { return m_sockets; }
-    static inline int threads()       { return m_totalThreads; }
-
-private:
-    static void initCommon();
-
-    static bool m_l2_exclusive;
-    static char m_brand[64];
-    static int m_flags;
-    static int m_l2_cache;
-    static int m_l3_cache;
-    static int m_sockets;
-    static int m_totalCores;
-    static size_t m_totalThreads;
-};
-
-
-#endif /* __CPU_H__ */

src/Summary.cpp

@@ -27,16 +27,33 @@
 #include <uv.h>
 
 
+#include "common/cpu/Cpu.h"
 #include "common/log/Log.h"
 #include "common/net/Pool.h"
 #include "core/Config.h"
 #include "core/Controller.h"
-#include "Cpu.h"
+#include "crypto/Asm.h"
 #include "Mem.h"
 #include "Summary.h"
 #include "version.h"
 
 
+#ifndef XMRIG_NO_ASM
+static const char *coloredAsmNames[] = {
+    "\x1B[1;31mnone\x1B[0m",
+    "auto",
+    "\x1B[1;32mintel\x1B[0m",
+    "\x1B[1;32mryzen\x1B[0m"
+};
+
+
+inline static const char *asmName(xmrig::Assembly assembly, bool colors)
+{
+    return colors ? coloredAsmNames[assembly] : xmrig::Asm::toString(assembly);
+}
+#endif
+
+
 static void print_memory(xmrig::Config *config) {
 #   ifdef _WIN32
     if (config->isColors()) {
@@ -52,21 +69,23 @@ static void print_memory(xmrig::Config *config) {
 
 static void print_cpu(xmrig::Config *config)
 {
+    using namespace xmrig;
+
     if (config->isColors()) {
         Log::i()->text(GREEN_BOLD(" * ") WHITE_BOLD("%-13s%s (%d)") " %sx64 %sAES",
                        "CPU",
-                       Cpu::brand(),
+                       Cpu::info()->brand(),
-                       Cpu::sockets(),
+                       Cpu::info()->sockets(),
-                       Cpu::isX64() ? "\x1B[1;32m" : "\x1B[1;31m-",
+                       Cpu::info()->isX64() ? "\x1B[1;32m" : "\x1B[1;31m-",
-                       Cpu::hasAES() ? "\x1B[1;32m" : "\x1B[1;31m-");
+                       Cpu::info()->hasAES() ? "\x1B[1;32m" : "\x1B[1;31m-");
 #       ifndef XMRIG_NO_LIBCPUID
-        Log::i()->text(GREEN_BOLD(" * ") WHITE_BOLD("%-13s%.1f MB/%.1f MB"), "CPU L2/L3", Cpu::l2() / 1024.0, Cpu::l3() / 1024.0);
+        Log::i()->text(GREEN_BOLD(" * ") WHITE_BOLD("%-13s%.1f MB/%.1f MB"), "CPU L2/L3", Cpu::info()->L2() / 1024.0, Cpu::info()->L3() / 1024.0);
 #       endif
     }
     else {
-        Log::i()->text(" * %-13s%s (%d) %sx64 %sAES", "CPU", Cpu::brand(), Cpu::sockets(), Cpu::isX64() ? "" : "-", Cpu::hasAES() ? "" : "-");
+        Log::i()->text(" * %-13s%s (%d) %sx64 %sAES", "CPU", Cpu::info()->brand(), Cpu::info()->sockets(), Cpu::info()->isX64() ? "" : "-", Cpu::info()->hasAES() ? "" : "-");
 #       ifndef XMRIG_NO_LIBCPUID
-        Log::i()->text(" * %-13s%.1f MB/%.1f MB", "CPU L2/L3", Cpu::l2() / 1024.0, Cpu::l3() / 1024.0);
+        Log::i()->text(" * %-13s%.1f MB/%.1f MB", "CPU L2/L3", Cpu::info()->L2() / 1024.0, Cpu::info()->L3() / 1024.0);
 #       endif
     }
 }
@@ -99,6 +118,18 @@ static void print_threads(xmrig::Config *config)
                        config->isColors() && config->donateLevel() == 0 ? "\x1B[1;31m" : "",
                        config->donateLevel());
     }
+
+#   ifndef XMRIG_NO_ASM
+    if (config->assembly() == xmrig::ASM_AUTO) {
+        const xmrig::Assembly assembly = xmrig::Cpu::info()->assembly();
+
+        Log::i()->text(config->isColors() ? GREEN_BOLD(" * ") WHITE_BOLD("%-13sauto:%s")
+                                          : " * %-13sauto:%s", "ASSEMBLY", asmName(assembly, config->isColors()));
+    }
+    else {
+        Log::i()->text(config->isColors() ? GREEN_BOLD(" * ") WHITE_BOLD("%-13s%s") : " * %-13s%s", "ASSEMBLY", asmName(config->assembly(), config->isColors()));
+    }
+#   endif
 }
 
 

src/api/ApiRouter.cpp

@@ -35,12 +35,12 @@
 #include "api/ApiRouter.h"
 #include "common/api/HttpReply.h"
 #include "common/api/HttpRequest.h"
+#include "common/cpu/Cpu.h"
 #include "common/crypto/keccak.h"
 #include "common/net/Job.h"
 #include "common/Platform.h"
 #include "core/Config.h"
 #include "core/Controller.h"
-#include "Cpu.h"
 #include "interfaces/IThread.h"
 #include "rapidjson/document.h"
 #include "rapidjson/prettywriter.h"
@@ -238,13 +238,14 @@ void ApiRouter::getIdentify(rapidjson::Document &doc) const
 
 void ApiRouter::getMiner(rapidjson::Document &doc) const
 {
+    using namespace xmrig;
     auto &allocator = doc.GetAllocator();
 
     rapidjson::Value cpu(rapidjson::kObjectType);
-    cpu.AddMember("brand",   rapidjson::StringRef(Cpu::brand()), allocator);
+    cpu.AddMember("brand",   rapidjson::StringRef(Cpu::info()->brand()), allocator);
-    cpu.AddMember("aes",     Cpu::hasAES(), allocator);
+    cpu.AddMember("aes",     Cpu::info()->hasAES(), allocator);
-    cpu.AddMember("x64",     Cpu::isX64(), allocator);
+    cpu.AddMember("x64",     Cpu::info()->isX64(), allocator);
-    cpu.AddMember("sockets", Cpu::sockets(), allocator);
+    cpu.AddMember("sockets", Cpu::info()->sockets(), allocator);
 
     doc.AddMember("version",      APP_VERSION, allocator);
     doc.AddMember("kind",         APP_KIND, allocator);

src/common/cpu/BasicCpuInfo.cpp

@@ -21,6 +21,9 @@
  *   along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
+#include <string.h>
+#include <thread>
+
 
 #ifdef _MSC_VER
 #   include <intrin.h>
@@ -32,14 +35,8 @@
 #   define bit_AES (1 << 25)
 #endif
 
-#ifndef bit_BMI2
-#   define bit_BMI2 (1 << 8)
-#endif
 
-#include <string.h>
+#include "common/cpu/BasicCpuInfo.h"
-
-
-#include "Cpu.h"
 
 
 #define VENDOR_ID                  (0)
@@ -96,43 +93,18 @@ static inline bool has_aes_ni()
 }
 
 
-static inline bool has_bmi2() {
+xmrig::BasicCpuInfo::BasicCpuInfo() :
-    int cpu_info[4] = { 0 };
+    m_aes(has_aes_ni()),
-    cpuid(EXTENDED_FEATURES, cpu_info);
+    m_brand(),
-
+    m_threads(std::thread::hardware_concurrency())
-    return (cpu_info[EBX_Reg] & bit_BMI2) != 0;
-}
-
-
-char Cpu::m_brand[64]   = { 0 };
-int Cpu::m_flags        = 0;
-int Cpu::m_l2_cache     = 0;
-int Cpu::m_l3_cache     = 0;
-int Cpu::m_sockets      = 1;
-int Cpu::m_totalCores   = 0;
-size_t Cpu::m_totalThreads = 0;
-
-
-size_t Cpu::optimalThreadsCount(size_t size, int maxCpuUsage)
-{
-    const size_t count = m_totalThreads / 2;
-    return count < 1 ? 1 : count;
-}
-
-
-void Cpu::initCommon()
 {
     cpu_brand_string(m_brand);
-
+}
-#   if defined(__x86_64__) || defined(_M_AMD64)
+
-    m_flags |= X86_64;
+
-#   endif
+size_t xmrig::BasicCpuInfo::optimalThreadsCount(size_t memSize, int maxCpuUsage) const
-
+{
-    if (has_aes_ni()) {
+    const size_t count = threads() / 2;
-        m_flags |= AES;
+
-    }
+    return count < 1 ? 1 : count;
-
-    if (has_bmi2()) {
-        m_flags |= BMI2;
-    }
 }

src/common/cpu/BasicCpuInfo.h

@@ -0,0 +1,70 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2016-2018 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *
+ *   This program 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 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program 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.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef XMRIG_BASICCPUINFO_H
+#define XMRIG_BASICCPUINFO_H
+
+
+#include "common/interfaces/ICpuInfo.h"
+
+
+namespace xmrig {
+
+
+class BasicCpuInfo : public ICpuInfo
+{
+public:
+    BasicCpuInfo();
+
+protected:
+    size_t optimalThreadsCount(size_t memSize, int maxCpuUsage) const override;
+
+    inline Assembly assembly() const override       { return ASM_NONE; }
+    inline bool hasAES() const override             { return m_aes; }
+    inline bool isSupported() const override        { return true; }
+    inline const char *brand() const override       { return m_brand; }
+    inline int32_t cores() const override           { return -1; }
+    inline int32_t L2() const override              { return -1; }
+    inline int32_t L3() const override              { return -1; }
+    inline int32_t nodes() const override           { return -1; }
+    inline int32_t sockets() const override         { return 1; }
+    inline int32_t threads() const override         { return m_threads; }
+
+#   if defined(__x86_64__) || defined(_M_AMD64) || defined (__arm64__) || defined (__aarch64__)
+    inline bool isX64() const override { return true; }
+#   else
+    inline bool isX64() const override { return false; }
+#   endif
+
+private:
+    bool m_aes;
+    char m_brand[64];
+    int32_t m_threads;
+};
+
+
+} /* namespace xmrig */
+
+
+#endif /* XMRIG_BASICCPUINFO_H */

src/common/cpu/BasicCpuInfo_arm.cpp

@@ -21,37 +21,27 @@
  *   along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-
 #include <string.h>
+#include <thread>
 
 
-#include "Cpu.h"
+#include "common/cpu/BasicCpuInfo.h"
 
 
-char Cpu::m_brand[64]      = { 0 };
+xmrig::BasicCpuInfo::BasicCpuInfo() :
-int Cpu::m_flags           = 0;
+    m_aes(false),
-int Cpu::m_l2_cache        = 0;
+    m_brand(),
-int Cpu::m_l3_cache        = 0;
+    m_threads(std::thread::hardware_concurrency())
-int Cpu::m_sockets         = 1;
-int Cpu::m_totalCores      = 0;
-size_t Cpu::m_totalThreads = 0;
-
-
-size_t Cpu::optimalThreadsCount(size_t size, int maxCpuUsage)
-{
-    return m_totalThreads;
-}
-
-
-void Cpu::initCommon()
 {
     memcpy(m_brand, "Unknown", 7);
 
-#   if defined (__arm64__) || defined (__aarch64__)
-    m_flags |= X86_64;
-#   endif
-
 #   if __ARM_FEATURE_CRYPTO
-    m_flags |= AES;
+    m_aes = true;
 #   endif
 }
+
+
+size_t xmrig::BasicCpuInfo::optimalThreadsCount(size_t memSize, int maxCpuUsage) const
+{
+    return threads();
+}

src/common/cpu/Cpu.cpp

@@ -22,33 +22,36 @@
  */
 
 
-#ifdef __FreeBSD__
+#include <assert.h>
-#   include <sys/types.h>
-#   include <sys/param.h>
-#   include <sys/cpuset.h>
-#   include <pthread_np.h>
-#endif
 
 
-#include <pthread.h>
+#include "common/cpu/BasicCpuInfo.h"
-#include <sched.h>
+#include "common/cpu/Cpu.h"
-#include <unistd.h>
-#include <string.h>
 
 
-#include "Cpu.h"
+static xmrig::ICpuInfo *cpuInfo = nullptr;
 
 
-#ifdef __FreeBSD__
+xmrig::ICpuInfo *xmrig::Cpu::info()
-typedef cpuset_t cpu_set_t;
-#endif
-
-
-void Cpu::init()
 {
-#   ifdef XMRIG_NO_LIBCPUID
+    assert(cpuInfo != nullptr);
-    m_totalThreads = sysconf(_SC_NPROCESSORS_CONF);
-#   endif
 
-    initCommon();
+    return cpuInfo;
+}
+
+
+void xmrig::Cpu::init()
+{
+    assert(cpuInfo == nullptr);
+
+    cpuInfo = new BasicCpuInfo();
+}
+
+
+void xmrig::Cpu::release()
+{
+    assert(cpuInfo != nullptr);
+
+    delete cpuInfo;
+    cpuInfo = nullptr;
 }

src/common/cpu/Cpu.h

@@ -4,7 +4,7 @@
  * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
  * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
  * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
- * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
  * Copyright 2016-2018 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
  *
  *   This program is free software: you can redistribute it and/or modify
@@ -21,20 +21,26 @@
  *   along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-
+#ifndef XMRIG_CPU_H
-#include <pthread.h>
+#define XMRIG_CPU_H
-#include <sched.h>
-#include <unistd.h>
 
 
-#include "Cpu.h"
+#include "common/interfaces/ICpuInfo.h"
 
 
-void Cpu::init()
+namespace xmrig {
+
+
+class Cpu
 {
-#   ifdef XMRIG_NO_LIBCPUID
+public:
-    m_totalThreads = sysconf(_SC_NPROCESSORS_CONF);
+    static ICpuInfo *info();
-#   endif
+    static void init();
+    static void release();
+};
 
-    initCommon();
+
-}
+} /* namespace xmrig */
+
+
+#endif /* XMRIG_CPU_H */

src/common/interfaces/IConfig.h

@@ -80,6 +80,7 @@ public:
         SafeKey           = 1005,
         ThreadsKey        = 't',
         HardwareAESKey    = 1011,
+        AssemblyKey       = 1015,
 
         // xmrig amd
         OclPlatformKey    = 1400,

src/common/interfaces/ICpuInfo.h

@@ -0,0 +1,60 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2016-2018 XMRig       <support@xmrig.com>
+ *
+ *   This program 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 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program 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.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef XMRIG_CPUINFO_H
+#define XMRIG_CPUINFO_H
+
+
+#include <stddef.h>
+#include <stdint.h>
+
+
+#include "common/xmrig.h"
+
+
+namespace xmrig {
+
+
+class ICpuInfo
+{
+public:
+    virtual ~ICpuInfo() {}
+
+    virtual bool hasAES() const                                               = 0;
+    virtual bool isSupported() const                                          = 0;
+    virtual bool isX64() const                                                = 0;
+    virtual const char *brand() const                                         = 0;
+    virtual int32_t cores() const                                             = 0;
+    virtual int32_t L2() const                                                = 0;
+    virtual int32_t L3() const                                                = 0;
+    virtual int32_t nodes() const                                             = 0;
+    virtual int32_t sockets() const                                           = 0;
+    virtual int32_t threads() const                                           = 0;
+    virtual size_t optimalThreadsCount(size_t memSize, int maxCpuUsage) const = 0;
+    virtual xmrig::Assembly assembly() const                                  = 0;
+};
+
+
+} /* namespace xmrig */
+
+
+#endif // XMRIG_CPUINFO_H

src/common/xmrig.h

@@ -94,6 +94,15 @@ enum OclVendor {
 };
 
 
+enum Assembly {
+    ASM_NONE,
+    ASM_AUTO,
+    ASM_INTEL,
+    ASM_RYZEN,
+    ASM_MAX
+};
+
+
 } /* namespace xmrig */
 
 

src/core/Config.cpp

@@ -27,9 +27,10 @@
 
 
 #include "common/config/ConfigLoader.h"
+#include "common/cpu/Cpu.h"
 #include "core/Config.h"
 #include "core/ConfigCreator.h"
-#include "Cpu.h"
+#include "crypto/Asm.h"
 #include "crypto/CryptoNight_constants.h"
 #include "rapidjson/document.h"
 #include "rapidjson/filewritestream.h"
@@ -43,6 +44,7 @@ static char affinity_tmp[20] = { 0 };
 xmrig::Config::Config() : xmrig::CommonConfig(),
     m_aesMode(AES_AUTO),
     m_algoVariant(AV_AUTO),
+    m_assembly(ASM_AUTO),
     m_hugePages(true),
     m_safe(false),
     m_maxCpuUsage(75),
@@ -51,11 +53,6 @@ xmrig::Config::Config() : xmrig::CommonConfig(),
 }
 
 
-xmrig::Config::~Config()
-{
-}
-
-
 bool xmrig::Config::reload(const char *json)
 {
     return xmrig::ConfigLoader::reload(this, json);
@@ -153,7 +150,7 @@ bool xmrig::Config::finalize()
 
     if (!m_threads.cpu.empty()) {
         m_threads.mode     = Advanced;
-        const bool softAES = (m_aesMode == AES_AUTO ? (Cpu::hasAES() ? AES_HW : AES_SOFT) : m_aesMode) == AES_SOFT;
+        const bool softAES = (m_aesMode == AES_AUTO ? (Cpu::info()->hasAES() ? AES_HW : AES_SOFT) : m_aesMode) == AES_SOFT;
 
         for (size_t i = 0; i < m_threads.cpu.size(); ++i) {
             m_threads.list.push_back(CpuThread::createFromData(i, m_algorithm.algo(), m_threads.cpu[i], m_priority, softAES));
@@ -168,17 +165,17 @@ bool xmrig::Config::finalize()
     const size_t size = CpuThread::multiway(av) * cn_select_memory(m_algorithm.algo()) / 1024;
 
     if (!m_threads.count) {
-        m_threads.count = Cpu::optimalThreadsCount(size, m_maxCpuUsage);
+        m_threads.count = Cpu::info()->optimalThreadsCount(size, m_maxCpuUsage);
     }
     else if (m_safe) {
-        const size_t count = Cpu::optimalThreadsCount(size, m_maxCpuUsage);
+        const size_t count = Cpu::info()->optimalThreadsCount(size, m_maxCpuUsage);
         if (m_threads.count > count) {
             m_threads.count = count;
         }
     }
 
     for (size_t i = 0; i < m_threads.count; ++i) {
-        m_threads.list.push_back(CpuThread::createFromAV(i, m_algorithm.algo(), av, m_threads.mask, m_priority));
+        m_threads.list.push_back(CpuThread::createFromAV(i, m_algorithm.algo(), av, m_threads.mask, m_priority, m_assembly));
     }
 
     return true;
@@ -204,6 +201,12 @@ bool xmrig::Config::parseBoolean(int key, bool enable)
         m_aesMode = enable ? AES_HW : AES_SOFT;
         break;
 
+#   ifndef XMRIG_NO_ASM
+    case AssemblyKey:
+        m_assembly = Asm::parse(enable);
+        break;
+#   endif
+
     default:
         break;
     }
@@ -232,7 +235,7 @@ bool xmrig::Config::parseString(int key, const char *arg)
 
     case ThreadsKey:  /* --threads */
         if (strncmp(arg, "all", 3) == 0) {
-            m_threads.count = Cpu::threads();
+            m_threads.count = Cpu::info()->threads();
             return true;
         }
 
@@ -244,6 +247,12 @@ bool xmrig::Config::parseString(int key, const char *arg)
             return parseUint64(key, p ? strtoull(p, nullptr, 16) : strtoull(arg, nullptr, 10));
         }
 
+#   ifndef XMRIG_NO_ASM
+    case AssemblyKey: /* --asm */
+        m_assembly = Asm::parse(arg);
+        break;
+#   endif
+
     default:
         break;
     }
@@ -339,10 +348,10 @@ xmrig::AlgoVariant xmrig::Config::getAlgoVariant() const
 #   endif
 
     if (m_algoVariant <= AV_AUTO || m_algoVariant >= AV_MAX) {
-        return Cpu::hasAES() ? AV_SINGLE : AV_SINGLE_SOFT;
+        return Cpu::info()->hasAES() ? AV_SINGLE : AV_SINGLE_SOFT;
     }
 
-    if (m_safe && !Cpu::hasAES() && m_algoVariant <= AV_DOUBLE) {
+    if (m_safe && !Cpu::info()->hasAES() && m_algoVariant <= AV_DOUBLE) {
         return static_cast<AlgoVariant>(m_algoVariant + 2);
     }
 
@@ -354,10 +363,10 @@ xmrig::AlgoVariant xmrig::Config::getAlgoVariant() const
 xmrig::AlgoVariant xmrig::Config::getAlgoVariantLite() const
 {
     if (m_algoVariant <= AV_AUTO || m_algoVariant >= AV_MAX) {
-        return Cpu::hasAES() ? AV_DOUBLE : AV_DOUBLE_SOFT;
+        return Cpu::info()->hasAES() ? AV_DOUBLE : AV_DOUBLE_SOFT;
     }
 
-    if (m_safe && !Cpu::hasAES() && m_algoVariant <= AV_DOUBLE) {
+    if (m_safe && !Cpu::info()->hasAES() && m_algoVariant <= AV_DOUBLE) {
         return static_cast<AlgoVariant>(m_algoVariant + 2);
     }
 

src/core/Config.h

@@ -21,8 +21,8 @@
  *   along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#ifndef __CONFIG_H__
+#ifndef XMRIG_CONFIG_H
-#define __CONFIG_H__
+#define XMRIG_CONFIG_H
 
 
 #include <stdint.h>
@@ -69,7 +69,6 @@ public:
 
 
     Config();
-    ~Config();
 
     bool reload(const char *json);
 
@@ -77,6 +76,7 @@ public:
 
     inline AesMode aesMode() const                       { return m_aesMode; }
     inline AlgoVariant algoVariant() const               { return m_algoVariant; }
+    inline Assembly assembly() const                     { return m_assembly; }
     inline bool isHugePages() const                      { return m_hugePages; }
     inline const std::vector<IThread *> &threads() const { return m_threads.list; }
     inline int priority() const                          { return m_priority; }
@@ -116,6 +116,7 @@ private:
 
     AesMode m_aesMode;
     AlgoVariant m_algoVariant;
+    Assembly m_assembly;
     bool m_hugePages;
     bool m_safe;
     int m_maxCpuUsage;
@@ -126,4 +127,4 @@ private:
 
 } /* namespace xmrig */
 
-#endif /* __CONFIG_H__ */
+#endif /* XMRIG_CONFIG_H */

src/core/ConfigLoader_platform.h

@@ -135,6 +135,7 @@ static struct option const options[] = {
     { "tls",               0, nullptr, xmrig::IConfig::TlsKey            },
     { "tls-fingerprint",   1, nullptr, xmrig::IConfig::FingerprintKey    },
     { "version",           0, nullptr, xmrig::IConfig::VersionKey        },
+    { "asm",               1, nullptr, xmrig::IConfig::AssemblyKey       },
     { nullptr,             0, nullptr, 0 }
 };
 
@@ -159,6 +160,7 @@ static struct option const config_options[] = {
     { "threads",       1, nullptr, xmrig::IConfig::ThreadsKey     },
     { "user-agent",    1, nullptr, xmrig::IConfig::UserAgentKey   },
     { "hw-aes",        0, nullptr, xmrig::IConfig::HardwareAESKey },
+    { "asm",           1, nullptr, xmrig::IConfig::AssemblyKey    },
     { nullptr,         0, nullptr, 0 }
 };
 

src/core/Controller.cpp

@@ -26,6 +26,7 @@
 
 
 #include "common/config/ConfigLoader.h"
+#include "common/cpu/Cpu.h"
 #include "common/interfaces/IControllerListener.h"
 #include "common/log/ConsoleLog.h"
 #include "common/log/FileLog.h"
@@ -33,7 +34,6 @@
 #include "common/Platform.h"
 #include "core/Config.h"
 #include "core/Controller.h"
-#include "Cpu.h"
 #include "net/Network.h"
 
 

src/core/cpu/AdvancedCpuInfo.cpp

@@ -21,65 +21,25 @@
  *   along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-
 #include <libcpuid.h>
 #include <math.h>
 #include <string.h>
+#include <thread>
 
 
-#include "Cpu.h"
+#include "core/cpu/AdvancedCpuInfo.h"
 
 
-bool Cpu::m_l2_exclusive    = false;
+xmrig::AdvancedCpuInfo::AdvancedCpuInfo() :
-char Cpu::m_brand[64]       = { 0 };
+    m_assembly(ASM_NONE),
-int Cpu::m_flags            = 0;
+    m_aes(false),
-int Cpu::m_l2_cache         = 0;
+    m_L2_exclusive(false),
-int Cpu::m_l3_cache         = 0;
+    m_brand(),
-int Cpu::m_sockets          = 1;
+    m_cores(0),
-int Cpu::m_totalCores       = 0;
+    m_L2(0),
-size_t Cpu::m_totalThreads  = 0;
+    m_L3(0),
-
+    m_sockets(1),
-
+    m_threads(std::thread::hardware_concurrency())
-size_t Cpu::optimalThreadsCount(size_t size, int maxCpuUsage)
-{
-    if (m_totalThreads == 1) {
-        return 1;
-    }
-
-    size_t cache = 0;
-    if (m_l3_cache) {
-        cache = m_l2_exclusive ? (m_l2_cache + m_l3_cache) : m_l3_cache;
-    }
-    else {
-        cache = m_l2_cache;
-    }
-
-    size_t count = 0;
-
-    if (cache) {
-        count = cache / size;
-
-        if (cache % size >= size / 2) {
-            count++;
-        }
-    }
-    else {
-        count = m_totalThreads / 2;
-    }
-
-    if (count > m_totalThreads) {
-        count = m_totalThreads;
-    }
-
-    if (((float) count / m_totalThreads * 100) > maxCpuUsage) {
-        count = (int) ceil((float) m_totalThreads * (maxCpuUsage / 100.0));
-    }
-
-    return count < 1 ? 1 : count;
-}
-
-
-void Cpu::initCommon()
 {
     struct cpu_raw_data_t raw = { 0 };
     struct cpu_id_t data = { 0 };
@@ -87,42 +47,78 @@ void Cpu::initCommon()
     cpuid_get_raw_data(&raw);
     cpu_identify(&raw, &data);
 
-    strncpy(m_brand, data.brand_str, sizeof(m_brand) - 1);
+    strncpy(m_brand, data.brand_str, sizeof(m_brand));
-
-    m_totalThreads = data.total_logical_cpus;
-    m_sockets      = m_totalThreads / data.num_logical_cpus;
 
+    m_sockets = threads() / data.num_logical_cpus;
     if (m_sockets == 0) {
         m_sockets = 1;
     }
 
-    m_totalCores = data.num_cores * m_sockets;
+    m_cores = data.num_cores * m_sockets;
-    m_l3_cache = data.l3_cache > 0 ? data.l3_cache * m_sockets : 0;
+    m_L3 = data.l3_cache > 0 ? data.l3_cache * m_sockets : 0;
 
     // Workaround for AMD CPUs https://github.com/anrieff/libcpuid/issues/97
     if (data.vendor == VENDOR_AMD && data.ext_family >= 0x15 && data.ext_family < 0x17) {
-        m_l2_cache = data.l2_cache * (m_totalCores / 2) * m_sockets;
+        m_L2 = data.l2_cache * (cores() / 2) * m_sockets;
-        m_l2_exclusive = true;
+        m_L2_exclusive = true;
     }
     // Workaround for Intel Pentium Dual-Core, Core Duo, Core 2 Duo, Core 2 Quad and their Xeon homologue
     // These processors have L2 cache shared by 2 cores.
     else if (data.vendor == VENDOR_INTEL && data.ext_family == 0x06 && (data.ext_model == 0x0E || data.ext_model == 0x0F || data.ext_model == 0x17)) {
-        int l2_count_per_socket = m_totalCores > 1 ? m_totalCores / 2 : 1;
+        int l2_count_per_socket = cores() > 1 ? cores() / 2 : 1;
-        m_l2_cache = data.l2_cache > 0 ? data.l2_cache * l2_count_per_socket * m_sockets : 0;
+        m_L2 = data.l2_cache > 0 ? data.l2_cache * l2_count_per_socket * m_sockets : 0;
     }
     else{
-        m_l2_cache = data.l2_cache > 0 ? data.l2_cache * m_totalCores * m_sockets : 0;
+        m_L2 = data.l2_cache > 0 ? data.l2_cache * cores() * m_sockets : 0;
     }
 
-#   if defined(__x86_64__) || defined(_M_AMD64)
-    m_flags |= X86_64;
-#   endif
-
     if (data.flags[CPU_FEATURE_AES]) {
-        m_flags |= AES;
+        m_aes = true;
-    }
 
-    if (data.flags[CPU_FEATURE_BMI2]) {
+        if (data.vendor == VENDOR_AMD && data.ext_family >= 23) {
-        m_flags |= BMI2;
+            m_assembly = ASM_RYZEN;
+        }
+        else if (data.vendor == VENDOR_INTEL && data.ext_model >= 42) {
+            m_assembly = ASM_INTEL;
+        }
     }
 }
+
+
+size_t xmrig::AdvancedCpuInfo::optimalThreadsCount(size_t memSize, int maxCpuUsage) const
+{
+    if (threads() == 1) {
+        return 1;
+    }
+
+    size_t cache = 0;
+    if (m_L3) {
+        cache = m_L2_exclusive ? (m_L2 + m_L3) : m_L3;
+    }
+    else {
+        cache = m_L2;
+    }
+
+    size_t count = 0;
+
+    if (cache) {
+        count = cache / memSize;
+
+        if (cache % memSize >= memSize / 2) {
+            count++;
+        }
+    }
+    else {
+        count = threads() / 2;
+    }
+
+    if (count > (size_t) threads()) {
+        count = threads();
+    }
+
+    if (((float) count / threads() * 100) > maxCpuUsage) {
+        count = (int) ceil((float) threads() * (maxCpuUsage / 100.0));
+    }
+
+    return count < 1 ? 1 : count;
+}

src/core/cpu/AdvancedCpuInfo.h

@@ -0,0 +1,75 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2016-2018 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   This program 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 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program 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.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef XMRIG_ADVANCEDCPUINFO_H
+#define XMRIG_ADVANCEDCPUINFO_H
+
+
+#include "common/interfaces/ICpuInfo.h"
+
+
+namespace xmrig {
+
+
+class AdvancedCpuInfo : public ICpuInfo
+{
+public:
+    AdvancedCpuInfo();
+
+protected:
+    size_t optimalThreadsCount(size_t memSize, int maxCpuUsage) const override;
+
+    inline Assembly assembly() const override       { return m_assembly; }
+    inline bool hasAES() const override             { return m_aes; }
+    inline bool isSupported() const override        { return true; }
+    inline const char *brand() const override       { return m_brand; }
+    inline int32_t cores() const override           { return m_cores; }
+    inline int32_t L2() const override              { return m_L2; }
+    inline int32_t L3() const override              { return m_L3; }
+    inline int32_t nodes() const override           { return -1; }
+    inline int32_t sockets() const override         { return m_sockets; }
+    inline int32_t threads() const override         { return m_threads; }
+
+#   if defined(__x86_64__) || defined(_M_AMD64)
+    inline bool isX64() const override { return true; }
+#   else
+    inline bool isX64() const override { return false; }
+#   endif
+
+private:
+    Assembly m_assembly;
+    bool m_aes;
+    bool m_L2_exclusive;
+    char m_brand[64];
+    int32_t m_cores;
+    int32_t m_L2;
+    int32_t m_L3;
+    int32_t m_sockets;
+    int32_t m_threads;
+};
+
+
+} /* namespace xmrig */
+
+
+#endif /* XMRIG_ADVANCEDCPUINFO_H */

src/core/cpu/Cpu.cpp

@@ -22,20 +22,40 @@
  */
 
 
-#include <windows.h>
+#include <assert.h>
 
 
-#include "Cpu.h"
+#include "common/cpu/Cpu.h"
 
 
-void Cpu::init()
+#ifndef XMRIG_NO_LIBCPUID
+#   include "core/cpu/AdvancedCpuInfo.h"
+#endif
+
+
+static xmrig::ICpuInfo *cpuInfo = nullptr;
+
+
+xmrig::ICpuInfo *xmrig::Cpu::info()
 {
-#   ifdef XMRIG_NO_LIBCPUID
+    assert(cpuInfo != nullptr);
-    SYSTEM_INFO sysinfo;
-    GetSystemInfo(&sysinfo);
 
-    m_totalThreads = sysinfo.dwNumberOfProcessors;
+    return cpuInfo;
-#   endif
+}
-
+
-    initCommon();
+
+void xmrig::Cpu::init()
+{
+    assert(cpuInfo == nullptr);
+
+    cpuInfo = new AdvancedCpuInfo();
+}
+
+
+void xmrig::Cpu::release()
+{
+    assert(cpuInfo != nullptr);
+
+    delete cpuInfo;
+    cpuInfo = nullptr;
 }

src/crypto/Asm.cpp

@@ -0,0 +1,100 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2016-2018 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   This program 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 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program 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.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+
+#include <assert.h>
+#include <string.h>
+
+
+#ifdef _MSC_VER
+#   define strncasecmp _strnicmp
+#   define strcasecmp  _stricmp
+#endif
+
+
+#include "crypto/Asm.h"
+#include "rapidjson/document.h"
+
+
+static const char *asmNames[] = {
+    "none",
+    "auto",
+    "intel",
+    "ryzen"
+};
+
+
+xmrig::Assembly xmrig::Asm::parse(const char *assembly, Assembly defaultValue)
+{
+    constexpr size_t const size = sizeof(asmNames) / sizeof((asmNames)[0]);
+    assert(assembly != nullptr);
+    assert(ASM_MAX == size);
+
+    if (assembly == nullptr) {
+        return defaultValue;
+    }
+
+    for (size_t i = 0; i < size; i++) {
+        if (strcasecmp(assembly, asmNames[i]) == 0) {
+            return static_cast<Assembly>(i);
+        }
+    }
+
+    return defaultValue;
+}
+
+
+xmrig::Assembly xmrig::Asm::parse(const rapidjson::Value &value, Assembly defaultValue)
+{
+    if (value.IsBool()) {
+        return parse(value.IsBool());
+    }
+
+    if (value.IsString()) {
+        return parse(value.GetString(), defaultValue);
+    }
+
+    return defaultValue;
+}
+
+
+const char *xmrig::Asm::toString(Assembly assembly)
+{
+    return asmNames[assembly];
+}
+
+
+rapidjson::Value xmrig::Asm::toJSON(Assembly assembly)
+{
+    using namespace rapidjson;
+
+    if (assembly == ASM_NONE) {
+        return Value(false);
+    }
+
+    if (assembly == ASM_AUTO) {
+        return Value(true);
+    }
+
+    return Value(StringRef(toString(assembly)));
+}

src/crypto/Asm.h

@@ -0,0 +1,50 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2016-2018 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   This program 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 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program 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.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef XMRIG_ASM_H
+#define XMRIG_ASM_H
+
+
+#include "common/xmrig.h"
+#include "rapidjson/fwd.h"
+
+
+namespace xmrig {
+
+
+class Asm
+{
+public:
+    static Assembly parse(const char *assembly, Assembly defaultValue = ASM_AUTO);
+    static Assembly parse(const rapidjson::Value &value, Assembly defaultValue = ASM_AUTO);
+    static const char *toString(Assembly assembly);
+    static rapidjson::Value toJSON(Assembly assembly);
+
+    inline static Assembly parse(bool enable) { return enable ? ASM_AUTO : ASM_NONE; }
+};
+
+
+} /* namespace xmrig */
+
+
+#endif /* XMRIG_ASM_H */

src/crypto/CryptoNight.h

@@ -22,8 +22,8 @@
  *   along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#ifndef __CRYPTONIGHT_H__
+#ifndef XMRIG_CRYPTONIGHT_H
-#define __CRYPTONIGHT_H__
+#define XMRIG_CRYPTONIGHT_H
 
 
 #include <stddef.h>
@@ -31,9 +31,9 @@
 
 
 struct cryptonight_ctx {
-    alignas(16) uint8_t state[200];
+    alignas(16) uint8_t state[224];
-    alignas(16) uint8_t* memory;
+    alignas(16) uint8_t *memory;
 };
 
 
-#endif /* __CRYPTONIGHT_H__ */
+#endif /* XMRIG_CRYPTONIGHT_H */

src/crypto/CryptoNight_x86.h

@@ -561,6 +561,33 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si
 }
 
 
+#ifndef XMRIG_NO_ASM
+extern "C" void cnv2_mainloop_ivybridge_asm(cryptonight_ctx *ctx);
+extern "C" void cnv2_mainloop_ryzen_asm(cryptonight_ctx *ctx);
+
+
+template<xmrig::Algo ALGO, xmrig::Variant VARIANT, xmrig::Assembly ASM>
+inline void cryptonight_single_hash_asm(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, cryptonight_ctx **__restrict__ ctx)
+{
+    constexpr size_t MEM = xmrig::cn_select_memory<ALGO>();
+
+    xmrig::keccak(input, size, ctx[0]->state);
+    cn_explode_scratchpad<ALGO, MEM, false>((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    if (ASM == xmrig::ASM_INTEL) {
+        cnv2_mainloop_ivybridge_asm(ctx[0]);
+    }
+    else {
+        cnv2_mainloop_ryzen_asm(ctx[0]);
+    }
+
+    cn_implode_scratchpad<ALGO, MEM, false>((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+    xmrig::keccakf(reinterpret_cast<uint64_t*>(ctx[0]->state), 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}
+#endif
+
+
 template<xmrig::Algo ALGO, bool SOFT_AES, xmrig::Variant VARIANT>
 inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, cryptonight_ctx **__restrict__ ctx)
 {

src/crypto/asm/cnv2_main_loop.S

@@ -0,0 +1,27 @@
+#define ALIGN .align
+.intel_syntax noprefix
+#ifdef __APPLE__
+#   define FN_PREFIX(fn) _ ## fn
+.text
+#else
+#   define FN_PREFIX(fn) fn
+.section .text
+#endif
+.global FN_PREFIX(cnv2_mainloop_ivybridge_asm)
+.global FN_PREFIX(cnv2_mainloop_ryzen_asm)
+
+ALIGN 16
+FN_PREFIX(cnv2_mainloop_ivybridge_asm):
+	sub rsp, 48
+	mov rcx, rdi
+	#include "cnv2_main_loop_ivybridge.inc"
+	add rsp, 48
+	ret 0
+
+ALIGN 16
+FN_PREFIX(cnv2_mainloop_ryzen_asm):
+	sub rsp, 48
+	mov rcx, rdi
+	#include "cnv2_main_loop_ryzen.inc"
+	add rsp, 48
+	ret 0

src/crypto/asm/cnv2_main_loop.asm

@@ -0,0 +1,18 @@
+_TEXT_CNV2_MAINLOOP SEGMENT PAGE READ EXECUTE
+PUBLIC cnv2_mainloop_ivybridge_asm
+PUBLIC cnv2_mainloop_ryzen_asm
+
+ALIGN 64
+cnv2_mainloop_ivybridge_asm PROC
+	INCLUDE cnv2_main_loop_ivybridge.inc
+	ret 0
+cnv2_mainloop_ivybridge_asm ENDP
+
+ALIGN 64
+cnv2_mainloop_ryzen_asm PROC
+	INCLUDE cnv2_main_loop_ryzen.inc
+	ret 0
+cnv2_mainloop_ryzen_asm ENDP
+
+_TEXT_CNV2_MAINLOOP ENDS
+END

src/crypto/asm/cnv2_main_loop_ivybridge.inc

@@ -0,0 +1,186 @@
+	mov	 QWORD PTR [rsp+24], rbx
+	push	 rbp
+	push	 rsi
+	push	 rdi
+	push	 r12
+	push	 r13
+	push	 r14
+	push	 r15
+	sub	 rsp, 80
+
+	stmxcsr DWORD PTR [rsp]
+	mov DWORD PTR [rsp+4], 24448
+	ldmxcsr DWORD PTR [rsp+4]
+
+	mov	 rax, QWORD PTR [rcx+48]
+	mov	 r9, rcx
+	xor	 rax, QWORD PTR [rcx+16]
+	mov	 esi, 524288
+	mov	 r8, QWORD PTR [rcx+32]
+	mov	 r13d, -2147483647
+	xor	 r8, QWORD PTR [rcx]
+	mov	 r11, QWORD PTR [rcx+40]
+	mov	 r10, r8
+	mov	 rdx, QWORD PTR [rcx+56]
+	movq	 xmm4, rax
+	xor	 rdx, QWORD PTR [rcx+24]
+	xor	 r11, QWORD PTR [rcx+8]
+	mov	 rbx, QWORD PTR [rcx+224]
+	mov	 rax, QWORD PTR [r9+80]
+	xor	 rax, QWORD PTR [r9+64]
+	movq	 xmm0, rdx
+	mov	 rcx, QWORD PTR [rcx+88]
+	xor	 rcx, QWORD PTR [r9+72]
+	movq	 xmm3, QWORD PTR [r9+104]
+	movaps	 XMMWORD PTR [rsp+64], xmm6
+	movaps	 XMMWORD PTR [rsp+48], xmm7
+	movaps	 XMMWORD PTR [rsp+32], xmm8
+	and	 r10d, 2097136
+	movq	 xmm5, rax
+
+	xor eax, eax
+	mov QWORD PTR [rsp+16], rax
+
+	mov ax, 1023
+	shl rax, 52
+	movq xmm8, rax
+	mov r15, QWORD PTR [r9+96]
+	punpcklqdq xmm4, xmm0
+	movq	 xmm0, rcx
+	punpcklqdq xmm5, xmm0
+	movdqu	 xmm6, XMMWORD PTR [r10+rbx]
+
+	ALIGN 16
+main_loop_ivybridge:
+	lea	 rdx, QWORD PTR [r10+rbx]
+	mov	 ecx, r10d
+	mov	 eax, r10d
+	mov rdi, r15
+	xor	 ecx, 16
+	xor	 eax, 32
+	xor	 r10d, 48
+	movq	 xmm0, r11
+	movq	 xmm7, r8
+	punpcklqdq xmm7, xmm0
+	aesenc	 xmm6, xmm7
+	movq	 rbp, xmm6
+	mov	 r9, rbp
+	and	 r9d, 2097136
+	movdqu	 xmm2, XMMWORD PTR [rcx+rbx]
+	movdqu	 xmm1, XMMWORD PTR [rax+rbx]
+	movdqu	 xmm0, XMMWORD PTR [r10+rbx]
+	paddq	 xmm1, xmm7
+	paddq	 xmm0, xmm5
+	paddq	 xmm2, xmm4
+	movdqu	 XMMWORD PTR [rcx+rbx], xmm0
+	movdqu	 XMMWORD PTR [rax+rbx], xmm2
+	movdqu	 XMMWORD PTR [r10+rbx], xmm1
+	mov r10, r9
+	xor r10d, 32
+	movq	 rcx, xmm3
+	mov	 rax, rcx
+	shl	 rax, 32
+	xor	 rdi, rax
+	movdqa	 xmm0, xmm6
+	pxor	 xmm0, xmm4
+	movdqu	 XMMWORD PTR [rdx], xmm0
+	xor	 rdi, QWORD PTR [r9+rbx]
+	lea	 r14, QWORD PTR [r9+rbx]
+	mov	 r12, QWORD PTR [r14+8]
+	xor	 edx, edx
+	lea	 r9d, DWORD PTR [ecx+ecx]
+	add	 r9d, ebp
+	movdqa	 xmm0, xmm6
+	psrldq	 xmm0, 8
+	or	 r9d, r13d
+	movq	 rax, xmm0
+	div	 r9
+	xorps xmm3, xmm3
+	mov	 eax, eax
+	shl	 rdx, 32
+	add	 rdx, rax
+	lea	 r9, QWORD PTR [rdx+rbp]
+	mov r15, rdx
+	mov	 rax, r9
+	shr	 rax, 12
+	movq	 xmm0, rax
+	paddq	 xmm0, xmm8
+	sqrtsd	 xmm3, xmm0
+	psubq	 xmm3, XMMWORD PTR [rsp+16]
+	movq	 rdx, xmm3
+	test	 edx, 524287
+	je	 sqrt_fixup_ivybridge
+	psrlq	 xmm3, 19
+sqrt_fixup_ivybridge_ret:
+
+	mov	 ecx, r10d
+	mov	 rax, rdi
+	mul	 rbp
+	movq xmm2, rdx
+	xor rdx, [rcx+rbx]
+	add	 r8, rdx
+	mov	 QWORD PTR [r14], r8
+	xor	 r8, rdi
+	mov edi, r8d
+	and edi, 2097136
+	movq xmm0, rax
+	xor rax, [rcx+rbx+8]
+	add	 r11, rax
+	mov	 QWORD PTR [r14+8], r11
+	punpcklqdq xmm2, xmm0
+
+	mov	 r9d, r10d
+	xor	 r9d, 48
+	xor	 r10d, 16
+	pxor	 xmm2, XMMWORD PTR [r9+rbx]
+	movdqu	 xmm0, XMMWORD PTR [r10+rbx]
+	paddq	 xmm0, xmm5
+	movdqu	 xmm1, XMMWORD PTR [rcx+rbx]
+	paddq	 xmm2, xmm4
+	paddq	 xmm1, xmm7
+	movdqa	 xmm5, xmm4
+	movdqu	 XMMWORD PTR [r9+rbx], xmm0
+	movdqa	 xmm4, xmm6
+	movdqu	 XMMWORD PTR [rcx+rbx], xmm2
+	movdqu	 XMMWORD PTR [r10+rbx], xmm1
+	movdqu xmm6, [rdi+rbx]
+	mov	 r10d, edi
+	xor	 r11, r12
+	dec rsi
+	jne	 main_loop_ivybridge
+
+	ldmxcsr DWORD PTR [rsp]
+	mov	 rbx, QWORD PTR [rsp+160]
+	movaps	 xmm6, XMMWORD PTR [rsp+64]
+	movaps	 xmm7, XMMWORD PTR [rsp+48]
+	movaps	 xmm8, XMMWORD PTR [rsp+32]
+	add	 rsp, 80
+	pop	 r15
+	pop	 r14
+	pop	 r13
+	pop	 r12
+	pop	 rdi
+	pop	 rsi
+	pop	 rbp
+	jmp cnv2_main_loop_ivybridge_endp
+
+sqrt_fixup_ivybridge:
+	dec	 rdx
+	mov r13d, -1022
+	shl r13, 32
+	mov	 rax, rdx
+	shr	 rdx, 19
+	shr	 rax, 20
+	mov	 rcx, rdx
+	sub	 rcx, rax
+	add	 rax, r13
+	not r13
+	sub	 rcx, r13
+	mov	 r13d, -2147483647
+	imul	 rcx, rax
+	sub	 rcx, r9
+	adc	 rdx, 0
+	movq	 xmm3, rdx
+	jmp	 sqrt_fixup_ivybridge_ret
+
+cnv2_main_loop_ivybridge_endp:

src/crypto/asm/cnv2_main_loop_ryzen.inc

@@ -0,0 +1,179 @@
+	mov	QWORD PTR [rsp+16], rbx
+	mov	QWORD PTR [rsp+24], rbp
+	mov	QWORD PTR [rsp+32], rsi
+	push	rdi
+	push	r12
+	push	r13
+	push	r14
+	push	r15
+	sub	rsp, 64
+
+	stmxcsr DWORD PTR [rsp]
+	mov DWORD PTR [rsp+4], 24448
+	ldmxcsr DWORD PTR [rsp+4]
+
+	mov	rax, QWORD PTR [rcx+48]
+	mov	r9, rcx
+	xor	rax, QWORD PTR [rcx+16]
+	mov	ebp, 524288
+	mov	r8, QWORD PTR [rcx+32]
+	xor	r8, QWORD PTR [rcx]
+	mov	r11, QWORD PTR [rcx+40]
+	mov	r10, r8
+	mov	rdx, QWORD PTR [rcx+56]
+	movq	xmm3, rax
+	xor	rdx, QWORD PTR [rcx+24]
+	xor	r11, QWORD PTR [rcx+8]
+	mov	rbx, QWORD PTR [rcx+224]
+	mov	rax, QWORD PTR [r9+80]
+	xor	rax, QWORD PTR [r9+64]
+	movq	xmm0, rdx
+	mov	rcx, QWORD PTR [rcx+88]
+	xor	rcx, QWORD PTR [r9+72]
+	mov	rdi, QWORD PTR [r9+104]
+	and	r10d, 2097136
+	movaps	XMMWORD PTR [rsp+48], xmm6
+	movq	xmm4, rax
+	movaps	XMMWORD PTR [rsp+32], xmm7
+	movaps	XMMWORD PTR [rsp+16], xmm8
+	xorps	xmm8, xmm8
+	mov ax, 1023
+	shl rax, 52
+	movq xmm7, rax
+	mov	r15, QWORD PTR [r9+96]
+	punpcklqdq xmm3, xmm0
+	movq	xmm0, rcx
+	punpcklqdq xmm4, xmm0
+
+	ALIGN 16
+main_loop_ryzen:
+	movdqa	xmm5, XMMWORD PTR [r10+rbx]
+	movq	xmm0, r11
+	movq	xmm6, r8
+	punpcklqdq xmm6, xmm0
+	lea	rdx, QWORD PTR [r10+rbx]
+	lea	r9, QWORD PTR [rdi+rdi]
+	shl	rdi, 32
+
+	mov	ecx, r10d
+	mov	eax, r10d
+	xor	ecx, 16
+	xor	eax, 32
+	xor	r10d, 48
+	aesenc	xmm5, xmm6
+	movdqa	xmm2, XMMWORD PTR [rcx+rbx]
+	movdqa	xmm1, XMMWORD PTR [rax+rbx]
+	movdqa	xmm0, XMMWORD PTR [r10+rbx]
+	paddq	xmm2, xmm3
+	paddq	xmm1, xmm6
+	paddq	xmm0, xmm4
+	movdqa	XMMWORD PTR [rcx+rbx], xmm0
+	movdqa	XMMWORD PTR [rax+rbx], xmm2
+	movdqa	XMMWORD PTR [r10+rbx], xmm1
+
+	movaps	xmm1, xmm8
+	mov	rsi, r15
+	xor	rsi, rdi
+	movq	r14, xmm5
+	movdqa	xmm0, xmm5
+	pxor	xmm0, xmm3
+	mov	r10, r14
+	and	r10d, 2097136
+	movdqa	XMMWORD PTR [rdx], xmm0
+	xor	rsi, QWORD PTR [r10+rbx]
+	lea	r12, QWORD PTR [r10+rbx]
+	mov	r13, QWORD PTR [r10+rbx+8]
+
+	add	r9d, r14d
+	or	r9d, -2147483647
+	xor	edx, edx
+	movdqa	xmm0, xmm5
+	psrldq	xmm0, 8
+	movq	rax, xmm0
+
+	div	r9
+	movq xmm0, rax
+	movq xmm1, rdx
+	punpckldq xmm0, xmm1
+	movq r15, xmm0
+	paddq xmm0, xmm5
+	movdqa xmm2, xmm0
+	psrlq xmm0, 12
+	paddq	xmm0, xmm7
+	sqrtsd	xmm1, xmm0
+	movq	rdi, xmm1
+	test	rdi, 524287
+	je	sqrt_fixup_ryzen
+	shr	rdi, 19
+
+sqrt_fixup_ryzen_ret:
+	mov	rax, rsi
+	mul	r14
+	movq xmm1, rax
+	movq xmm0, rdx
+	punpcklqdq xmm0, xmm1
+
+	mov	r9d, r10d
+	mov	ecx, r10d
+	xor	r9d, 16
+	xor	ecx, 32
+	xor	r10d, 48
+	movdqa	xmm1, XMMWORD PTR [rcx+rbx]
+	xor rdx, [rcx+rbx]
+	xor rax, [rcx+rbx+8]
+	movdqa	xmm2, XMMWORD PTR [r9+rbx]
+	pxor xmm2, xmm0
+	paddq xmm4, XMMWORD PTR [r10+rbx]
+	paddq	xmm2, xmm3
+	paddq	xmm1, xmm6
+	movdqa	XMMWORD PTR [r9+rbx], xmm4
+	movdqa	XMMWORD PTR [rcx+rbx], xmm2
+	movdqa	XMMWORD PTR [r10+rbx], xmm1
+
+	movdqa	xmm4, xmm3
+	add	r8, rdx
+	add	r11, rax
+	mov	QWORD PTR [r12], r8
+	xor	r8, rsi
+	mov	QWORD PTR [r12+8], r11
+	mov	r10, r8
+	xor	r11, r13
+	and	r10d, 2097136
+	movdqa	xmm3, xmm5
+	dec	ebp
+	jne	main_loop_ryzen
+
+	ldmxcsr DWORD PTR [rsp]
+	movaps	xmm6, XMMWORD PTR [rsp+48]
+	lea	r11, QWORD PTR [rsp+64]
+	mov	rbx, QWORD PTR [r11+56]
+	mov	rbp, QWORD PTR [r11+64]
+	mov	rsi, QWORD PTR [r11+72]
+	movaps	xmm8, XMMWORD PTR [r11-48]
+	movaps	xmm7, XMMWORD PTR [rsp+32]
+	mov	rsp, r11
+	pop	r15
+	pop	r14
+	pop	r13
+	pop	r12
+	pop	rdi
+	jmp cnv2_main_loop_ryzen_endp
+
+sqrt_fixup_ryzen:
+	movq r9, xmm2
+	dec	rdi
+	mov edx, -1022
+	shl rdx, 32
+	mov	rax, rdi
+	shr	rdi, 19
+	shr	rax, 20
+	mov	rcx, rdi
+	sub	rcx, rax
+	lea	rcx, [rcx+rdx+1]
+	add	rax, rdx
+	imul	rcx, rax
+	sub	rcx, r9
+	adc	rdi, 0
+	jmp	sqrt_fixup_ryzen_ret
+
+cnv2_main_loop_ryzen_endp:

src/crypto/asm/cnv2_main_loop_win.S

@@ -0,0 +1,15 @@
+#define ALIGN .align
+.intel_syntax noprefix
+.section .text
+.global cnv2_mainloop_ivybridge_asm
+.global cnv2_mainloop_ryzen_asm
+
+ALIGN 16
+cnv2_mainloop_ivybridge_asm:
+	#include "cnv2_main_loop_ivybridge.inc"
+	ret 0
+
+ALIGN 16
+cnv2_mainloop_ryzen_asm:
+	#include "cnv2_main_loop_ryzen.inc"
+	ret 0

src/workers/CpuThread.cpp

@@ -24,8 +24,10 @@
 #include <assert.h>
 
 
+#include "common/cpu/Cpu.h"
 #include "common/log/Log.h"
 #include "common/net/Pool.h"
+#include "crypto/Asm.h"
 #include "rapidjson/document.h"
 #include "workers/CpuThread.h"
 
@@ -37,9 +39,10 @@
 #endif
 
 
-xmrig::CpuThread::CpuThread(size_t index, Algo algorithm, AlgoVariant av, Multiway multiway, int64_t affinity, int priority, bool softAES, bool prefetch) :
+xmrig::CpuThread::CpuThread(size_t index, Algo algorithm, AlgoVariant av, Multiway multiway, int64_t affinity, int priority, bool softAES, bool prefetch, Assembly assembly) :
     m_algorithm(algorithm),
     m_av(av),
+    m_assembly(assembly),
     m_prefetch(prefetch),
     m_softAES(softAES),
     m_priority(priority),
@@ -50,22 +53,23 @@ xmrig::CpuThread::CpuThread(size_t index, Algo algorithm, AlgoVariant av, Multiw
 }
 
 
-xmrig::CpuThread::~CpuThread()
-{
-}
-
-
 bool xmrig::CpuThread::isSoftAES(AlgoVariant av)
 {
     return av == AV_SINGLE_SOFT || av == AV_DOUBLE_SOFT || av > AV_PENTA;
 }
 
 
-xmrig::CpuThread::cn_hash_fun xmrig::CpuThread::fn(Algo algorithm, AlgoVariant av, Variant variant)
+xmrig::CpuThread::cn_hash_fun xmrig::CpuThread::fn(Algo algorithm, AlgoVariant av, Variant variant, Assembly assembly)
 {
     assert(variant >= VARIANT_0 && variant < VARIANT_MAX);
 
-    static const cn_hash_fun func_table[VARIANT_MAX * 10 * 3] = {
+#   ifndef XMRIG_NO_ASM
+    constexpr const size_t count = VARIANT_MAX * 10 * 3 + 2;
+#   else
+    constexpr const size_t count = VARIANT_MAX * 10 * 3;
+#   endif
+
+    static const cn_hash_fun func_table[count] = {
         cryptonight_single_hash<CRYPTONIGHT, false, VARIANT_0>,
         cryptonight_double_hash<CRYPTONIGHT, false, VARIANT_0>,
         cryptonight_single_hash<CRYPTONIGHT, true,  VARIANT_0>,
@@ -241,12 +245,15 @@ xmrig::CpuThread::cn_hash_fun xmrig::CpuThread::fn(Algo algorithm, AlgoVariant a
         nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
         nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
         nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
+#       endif
+#       ifndef XMRIG_NO_ASM
+        cryptonight_single_hash_asm<CRYPTONIGHT, VARIANT_2, ASM_INTEL>,
+        cryptonight_single_hash_asm<CRYPTONIGHT, VARIANT_2, ASM_RYZEN>
 #       endif
     };
 
-    const size_t index = VARIANT_MAX * 10 * algorithm + 10 * variant + av - 1;
-
 #   ifndef NDEBUG
+    const size_t index = fnIndex(algorithm, av, variant, assembly);
     cn_hash_fun func = func_table[index];
 
     assert(index < sizeof(func_table) / sizeof(func_table[0]));
@@ -254,12 +261,12 @@ xmrig::CpuThread::cn_hash_fun xmrig::CpuThread::fn(Algo algorithm, AlgoVariant a
 
     return func;
 #   else
-    return func_table[index];
+    return func_table[fnIndex(algorithm, av, variant, assembly)];
 #   endif
 }
 
 
-xmrig::CpuThread *xmrig::CpuThread::createFromAV(size_t index, Algo algorithm, AlgoVariant av, int64_t affinity, int priority)
+xmrig::CpuThread *xmrig::CpuThread::createFromAV(size_t index, Algo algorithm, AlgoVariant av, int64_t affinity, int priority, Assembly assembly)
 {
     assert(av > AV_AUTO && av < AV_MAX);
 
@@ -282,7 +289,7 @@ xmrig::CpuThread *xmrig::CpuThread::createFromAV(size_t index, Algo algorithm, A
         }
     }
 
-    return new CpuThread(index, algorithm, av, multiway(av), cpuId, priority, isSoftAES(av), false);
+    return new CpuThread(index, algorithm, av, multiway(av), cpuId, priority, isSoftAES(av), false, assembly);
 }
 
 
@@ -300,7 +307,7 @@ xmrig::CpuThread *xmrig::CpuThread::createFromData(size_t index, Algo algorithm,
 
     assert(av > AV_AUTO && av < AV_MAX);
 
-    return new CpuThread(index, algorithm, static_cast<AlgoVariant>(av), multiway, data.affinity, priority, softAES, false);
+    return new CpuThread(index, algorithm, static_cast<AlgoVariant>(av), multiway, data.affinity, priority, softAES, false, data.assembly);
 }
 
 
@@ -322,11 +329,14 @@ xmrig::CpuThread::Data xmrig::CpuThread::parse(const rapidjson::Value &object)
     }
 
     const auto &affinity = object["affine_to_cpu"];
-
     if (affinity.IsUint64()) {
         data.affinity = affinity.GetInt64();
     }
 
+#   ifndef XMRIG_NO_ASM
+    data.assembly = Asm::parse(object["asm"]);
+#   endif
+
     return data;
 }
 
@@ -368,7 +378,11 @@ void xmrig::CpuThread::print() const
     LOG_DEBUG(GREEN_BOLD("CPU thread:   ") " index " WHITE_BOLD("%zu") ", multiway " WHITE_BOLD("%d") ", av " WHITE_BOLD("%d") ",",
               index(), static_cast<int>(multiway()), static_cast<int>(m_av));
 
+#   ifndef XMRIG_NO_ASM
+    LOG_DEBUG("               assembly: %s, affine_to_cpu: %" PRId64, Asm::toString(m_assembly), affinity());
+#   else
     LOG_DEBUG("               affine_to_cpu: %" PRId64, affinity());
+#   endif
 }
 #endif
 
@@ -403,5 +417,35 @@ rapidjson::Value xmrig::CpuThread::toConfig(rapidjson::Document &doc) const
     obj.AddMember("low_power_mode", multiway(), allocator);
     obj.AddMember("affine_to_cpu",  affinity() == -1L ? Value(kFalseType) : Value(affinity()), allocator);
 
+#   ifndef XMRIG_NO_ASM
+    obj.AddMember("asm", Asm::toJSON(m_assembly), allocator);
+#   endif
+
     return obj;
 }
+
+
+size_t xmrig::CpuThread::fnIndex(Algo algorithm, AlgoVariant av, Variant variant, Assembly assembly)
+{
+    const size_t index = VARIANT_MAX * 10 * algorithm + 10 * variant + av - 1;
+
+#   ifndef XMRIG_NO_ASM
+    if (assembly == ASM_AUTO) {
+        assembly = Cpu::info()->assembly();
+    }
+
+    if (assembly == ASM_NONE) {
+        return index;
+    }
+
+    constexpr const size_t offset = VARIANT_MAX * 10 * 3;
+
+    if (algorithm == CRYPTONIGHT && variant == VARIANT_2) {
+        if (av == AV_SINGLE) {
+            return offset + assembly - 2;
+        }
+    }
+#   endif
+
+    return index;
+}

src/workers/CpuThread.h

@@ -40,7 +40,7 @@ class CpuThread : public IThread
 public:
     struct Data
     {
-        inline Data() : valid(false), affinity(-1L), multiway(SingleWay) {}
+        inline Data() : assembly(ASM_AUTO), valid(false), affinity(-1L), multiway(SingleWay) {}
 
         inline void setMultiway(int value)
         {
@@ -50,27 +50,27 @@ public:
             }
         }
 
+        Assembly assembly;
         bool valid;
         int64_t affinity;
         Multiway multiway;
     };
 
 
-    CpuThread(size_t index, Algo algorithm, AlgoVariant av, Multiway multiway, int64_t affinity, int priority, bool softAES, bool prefetch);
+    CpuThread(size_t index, Algo algorithm, AlgoVariant av, Multiway multiway, int64_t affinity, int priority, bool softAES, bool prefetch, Assembly assembly);
-    ~CpuThread();
 
     typedef void (*cn_hash_fun)(const uint8_t *input, size_t size, uint8_t *output, cryptonight_ctx **ctx);
 
     static bool isSoftAES(AlgoVariant av);
-    static cn_hash_fun fn(Algo algorithm, AlgoVariant av, Variant variant);
+    static cn_hash_fun fn(Algo algorithm, AlgoVariant av, Variant variant, Assembly assembly);
-    static CpuThread *createFromAV(size_t index, Algo algorithm, AlgoVariant av, int64_t affinity, int priority);
+    static CpuThread *createFromAV(size_t index, Algo algorithm, AlgoVariant av, int64_t affinity, int priority, Assembly assembly);
     static CpuThread *createFromData(size_t index, Algo algorithm, const CpuThread::Data &data, int priority, bool softAES);
     static Data parse(const rapidjson::Value &object);
     static Multiway multiway(AlgoVariant av);
 
     inline bool isPrefetch() const               { return m_prefetch; }
     inline bool isSoftAES() const                { return m_softAES; }
-    inline cn_hash_fun fn(Variant variant) const { return fn(m_algorithm, m_av, variant); }
+    inline cn_hash_fun fn(Variant variant) const { return fn(m_algorithm, m_av, variant, m_assembly); }
 
     inline Algo algorithm() const override       { return m_algorithm; }
     inline int priority() const override         { return m_priority; }
@@ -91,8 +91,11 @@ protected:
     rapidjson::Value toConfig(rapidjson::Document &doc) const override;
 
 private:
+    static size_t fnIndex(Algo algorithm, AlgoVariant av, Variant variant, Assembly assembly);
+
     const Algo m_algorithm;
     const AlgoVariant m_av;
+    const Assembly m_assembly;
     const bool m_prefetch;
     const bool m_softAES;
     const int m_priority;

src/workers/Worker.cpp

@@ -24,8 +24,8 @@
 #include <chrono>
 
 
+#include "common/cpu/Cpu.h"
 #include "common/Platform.h"
-#include "Cpu.h"
 #include "workers/CpuThread.h"
 #include "workers/Handle.h"
 #include "workers/Worker.h"
@@ -41,7 +41,7 @@ Worker::Worker(Handle *handle) :
     m_sequence(0),
     m_thread(static_cast<xmrig::CpuThread *>(handle->config()))
 {
-    if (Cpu::threads() > 1 && m_thread->affinity() != -1L) {
+    if (xmrig::Cpu::info()->threads() > 1 && m_thread->affinity() != -1L) {
         Platform::setThreadAffinity(m_thread->affinity());
     }
 

src/workers/Worker.h

@@ -21,8 +21,8 @@
  *   along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#ifndef __WORKER_H__
+#ifndef XMRIG_WORKER_H
-#define __WORKER_H__
+#define XMRIG_WORKER_H
 
 
 #include <atomic>
@@ -33,7 +33,6 @@
 #include "Mem.h"
 
 
-struct cryptonight_ctx;
 class Handle;
 
 
@@ -67,4 +66,4 @@ protected:
 };
 
 
-#endif /* __WORKER_H__ */
+#endif /* XMRIG_WORKER_H */