Merge pull request #53 from Asherda/master

MSVC compilation support

.gitlab-ci.yml

@@ -1,7 +1,7 @@
 stages:
 - build
 variables:
-  VERSION: 0.7.2
+  VERSION: 0.8.0
   POST_MESSAGE: "Pipeline Trigger: ${CI_PIPELINE_SOURCE}\n
   Branch: ${CI_COMMIT_REF_NAME}\n
   Commit: ${CI_COMMIT_SHA}\n

doc/Linux/Readme.txt

@@ -1,4 +1,4 @@
-VerusCoin nheqminer v0.7.2
+VerusCoin nheqminer v0.8.0
 Using VerusHash v2.0
 
 This software needs to be run on a terminal.

doc/Mac/Readme.txt

@@ -1,4 +1,4 @@
-VerusCoin nheqminer v0.7.2
+VerusCoin nheqminer v0.8.0
 Using VerusHash v2.0
 
 This software needs to be run on a terminal. To find it, open the Applications folder, then open the Utilities folder and finally open the Terminal application. The terminal can also be found using spotlight and searching for “terminal”.

doc/Windows/Readme.txt

@@ -1,4 +1,4 @@
-VerusCoin nheqminer v0.7.2
+VerusCoin nheqminer v0.8.0
 Using VerusHash v2.0
 
 This software needs to be run on a terminal. From the start menu, type cmd and press enter to open a command prompt.

nheqminer/crypto/verus_clhash.cpp

@@ -32,8 +32,10 @@ int __cpuverusoptimized = 0x80;
 #if defined(__arm__)  || defined(__aarch64__)
 #include "crypto/SSE2NEON.h"
 #else
+#ifndef _WIN32
 #include <x86intrin.h>
 #endif
+#endif
 
 #include "../../cpu_verushash/cpu_verushash.hpp"
 
@@ -150,15 +152,22 @@ __m128i _mm_loadl_epi64(__m128i *a)
 #endif 
 
 // multiply the length and the some key, no modulo
-static inline __attribute__((always_inline)) __m128i lazyLengthHash(uint64_t keylength, uint64_t length) {
-
+#ifdef _MSC_VER
+    static __forceinline  __m128i lazyLengthHash(uint64_t keylength, uint64_t length) {
+#else
+    static inline __attribute__((always_inline)) __m128i lazyLengthHash(uint64_t keylength, uint64_t length) {
+#endif
     const __m128i lengthvector = _mm_set_epi64x(keylength,length);
     const __m128i clprod1 = _mm_clmulepi64_si128( lengthvector, lengthvector, 0x10);
     return clprod1;
 }
 
 // modulo reduction to 64-bit value. The high 64 bits contain garbage, see precompReduction64
-static inline __attribute__((always_inline)) __m128i precompReduction64_si128( __m128i A) {
+#ifdef _MSC_VER
+  static __forceinline  __m128i precompReduction64_si128( __m128i A) {
+#else
+  static inline __attribute__((always_inline)) __m128i precompReduction64_si128( __m128i A) {
+#endif
     //const __m128i C = _mm_set_epi64x(1U,(1U<<4)+(1U<<3)+(1U<<1)+(1U<<0)); // C is the irreducible poly. (64,4,3,1,0)
     const __m128i C = _mm_cvtsi64_si128((1U<<4)+(1U<<3)+(1U<<1)+(1U<<0));
     __m128i Q2 = _mm_clmulepi64_si128( A, C, 0x01);
@@ -169,11 +178,19 @@ static inline __attribute__((always_inline)) __m128i precompReduction64_si128( _
     return final;/// WARNING: HIGH 64 BITS CONTAIN GARBAGE
 }
 
-static inline __attribute__((always_inline)) uint64_t precompReduction64( __m128i A) {
+#ifdef _MSC_VER
+    static __forceinline uint64_t precompReduction64( __m128i A) {
+#else
+    static inline __attribute__((always_inline)) uint64_t precompReduction64( __m128i A) {
+#endif
     return _mm_cvtsi128_si64(precompReduction64_si128(A));
 }
 
-static inline __attribute__((always_inline)) void fixupkey(__m128i **pMoveScratch, verusclhash_descr *pdesc) {
+#ifdef _MSC_VER
+    static __forceinline  void fixupkey(__m128i **pMoveScratch, verusclhash_descr *pdesc) {
+#else
+    static inline __attribute__((always_inline)) void fixupkey(__m128i **pMoveScratch, verusclhash_descr *pdesc) {
+#endif
     uint32_t ofs = pdesc->keySizeInBytes >> 4;
     for (__m128i *pfixup = *pMoveScratch; pfixup; pfixup = *++pMoveScratch)
     {
@@ -182,7 +199,11 @@ static inline __attribute__((always_inline)) void fixupkey(__m128i **pMoveScratc
     }
 }
 
-static inline __attribute__((always_inline)) void haraka512_keyed_local(unsigned char *out, const unsigned char *in, const u128 *rc) {
+#ifdef _MSC_VER
+    static __forceinline void haraka512_keyed_local(unsigned char *out, const unsigned char *in, const u128 *rc) {
+#else
+    static inline __attribute__((always_inline)) void haraka512_keyed_local(unsigned char *out, const unsigned char *in, const u128 *rc) {
+#endif
   u128 s[4], tmp;
 
   s[0] = LOAD(in);

nheqminer/crypto/verus_clhash.h

@@ -30,7 +30,7 @@
 #ifdef _WIN32
 #undef __cpuid
 #include <intrin.h>
-#endif
+#else
 
 #if defined(__arm__)  || defined(__aarch64__)
 #include "crypto/SSE2NEON.h"
@@ -40,6 +40,7 @@
 #include <cpuid.h>
 #include <x86intrin.h>
 #endif // !WIN32
+#endif
 
 #include <boost/thread.hpp>
 #include "tinyformat.h"
@@ -118,6 +119,18 @@ inline bool IsCPUVerusOptimized()
     #else
     if (__cpuverusoptimized & 0x80)
     {
+        #ifdef _MSC_VER
+        #define bit_AVX		(1 << 28)
+        #define bit_AES		(1 << 25)
+        #define bit_PCLMUL  (1 << 1)
+        // https://insufficientlycomplicated.wordpress.com/2011/11/07/detecting-intel-advanced-vector-extensions-avx-in-visual-studio/
+        // bool cpuAVXSuport = cpuInfo[2] & (1 << 28) || false;
+
+        int cpuInfo[4];
+		__cpuid(cpuInfo, 1);
+        __cpuverusoptimized = ((cpuInfo[2] & (bit_AVX | bit_AES | bit_PCLMUL)) == (bit_AVX | bit_AES | bit_PCLMUL));
+
+        #else
         unsigned int eax,ebx,ecx,edx;
         if (!__get_cpuid(1,&eax,&ebx,&ecx,&edx))
         {
@@ -127,6 +140,7 @@ inline bool IsCPUVerusOptimized()
         {
             __cpuverusoptimized = ((ecx & (bit_AVX | bit_AES | bit_PCLMUL)) == (bit_AVX | bit_AES | bit_PCLMUL));
         }
+        #endif
     }
     #endif
     return __cpuverusoptimized;

nheqminer/version.h

@@ -34,7 +34,7 @@ static const int BIP0031_VERSION = 60000;
 //! "mempool" command, enhanced "getdata" behavior starts with this version
 static const int MEMPOOL_GD_VERSION = 60002;
 
-#define STANDALONE_MINER_VERSION	"0.7.2"
+#define STANDALONE_MINER_VERSION	"0.8.0"
 
 // uncomment to use with ZCash address
 //#define ZCASH_POOL