Merge pull request #7 from sirzooro/rank10-optimization

Rank10 optimization

.clang-format

@@ -0,0 +1,28 @@
+BasedOnStyle: LLVM
+AccessModifierOffset: -4
+AllowShortFunctionsOnASingleLine: None
+BreakBeforeBraces: Custom
+BraceWrapping:
+  AfterClass: true
+  AfterControlStatement: true
+  AfterEnum: true
+  AfterFunction: true
+  AfterNamespace: true
+  AfterStruct: true
+  AfterUnion: true
+  AfterExternBlock: true
+  BeforeCatch: true
+  BeforeElse: true
+  SplitEmptyFunction: true
+  SplitEmptyRecord: true
+  SplitEmptyNamespace: true
+ColumnLimit: 120
+IncludeBlocks: Preserve
+IndentWidth: 4
+PointerAlignment: Left
+ReflowComments: false
+SortIncludes: false
+Standard: C++11
+TabWidth: 4
+UseTab: Never
+...

.gitignore

@@ -1,32 +1,33 @@
-# Windows image file caches
-Thumbs.db
-ehthumbs.db
-
-# Folder config file
-Desktop.ini
-
-# Recycle Bin used on file shares
-$RECYCLE.BIN/
-
-# Windows Installer files
-*.cab
-*.msi
-*.msm
-*.msp
-
-# =========================
-# Operating System Files
-# =========================
-
-# OSX
-# =========================
-
+# Windows image file caches
+Thumbs.db
+ehthumbs.db
+
+# Folder config file
+Desktop.ini
+
+# Recycle Bin used on file shares
+$RECYCLE.BIN/
+
+# Windows Installer files
+*.cab
+*.msi
+*.msm
+*.msp
+
+# =========================
+# Operating System Files
+# =========================
+
+# OSX
+# =========================
+
 .DS_Store
 .AppleDouble
 .LSOverride
 
 # Icon must end with two \r
-Icon
+Icon
+
 
 # Thumbnails
 ._*
@@ -41,3 +42,12 @@ Icon
 Network Trash Folder
 Temporary Items
 .apdisk
+
+# Files created during build
+*.o
+*.d
+*.bak
+RakeSearch/RakeSearchV3/RakeDiagSearchV3/RakeDiagSearchV3/rakesearch
+RakeSearch/RakeSearchV3/RakeDiagSearchV3/RakeDiagSearchV3/rakesearch.exe
+RakeSearch/RakeSearchV3/RakeDiagSearchV3/RakeDiagSearchV3/tests/tests
+RakeSearch/RakeSearchV3/RakeDiagSearchV3/RakeDiagSearchV3/tests/tests.exe

RakeSearchV3/RakeDiagSearchV3/RakeDiagSearchV3/Helpers.h

@@ -0,0 +1,33 @@
+// Various helper stuff
+
+#if !defined Helpers_h
+#define Helpers_h
+
+#ifndef NO_SIMD
+
+#if defined(__SSE2__) || defined(__ARM_NEON)
+#define HAS_SIMD 1
+#endif
+
+#ifdef __AVX512F__
+#define ALIGNED __attribute__((aligned(64)))
+#elif defined(__SSE2__) || defined(__ARM_NEON)
+#define ALIGNED __attribute__((aligned(32)))
+#else
+#define ALIGNED
+#endif
+
+#else // !NO_SIMD
+#define ALIGNED
+#endif // !NO_SIMD
+
+#ifdef UT_BUILD
+#define private public
+#define UT_VIRTUAL virtual
+
+#else
+#define UT_VIRTUAL
+
+#endif
+
+#endif

RakeSearchV3/RakeDiagSearchV3/RakeDiagSearchV3/Makefile

@@ -0,0 +1,101 @@
+LD_PTHREAD = -pthread
+
+ifeq ($(MinGW32),1)
+$(info ===== Compiling MinGW 32-bit app version =====)
+CXX = i686-w64-mingw32-g++
+BOINC_DIR = /boinc930/mingw32
+else ifeq ($(MinGW64),1)
+$(info ===== Compiling MinGW 64-bit app version =====)
+CXX = x86_64-w64-mingw32-g++
+BOINC_DIR = /boinc930/mingw64
+else ifeq ($(AARCH64),1)
+$(info ===== Compiling AARCH64 app version =====)
+CXX = aarch64-unknown-linux-gnu-g++
+BOINC_DIR = /boinc930/aarch64
+else ifeq ($(ARM),1)
+# Fix for "terminate called after throwing an instance of '__gnu_cxx::__concurrence_broadcast_error'" on ARM
+LD_PTHREAD = -Wl,-whole-archive -lpthread -Wl,-no-whole-archive
+ifeq ($(NEON),1)
+$(info ===== Compiling ARMv7 HF app version =====)
+CXX = arm-unknown-linux-gnueabihf-g++
+BOINC_DIR = /boinc930/armhf
+else
+$(info ===== Compiling ARMv7 SF app version =====)
+CXX = arm-unknown-linux-gnueabi-g++
+BOINC_DIR = /boinc930/arm
+# Fix for "terminate called after throwing an instance of '__gnu_cxx::__concurrence_broadcast_error'" on ARM
+LD_PTHREAD = -Wl,-whole-archive -lpthread -Wl,-no-whole-archive
+endif
+else ifeq ($(ARMv6),1)
+$(info ===== Compiling ARMv6 app version =====)
+CXX = armv6-unknown-linux-gnueabi-g++
+BOINC_DIR = /boinc930/armv6
+else ifeq ($(M32),1)
+$(info ===== Compiling 32-bit app version =====)
+CXX = g++ -m32 -mtune=generic -march=pentiumpro
+BOINC_DIR = /boinc930/linux32
+else
+$(info ===== Compiling default app version =====)
+CXX = g++
+BOINC_DIR = /boinc930/linux64
+endif
+
+ifeq ($(Native),1)
+# Tune for machine where app is compiled
+$(info ===== Compiling native app version =====)
+TARGET_FLAGS = -march=native -mtune=native
+else ifeq ($(AVX512),1)
+# AVX512+BMI2
+$(info ===== Compiling AVX512+BMI2 app version =====)
+TARGET_FLAGS = -march=skylake-avx512 -mprefer-vector-width=256
+else ifeq ($(AVX2),1)
+# AVX2+BMI2
+$(info ===== Compiling AVX2+BMI2 app version =====)
+TARGET_FLAGS = -march=core2 -mtune=haswell -mpopcnt -mavx2 -mbmi -mbmi2
+else ifeq ($(AVX),1)
+# AVX
+$(info ===== Compiling AVX app version =====)
+TARGET_FLAGS = -march=core2 -mtune=sandybridge -mpopcnt -mavx -mprefer-vector-width=128
+else ifeq ($(SSE41),1)
+# SSE4.1
+$(info ===== Compiling SSE4.1 app version =====)
+TARGET_FLAGS = -mtune=core2 -msse4.1
+else ifeq ($(SSSE3),1)
+# SSSE3
+$(info ===== Compiling SSSE3 app version =====)
+TARGET_FLAGS = -mtune=core2 -mssse3
+else ifeq ($(SSE2),1)
+# SSE2
+$(info ===== Compiling SSE2 app version =====)
+TARGET_FLAGS = -mtune=core2 -msse2
+else ifeq ($(NEON),1)
+# NEON (ARM only; AARCH64 has NEON by default)
+$(info ===== Compiling ARM NEON app version =====)
+TARGET_FLAGS = -mfloat-abi=hard -mfpu=neon
+else ifeq ($(NOSIMD),1)
+$(info ===== Compiling No-SIMD app version =====)
+TARGET_FLAGS = -DNO_SIMD=1
+else
+$(info ===== Compiling default app version =====)
+endif
+
+CXXFLAGS = $(TARGET_FLAGS) -O3 -ftree-vectorize -pthread -std=c++11 -Wall \
+    -I$(BOINC_DIR)/include/boinc
+
+LDFLAGS = $(TARGET_FLAGS) -O3 -ftree-vectorize -static -static-libgcc -static-libstdc++ $(LD_PTHREAD) -std=c++11 -Wall \
+    -L$(BOINC_DIR)/lib
+
+PROGRAM = rakesearch10
+
+all: $(PROGRAM)
+
+OBJ_FILES = main.o Square.o RakeSearch.o
+
+clean:
+	rm -f $(PROGRAM) $(PROGRAM).exe *.o
+
+$(PROGRAM): $(OBJ_FILES)
+	$(CXX) $(LDFLAGS) -o $(PROGRAM) $(OBJ_FILES) -lboinc_api -lboinc
+
+%.o: %.cpp
+	$(CXX) $(CXXFLAGS) -c $< -o $@