Fixes compilation w/ CMake & naive SIMD implementation on macOS/arm64

.gitignore

@@ -0,0 +1 @@
+build/

PLA-SeedFinder/CMakeLists.txt

@@ -16,31 +16,31 @@ set(CMAKE_CXX_STANDARD_REQUIRED ON)
 set(CMAKE_EXPORT_COMPILE_COMMANDS ON)
 #set(CMAKE_VERBOSE_MAKEFILE ON)
 
-set(CMAKE_AUTOMOC ON)
-set(CMAKE_AUTORCC ON)
-set(CMAKE_AUTOUIC ON)
-
-add_custom_target(build-time-make-directory ALL
-    COMMAND ${CMAKE_COMMAND} -E make_directory Assembly/)
-
 #Find threads library
 set(THREADS_PREFER_PTHREAD_FLAG ON)
 find_package(Threads REQUIRED)
 
+#detect cpu arch
+if (CMAKE_SYSTEM_PROCESSOR MATCHES "(x86)|(X86)|(amd64)|(AMD64)")
+    set (X86 TRUE)
+else ()
+    set (X86 FALSE)
+endif ()
 
 #add current directory to find tesseractPA.lib
 #link_directories(${CMAKE_CURRENT_LIST_DIR})
 
-file(GLOB MAIN_SOURCES
+if (X86)
+    file(GLOB MAIN_SOURCES
     Source/Compiler.h
-    Source/CpuId.cpp
+    Source/CpuFeatures.cpp
     Source/CpuFeatures.h
     Source/DynamicParallelizer.cpp
     Source/DynamicParallelizer.h
     Source/Kernels/SeedScan_Default.cpp
-    Source/Kernels/SeedScan_x64_AVX2.cpp
-    Source/Kernels/SeedScan_x64_AVX512.cpp
-    Source/Kernels/SeedScan_x64_SSE41.cpp
+    Source/Kernels/SeedScan_x86_AVX2.cpp
+    Source/Kernels/SeedScan_x86_AVX512.cpp
+    Source/Kernels/SeedScan_x86_SSE41.cpp
     Source/Kernels/XoroShiro1_Default.h
     Source/Kernels/XoroShiro2_SSE2.h
     Source/Kernels/XoroShiro4_AVX2.h
@@ -54,8 +54,25 @@ file(GLOB MAIN_SOURCES
     Source/SeedScan.h
     Source/Tools.cpp
     Source/Tools.h
-)
-add_executable(PLA-SeedFinder WIN32 ${MAIN_SOURCES})
+    )
+else ()
+    file(GLOB MAIN_SOURCES
+        Source/DynamicParallelizer.cpp
+        Source/Kernels/SeedScan_Default.cpp
+        Source/Kernels/SeedScan_aarch64.cpp
+        Source/Main.cpp
+        Source/PLA-SeedFinder.cpp
+        Source/ReportCandidates.cpp
+        Source/SeedScan.cpp
+        Source/Tools.cpp
+    )
+endif (X86)
+
+if (WIN32)
+    add_executable(PLA-SeedFinder WIN32 ${MAIN_SOURCES})
+else ()
+    add_executable(PLA-SeedFinder ${MAIN_SOURCES})
+endif ()
 set_target_properties(PLA-SeedFinder PROPERTIES LINKER_LANGUAGE CXX)
 target_link_libraries(PLA-SeedFinder)
 target_link_libraries(PLA-SeedFinder Threads::Threads)
@@ -68,6 +85,9 @@ target_include_directories(PLA-SeedFinder PRIVATE Source/)
 
 #enable MP with MSVC (Build with Multiple Processes)
 if (MSVC)
+    add_custom_target(build-time-make-directory ALL
+    COMMAND ${CMAKE_COMMAND} -E make_directory Assembly/)
+
     target_compile_options(PLA-SeedFinder PRIVATE /FAs /FaAssembly/ /MP /W4)
     target_compile_options(PLA-SeedFinder PRIVATE /wd5054)  # Deprecated enum arithemtic
     target_compile_options(PLA-SeedFinder PRIVATE /wd4505)  # unreferenced local function has been removed
@@ -82,50 +102,49 @@ if (MSVC)
     target_compile_definitions(PLA-SeedFinder PRIVATE PA_AutoDispatch_17_Skylake)
 
 else()
-    target_compile_options(PLA-SeedFinder PRIVATE -msse4.2)
-
-    target_compile_options(PLA-SeedFinder PRIVATE -Wall -Wpedantic -DPA_STATIC)
-
-    set(ARCH_FLAGS_09_Nehalem       -march=nehalem)
-    set(ARCH_FLAGS_13_Haswell       -march=haswell)
-    set(ARCH_FLAGS_17_Skylake       -march=skylake-avx512)
-
-    #   Run-time ISA dispatching
-    target_compile_definitions(PLA-SeedFinder PRIVATE PA_AutoDispatch_08_Nehalem)
-    target_compile_definitions(PLA-SeedFinder PRIVATE PA_AutoDispatch_13_Haswell)
-    target_compile_definitions(PLA-SeedFinder PRIVATE PA_AutoDispatch_17_Skylake)
-endif()
-
-
-
-#   Run-time CPU dispatching.
-if (ARCH_FLAGS_09_Nehalem)
-SET_SOURCE_FILES_PROPERTIES(
-    Source/Kernels/SeedScan_x64_SSE41.cpp
-    PROPERTIES COMPILE_FLAGS ${ARCH_FLAGS_09_Nehalem}
-)
-endif()
-if (ARCH_FLAGS_13_Haswell)
-SET_SOURCE_FILES_PROPERTIES(
-    Source/Kernels/SeedScan_x64_AVX2.cpp
-    PROPERTIES COMPILE_FLAGS ${ARCH_FLAGS_13_Haswell}
-)
+    target_compile_options(PLA-SeedFinder PRIVATE -Wall -Wpedantic -O2)
+    if (X86)
+        target_compile_options(PLA-SeedFinder PRIVATE -msse4.2)
+
+        set(ARCH_FLAGS_09_Nehalem       -march=nehalem)
+        set(ARCH_FLAGS_13_Haswell       -march=haswell)
+        set(ARCH_FLAGS_17_Skylake       -march=skylake-avx512)
+
+        #   Run-time ISA dispatching
+        target_compile_definitions(PLA-SeedFinder PRIVATE PA_AutoDispatch_08_Nehalem)
+        target_compile_definitions(PLA-SeedFinder PRIVATE PA_AutoDispatch_13_Haswell)
+        target_compile_definitions(PLA-SeedFinder PRIVATE PA_AutoDispatch_17_Skylake)
+    endif (X86)
 endif()
-if (ARCH_FLAGS_17_Skylake)
-SET_SOURCE_FILES_PROPERTIES(
-    Source/Kernels/SeedScan_x64_AVX512.cpp
-    PROPERTIES COMPILE_FLAGS ${ARCH_FLAGS_17_Skylake}
-)
-endif()
-
-
-
-
 
 
-#copy needed dlls
-#file(COPY *.dll DESTINATION ${CMAKE_CURRENT_BINARY_DIR})
-file(GLOB MY_DLLS
-  "*.dll"
-)
-file(COPY ${MY_DLLS} DESTINATION ${CMAKE_CURRENT_BINARY_DIR})
+if(X86)
+    #   Run-time CPU dispatching.
+    if (ARCH_FLAGS_09_Nehalem)
+    SET_SOURCE_FILES_PROPERTIES(
+        Source/Kernels/SeedScan_x86_SSE41.cpp
+        PROPERTIES COMPILE_FLAGS ${ARCH_FLAGS_09_Nehalem}
+    )
+    endif()
+    if (ARCH_FLAGS_13_Haswell)
+    SET_SOURCE_FILES_PROPERTIES(
+        Source/Kernels/SeedScan_x86_AVX2.cpp
+        PROPERTIES COMPILE_FLAGS ${ARCH_FLAGS_13_Haswell}
+    )
+    endif()
+    if (ARCH_FLAGS_17_Skylake)
+    SET_SOURCE_FILES_PROPERTIES(
+        Source/Kernels/SeedScan_x86_AVX512.cpp
+        PROPERTIES COMPILE_FLAGS ${ARCH_FLAGS_17_Skylake}
+    )
+    endif()
+endif(X86)
+
+if (WIN32)
+    #copy needed dlls
+    #file(COPY *.dll DESTINATION ${CMAKE_CURRENT_BINARY_DIR})
+    file(GLOB MY_DLLS
+    "*.dll"
+    )
+    file(COPY ${MY_DLLS} DESTINATION ${CMAKE_CURRENT_BINARY_DIR})
+endif(WIN32)

PLA-SeedFinder/Source/Kernels/SeedScan_aarch64.cpp

@@ -0,0 +1,119 @@
+#include <stdint.h>
+#include <cstddef>
+#include "XoroShiro_aarch64.h"
+
+namespace PokemonAutomation{
+
+#if defined __aarch64__ && defined __APPLE__
+
+bool seed_scan_common_unroll4_NEON(size_t rolls, uint32_t desired_pid, uint64_t start_seed, uint64_t iterations){
+    simd_ulong4 t = simd_make_ulong4((uint64_t)desired_pid, (uint64_t)desired_pid, (uint64_t)desired_pid, (uint64_t)desired_pid);
+    simd_ulong4 seed = simd_make_ulong4(start_seed, start_seed+0x200000000, start_seed+0x300000000, start_seed+0x400000000);
+    simd_ulong4 delta = simd_make_ulong4(0x500000000, 0x500000000, 0x500000000, 0x500000000);
+    simd_ulong4 pid;
+    iterations /= 4;
+
+    do {
+        XoroShiro4 rng(seed);
+        rng.next();
+        size_t lc = rolls;
+        do {
+            rng.next();
+            pid = rng.get_masked();
+            if (simd_reduce_min(pid ^ t) == 0) {
+                return true;
+            }
+        } while (--lc);
+        seed += delta;
+    } while (--iterations);
+    return false;
+}
+
+bool seed_scan_thorough_unroll4_NEON(size_t rolls, uint32_t desired_pid, uint64_t start_seed, uint64_t iterations){
+    desired_pid &= 0xefffffff;
+    simd_ulong4 t = simd_make_ulong4((uint64_t)desired_pid, (uint64_t)desired_pid, (uint64_t)desired_pid, (uint64_t)desired_pid);
+    simd_ulong4 seed = simd_make_ulong4(start_seed, start_seed+0x200000000, start_seed+0x300000000, start_seed+0x400000000);
+    simd_ulong4 delta = simd_make_ulong4(0x500000000, 0x500000000, 0x500000000, 0x500000000);
+    simd_ulong4 mask = simd_make_ulong4(0xefffffff, 0xefffffff, 0xefffffff, 0xefffffff);
+    simd_ulong4 pid;
+    iterations /= 4;
+
+    do {
+        XoroShiro4 rng(seed);
+        size_t lc = rolls + 1;
+        do {
+            rng.next();
+            pid = rng.get_masked();
+            if (simd_reduce_max(pid) == UINT32_MAX) {
+                return true;
+            }
+            pid &= mask;
+            if (simd_reduce_min(pid ^ t) == 0) {
+                return true;
+            }
+        } while (--lc);
+        seed += delta;
+    } while (--iterations);
+    return false;
+}
+
+bool seed_scan_common_unroll8_NEON(size_t rolls, uint32_t desired_pid, uint64_t start_seed, uint64_t iterations){
+    simd_ulong8 t = simd_make_ulong8(simd_make_ulong4((uint64_t)desired_pid, (uint64_t)desired_pid, (uint64_t)desired_pid, (uint64_t)desired_pid),
+                                     simd_make_ulong4((uint64_t)desired_pid, (uint64_t)desired_pid, (uint64_t)desired_pid, (uint64_t)desired_pid));
+    simd_ulong8 seed = simd_make_ulong8(simd_make_ulong4(start_seed, start_seed+0x200000000, start_seed+0x300000000, start_seed+0x400000000),
+                                        simd_make_ulong4(start_seed+0x500000000, start_seed+0x600000000, start_seed+0x700000000, start_seed+0x800000000));
+    simd_ulong8 delta = simd_make_ulong8(simd_make_ulong4(0x900000000, 0x900000000, 0x900000000, 0x900000000),
+                                         simd_make_ulong4(0x900000000, 0x900000000, 0x900000000, 0x900000000));
+    simd_ulong8 pid;
+    iterations /= 8;
+
+    do {
+        XoroShiro8 rng(seed);
+        rng.next();
+        size_t lc = rolls;
+        do {
+            rng.next();
+            pid = rng.get_masked();
+            if (simd_reduce_min(pid ^ t) == 0) {
+                return true;
+            }
+        } while (--lc);
+        seed += delta;
+    } while (--iterations);
+    return false;
+}
+
+bool seed_scan_thorough_unroll8_NEON(size_t rolls, uint32_t desired_pid, uint64_t start_seed, uint64_t iterations){
+    desired_pid &= 0xefffffff;
+    simd_ulong8 t = simd_make_ulong8(simd_make_ulong4((uint64_t)desired_pid, (uint64_t)desired_pid, (uint64_t)desired_pid, (uint64_t)desired_pid),
+                                     simd_make_ulong4((uint64_t)desired_pid, (uint64_t)desired_pid, (uint64_t)desired_pid, (uint64_t)desired_pid));
+    simd_ulong8 seed = simd_make_ulong8(simd_make_ulong4(start_seed, start_seed+0x200000000, start_seed+0x300000000, start_seed+0x400000000),
+                                        simd_make_ulong4(start_seed+0x500000000, start_seed+0x600000000, start_seed+0x700000000, start_seed+0x800000000));
+    simd_ulong8 delta = simd_make_ulong8(simd_make_ulong4(0x900000000, 0x900000000, 0x900000000, 0x900000000),
+                                         simd_make_ulong4(0x900000000, 0x900000000, 0x900000000, 0x900000000));
+    simd_ulong8 u1 = simd_make_ulong8(simd_make_ulong4(0xefffffff, 0xefffffff, 0xefffffff, 0xefffffff), 
+                                     simd_make_ulong4(0xefffffff, 0xefffffff, 0xefffffff, 0xefffffff));
+    simd_ulong8 pid;
+    iterations /= 8;
+
+    do {
+        XoroShiro8 rng(seed);
+        size_t lc = rolls + 1;
+        do {
+            rng.next();
+            pid = rng.get_masked();
+            if (simd_reduce_max(pid) == UINT32_MAX) {
+                return true;
+            }
+            pid &= u1;
+            if (simd_reduce_min(pid ^ t) == 0) {
+                return true;
+            }
+        } while (--lc);
+        seed += delta;
+    } while (--iterations);
+    return false;
+}
+
+#endif
+}

PLA-SeedFinder/Source/Kernels/XoroShiro1_Default.h

@@ -18,11 +18,7 @@ namespace PokemonAutomation{
 
 class XoroShiroX1_Default{
 public:
-    PA_FORCE_INLINE XoroShiroX1_Default(uint64_t seed){
-        state[0] = seed;
-        state[1] = 0x82A2B175229D6A5B;
-    }
-
+    PA_FORCE_INLINE XoroShiroX1_Default(uint64_t seed) : state {seed, 0x82A2B175229D6A5B} {}
     PA_FORCE_INLINE uint64_t get_int64(){
         return state[0] + state[1];
     }

PLA-SeedFinder/Source/Kernels/XoroShiro_aarch64.h

@@ -0,0 +1,61 @@
+#pragma once
+
+#if defined __aarch64__ && defined __APPLE__
+#include <simd/simd.h>
+#include "Compiler.h"
+
+#define MAGIC_NUMBER 0x82A2B175229D6A5B
+
+namespace PokemonAutomation{
+
+static const simd_ulong4 magic_ulong4 = simd_make_ulong4(MAGIC_NUMBER, MAGIC_NUMBER, MAGIC_NUMBER, MAGIC_NUMBER);
+static const simd_ulong4 mask_ulong4 = simd_make_ulong4(UINT32_MAX, UINT32_MAX, UINT32_MAX, UINT32_MAX);
+static const simd_ulong8 mask_ulong8 = simd_make_ulong8(simd_make_ulong4(UINT32_MAX, UINT32_MAX, UINT32_MAX, UINT32_MAX),
+                                                        simd_make_ulong4(UINT32_MAX, UINT32_MAX, UINT32_MAX, UINT32_MAX));
+
+class XoroShiro4{
+public:
+    PA_FORCE_INLINE XoroShiro4(simd_ulong4 seed)
+    : state {seed, magic_ulong4} {}
+    PA_FORCE_INLINE simd_ulong4 get(){ return state[0] + state[1]; }
+    PA_FORCE_INLINE simd_ulong4 get_masked(){ return (state[0] + state[1]) & mask_ulong4; }
+    PA_FORCE_INLINE void next(){
+        simd_ulong4 s0 = state[0];
+        simd_ulong4 s1 = state[1];        
+        s1 = s1 ^ s0;
+        s0 = (s0 << 24) | (s0 >> 40);
+        state[1] = (s1 << 37) | (s1 >> 27);
+        s1 = s1 ^ (s1 << 16);
+        state[0] = s0 ^ s1;
+    }
+
+private:
+    simd_ulong4 state[2];
+};
+
+
+class XoroShiro8{
+public:
+    PA_FORCE_INLINE XoroShiro8(simd_ulong8 seed)
+    : state {seed, simd_make_ulong8(magic_ulong4, magic_ulong4)} {}
+    PA_FORCE_INLINE simd_ulong8 get(){ return state[0] + state[1]; }
+    PA_FORCE_INLINE simd_ulong8 get_masked(){ return (state[0] + state[1]) & mask_ulong8; }
+    PA_FORCE_INLINE void next(){
+        simd_ulong8 s0 = state[0];
+        simd_ulong8 s1 = state[1];        
+        s1 = s1 ^ s0;
+        s0 = (s0 << 24) | (s0 >> 40);
+        state[1] = (s1 << 37) | (s1 >> 27);
+        s1 = s1 ^ (s1 << 16);
+        state[0] = s0 ^ s1;
+    }
+
+private:
+    simd_ulong8 state[2];
+};
+
+
+}
+
+#undef MAGIC_NUMBER
+#endif

PLA-SeedFinder/Source/Main.cpp

@@ -50,7 +50,6 @@ void test(){
 }
 #endif
 
-#include <immintrin.h>
 
 
 int main(){

PLA-SeedFinder/Source/PLA-SeedFinder.cpp

@@ -56,8 +56,6 @@ int32_t pa_PLA_find_seeds_threads(
                          << stats.ivs[4] << " "
                          << stats.ivs[5] << std::endl;
 
-    EcPidMatchReporter reporter(stats);
-
     std::cout << std::endl;
     print_isa();
     std::cout << "Threads: " << threads << std::endl;