Merge a776ebf394 into 16ecb8f085

CUDA backend: update RandomX dataset when it changes

CMakeLists.txt

@@ -14,7 +14,9 @@ option(WITH_HTTP            "Enable HTTP protocol support (client/server)" 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(WITH_MSR             "Enable MSR mod & 1st-gen Ryzen fix" ON)
+option(WITH_ASM_AMD         "Enable ASM for AMD processors" ON)
+option(WITH_MSR             "Enable MSR mod" ON)
+option(WITH_MSR_ZEN         "Enable MSR mod for AMD Zen-based processors" ON)
 option(WITH_ENV_VARS        "Enable environment variables support in config file" ON)
 option(WITH_EMBEDDED_CONFIG "Enable internal embedded JSON config" OFF)
 option(WITH_OPENCL          "Enable OpenCL backend" ON)

cmake/asm.cmake

@@ -44,9 +44,17 @@ if (WITH_ASM AND NOT XMRIG_ARM AND CMAKE_SIZEOF_VOID_P EQUAL 8)
     set_property(TARGET ${XMRIG_ASM_LIBRARY} PROPERTY LINKER_LANGUAGE C)
 
     add_definitions(/DXMRIG_FEATURE_ASM)
+    if (WITH_ASM_AMD)
+        add_definitions(/DXMRIG_FEATURE_ASM_AMD)
+        message("-- WITH_ASM=ON (+amd)")
+    else()
+        message("-- WITH_ASM=ON (-amd)")
+    endif()
 else()
     set(XMRIG_ASM_SOURCES "")
     set(XMRIG_ASM_LIBRARY "")
 
     remove_definitions(/DXMRIG_FEATURE_ASM)
+    remove_definitions(/DXMRIG_FEATURE_ASM_AMD)
+    message("-- WITH_ASM=OFF")
 endif()

cmake/randomx.cmake

@@ -104,8 +104,13 @@ if (WITH_RANDOMX)
 
     if (WITH_MSR AND NOT XMRIG_ARM AND CMAKE_SIZEOF_VOID_P EQUAL 8 AND (XMRIG_OS_WIN OR XMRIG_OS_LINUX))
         add_definitions(/DXMRIG_FEATURE_MSR)
-        add_definitions(/DXMRIG_FIX_RYZEN)
-        message("-- WITH_MSR=ON")
+        if (WITH_MSR_ZEN)
+            add_definitions(/DXMRIG_FIX_RYZEN)
+            message("-- WITH_MSR=ON (+zen)")
+        else()
+            remove_definitions(/DXMRIG_FIX_RYZEN)
+            message("-- WITH_MSR=ON (-zen)")
+        endif()
 
         if (XMRIG_OS_WIN)
             list(APPEND SOURCES_CRYPTO

src/backend/cuda/runners/CudaRxRunner.cpp

@@ -5,8 +5,8 @@
  * 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 2018-2019 SChernykh   <https://github.com/SChernykh>
- * Copyright 2016-2019 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ * Copyright 2018-2024 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2024 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
@@ -22,7 +22,6 @@
  *   along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-
 #include "backend/cuda/runners/CudaRxRunner.h"
 #include "backend/cuda/CudaLaunchData.h"
 #include "backend/cuda/wrappers/CudaLib.h"
@@ -55,12 +54,21 @@ bool xmrig::CudaRxRunner::run(uint32_t startNonce, uint32_t *rescount, uint32_t
 
 bool xmrig::CudaRxRunner::set(const Job &job, uint8_t *blob)
 {
+    if (!m_datasetHost && (m_seed != job.seed())) {
+        m_seed = job.seed();
+
+        if (m_ready) {
+            const auto *dataset = Rx::dataset(job, 0);
+            callWrapper(CudaLib::rxUpdateDataset(m_ctx, dataset->raw(), dataset->size(false)));
+        }
+    }
+
     const bool rc = CudaBaseRunner::set(job, blob);
     if (!rc || m_ready) {
         return rc;
     }
 
-    auto dataset = Rx::dataset(job, 0);
+    const auto *dataset = Rx::dataset(job, 0);
     m_ready = callWrapper(CudaLib::rxPrepare(m_ctx, dataset->raw(), dataset->size(false), m_datasetHost, m_intensity));
 
     return m_ready;

src/backend/cuda/runners/CudaRxRunner.h

@@ -5,8 +5,8 @@
  * 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 2018-2019 SChernykh   <https://github.com/SChernykh>
- * Copyright 2016-2019 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ * Copyright 2018-2024 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2024 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
@@ -27,6 +27,7 @@
 
 
 #include "backend/cuda/runners/CudaBaseRunner.h"
+#include "base/tools/Buffer.h"
 
 
 namespace xmrig {
@@ -46,6 +47,7 @@ protected:
 private:
     bool m_ready             = false;
     const bool m_datasetHost = false;
+    Buffer m_seed;
     size_t m_intensity       = 0;
 };
 

src/backend/cuda/wrappers/CudaLib.cpp

@@ -19,10 +19,10 @@
 #include <stdexcept>
 #include <uv.h>
 
-
 #include "backend/cuda/wrappers/CudaLib.h"
 #include "base/io/Env.h"
 #include "base/io/log/Log.h"
+#include "base/io/log/Tags.h"
 #include "base/kernel/Process.h"
 #include "crypto/rx/RxAlgo.h"
 
@@ -68,6 +68,7 @@ static const char *kPluginVersion                       = "pluginVersion";
 static const char *kRelease                             = "release";
 static const char *kRxHash                              = "rxHash";
 static const char *kRxPrepare                           = "rxPrepare";
+static const char *kRxUpdateDataset                     = "rxUpdateDataset";
 static const char *kSetJob                              = "setJob";
 static const char *kSetJob_v2                           = "setJob_v2";
 static const char *kVersion                             = "version";
@@ -92,6 +93,7 @@ using pluginVersion_t                                   = const char * (*)();
 using release_t                                         = void (*)(nvid_ctx *);
 using rxHash_t                                          = bool (*)(nvid_ctx *, uint32_t, uint64_t, uint32_t *, uint32_t *);
 using rxPrepare_t                                       = bool (*)(nvid_ctx *, const void *, size_t, bool, uint32_t);
+using rxUpdateDataset_t                                 = bool (*)(nvid_ctx *, const void *, size_t);
 using setJob_t                                          = bool (*)(nvid_ctx *, const void *, size_t, uint32_t);
 using setJob_v2_t                                       = bool (*)(nvid_ctx *, const void *, size_t, const char *);
 using version_t                                         = uint32_t (*)(Version);
@@ -116,6 +118,7 @@ static pluginVersion_t pPluginVersion                   = nullptr;
 static release_t pRelease                               = nullptr;
 static rxHash_t pRxHash                                 = nullptr;
 static rxPrepare_t pRxPrepare                           = nullptr;
+static rxUpdateDataset_t pRxUpdateDataset               = nullptr;
 static setJob_t pSetJob                                 = nullptr;
 static setJob_v2_t pSetJob_v2                           = nullptr;
 static version_t pVersion                               = nullptr;
@@ -202,10 +205,26 @@ bool xmrig::CudaLib::rxHash(nvid_ctx *ctx, uint32_t startNonce, uint64_t target,
 
 bool xmrig::CudaLib::rxPrepare(nvid_ctx *ctx, const void *dataset, size_t datasetSize, bool dataset_host, uint32_t batchSize) noexcept
 {
+#   ifdef XMRIG_ALGO_RANDOMX
+    if (!pRxUpdateDataset) {
+        LOG_WARN("%s" YELLOW_BOLD("CUDA plugin is outdated. Please update to the latest version"), Tags::randomx());
+    }
+#   endif
+
     return pRxPrepare(ctx, dataset, datasetSize, dataset_host, batchSize);
 }
 
 
+bool xmrig::CudaLib::rxUpdateDataset(nvid_ctx *ctx, const void *dataset, size_t datasetSize) noexcept
+{
+    if (pRxUpdateDataset) {
+        return pRxUpdateDataset(ctx, dataset, datasetSize);
+    }
+
+    return true;
+}
+
+
 bool xmrig::CudaLib::kawPowHash(nvid_ctx *ctx, uint8_t* job_blob, uint64_t target, uint32_t *rescount, uint32_t *resnonce, uint32_t *skipped_hashes) noexcept
 {
     return pKawPowHash(ctx, job_blob, target, rescount, resnonce, skipped_hashes);
@@ -401,5 +420,7 @@ void xmrig::CudaLib::load()
         DLSYM(SetJob_v2);
     }
 
+    uv_dlsym(&cudaLib, kRxUpdateDataset, reinterpret_cast<void**>(&pRxUpdateDataset));
+
     pInit();
 }

src/backend/cuda/wrappers/CudaLib.h

@@ -1,6 +1,6 @@
 /* XMRig
- * Copyright (c) 2018-2021 SChernykh   <https://github.com/SChernykh>
- * Copyright (c) 2016-2021 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ * Copyright (c) 2018-2024 SChernykh   <https://github.com/SChernykh>
+ * Copyright (c) 2016-2024 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
@@ -71,6 +71,7 @@ public:
     static bool deviceInit(nvid_ctx *ctx) noexcept;
     static bool rxHash(nvid_ctx *ctx, uint32_t startNonce, uint64_t target, uint32_t *rescount, uint32_t *resnonce) noexcept;
     static bool rxPrepare(nvid_ctx *ctx, const void *dataset, size_t datasetSize, bool dataset_host, uint32_t batchSize) noexcept;
+    static bool rxUpdateDataset(nvid_ctx *ctx, const void *dataset, size_t datasetSize) noexcept;
     static bool kawPowHash(nvid_ctx *ctx, uint8_t* job_blob, uint64_t target, uint32_t *rescount, uint32_t *resnonce, uint32_t *skipped_hashes) noexcept;
     static bool kawPowPrepare(nvid_ctx *ctx, const void* cache, size_t cache_size, const void* dag_precalc, size_t dag_size, uint32_t height, const uint64_t* dag_sizes) noexcept;
     static bool kawPowStopHash(nvid_ctx *ctx) noexcept;

src/core/config/usage.h

@@ -94,7 +94,13 @@ static inline const std::string &usage()
 #   ifdef XMRIG_ALGO_RANDOMX
     u += "      --huge-pages-jit          enable huge pages support for RandomX JIT code\n";
 #   endif
+#   ifdef XMRIG_FEATURE_ASM
+#   ifdef XMRIG_FEATURE_ASM_AMD
     u += "      --asm=ASM                 ASM optimizations, possible values: auto, none, intel, ryzen, bulldozer\n";
+#   else
+    u += "      --asm=ASM                 ASM optimizations, possible values: auto, none, intel\n";
+#   endif
+#   endif
 
 #   if defined(__x86_64__) || defined(_M_AMD64)
     u += "      --argon2-impl=IMPL        argon2 implementation: x86_64, SSE2, SSSE3, XOP, AVX2, AVX-512F\n";

src/crypto/cn/CnHash.cpp

@@ -55,6 +55,7 @@ bool cn_vaes_enabled = false;
 
 
 #ifdef XMRIG_FEATURE_ASM
+#ifdef XMRIG_FEATURE_ASM_AMD
 #   define ADD_FN_ASM(algo) do {                                                                                    \
         m_map[algo]->data[AV_SINGLE][Assembly::INTEL]     = cryptonight_single_hash_asm<algo, Assembly::INTEL>;     \
         m_map[algo]->data[AV_SINGLE][Assembly::RYZEN]     = cryptonight_single_hash_asm<algo, Assembly::RYZEN>;     \
@@ -63,34 +64,50 @@ bool cn_vaes_enabled = false;
         m_map[algo]->data[AV_DOUBLE][Assembly::RYZEN]     = cryptonight_double_hash_asm<algo, Assembly::RYZEN>;     \
         m_map[algo]->data[AV_DOUBLE][Assembly::BULLDOZER] = cryptonight_double_hash_asm<algo, Assembly::BULLDOZER>; \
     } while (0)
+#else
+#   define ADD_FN_ASM(algo) do {                                                                                    \
+        m_map[algo]->data[AV_SINGLE][Assembly::INTEL]     = cryptonight_single_hash_asm<algo, Assembly::INTEL>;     \
+        m_map[algo]->data[AV_DOUBLE][Assembly::INTEL]     = cryptonight_double_hash_asm<algo, Assembly::INTEL>;     \
+    } while (0)
+#endif
 
 
 namespace xmrig {
 
 
 cn_mainloop_fun        cn_half_mainloop_ivybridge_asm             = nullptr;
+#ifdef XMRIG_FEATURE_ASM_AMD
 cn_mainloop_fun        cn_half_mainloop_ryzen_asm                 = nullptr;
 cn_mainloop_fun        cn_half_mainloop_bulldozer_asm             = nullptr;
+#endif
 cn_mainloop_fun        cn_half_double_mainloop_sandybridge_asm    = nullptr;
 
 cn_mainloop_fun        cn_trtl_mainloop_ivybridge_asm             = nullptr;
+#ifdef XMRIG_FEATURE_ASM_AMD
 cn_mainloop_fun        cn_trtl_mainloop_ryzen_asm                 = nullptr;
 cn_mainloop_fun        cn_trtl_mainloop_bulldozer_asm             = nullptr;
+#endif
 cn_mainloop_fun        cn_trtl_double_mainloop_sandybridge_asm    = nullptr;
 
 cn_mainloop_fun        cn_tlo_mainloop_ivybridge_asm              = nullptr;
+#ifdef XMRIG_FEATURE_ASM_AMD
 cn_mainloop_fun        cn_tlo_mainloop_ryzen_asm                  = nullptr;
 cn_mainloop_fun        cn_tlo_mainloop_bulldozer_asm              = nullptr;
+#endif
 cn_mainloop_fun        cn_tlo_double_mainloop_sandybridge_asm     = nullptr;
 
 cn_mainloop_fun        cn_zls_mainloop_ivybridge_asm              = nullptr;
+#ifdef XMRIG_FEATURE_ASM_AMD
 cn_mainloop_fun        cn_zls_mainloop_ryzen_asm                  = nullptr;
 cn_mainloop_fun        cn_zls_mainloop_bulldozer_asm              = nullptr;
+#endif
 cn_mainloop_fun        cn_zls_double_mainloop_sandybridge_asm     = nullptr;
 
 cn_mainloop_fun        cn_double_mainloop_ivybridge_asm           = nullptr;
+#ifdef XMRIG_FEATURE_ASM_AMD
 cn_mainloop_fun        cn_double_mainloop_ryzen_asm               = nullptr;
 cn_mainloop_fun        cn_double_mainloop_bulldozer_asm           = nullptr;
+#endif
 cn_mainloop_fun        cn_double_double_mainloop_sandybridge_asm  = nullptr;
 
 cn_mainloop_fun        cn_upx2_mainloop_asm                       = nullptr;
@@ -160,31 +177,41 @@ static void patchAsmVariants()
     auto base = static_cast<uint8_t *>(VirtualMemory::allocateExecutableMemory(allocation_size, false));
 
     cn_half_mainloop_ivybridge_asm              = reinterpret_cast<cn_mainloop_fun>         (base + 0x0000);
+#   ifdef XMRIG_FEATURE_ASM_AMD
     cn_half_mainloop_ryzen_asm                  = reinterpret_cast<cn_mainloop_fun>         (base + 0x1000);
     cn_half_mainloop_bulldozer_asm              = reinterpret_cast<cn_mainloop_fun>         (base + 0x2000);
+#   endif
     cn_half_double_mainloop_sandybridge_asm     = reinterpret_cast<cn_mainloop_fun>         (base + 0x3000);
 
 #   ifdef XMRIG_ALGO_CN_PICO
     cn_trtl_mainloop_ivybridge_asm              = reinterpret_cast<cn_mainloop_fun>         (base + 0x4000);
+#   ifdef XMRIG_FEATURE_ASM_AMD
     cn_trtl_mainloop_ryzen_asm                  = reinterpret_cast<cn_mainloop_fun>         (base + 0x5000);
     cn_trtl_mainloop_bulldozer_asm              = reinterpret_cast<cn_mainloop_fun>         (base + 0x6000);
+#   endif
     cn_trtl_double_mainloop_sandybridge_asm     = reinterpret_cast<cn_mainloop_fun>         (base + 0x7000);
 #   endif
 
     cn_zls_mainloop_ivybridge_asm               = reinterpret_cast<cn_mainloop_fun>         (base + 0x8000);
+#   ifdef XMRIG_FEATURE_ASM_AMD
     cn_zls_mainloop_ryzen_asm                   = reinterpret_cast<cn_mainloop_fun>         (base + 0x9000);
     cn_zls_mainloop_bulldozer_asm               = reinterpret_cast<cn_mainloop_fun>         (base + 0xA000);
+#   endif
     cn_zls_double_mainloop_sandybridge_asm      = reinterpret_cast<cn_mainloop_fun>         (base + 0xB000);
 
     cn_double_mainloop_ivybridge_asm            = reinterpret_cast<cn_mainloop_fun>         (base + 0xC000);
+#   ifdef XMRIG_FEATURE_ASM_AMD
     cn_double_mainloop_ryzen_asm                = reinterpret_cast<cn_mainloop_fun>         (base + 0xD000);
     cn_double_mainloop_bulldozer_asm            = reinterpret_cast<cn_mainloop_fun>         (base + 0xE000);
+#   endif
     cn_double_double_mainloop_sandybridge_asm   = reinterpret_cast<cn_mainloop_fun>         (base + 0xF000);
 
 #   ifdef XMRIG_ALGO_CN_PICO
     cn_tlo_mainloop_ivybridge_asm               = reinterpret_cast<cn_mainloop_fun>         (base + 0x10000);
+#   ifdef XMRIG_FEATURE_ASM_AMD
     cn_tlo_mainloop_ryzen_asm                   = reinterpret_cast<cn_mainloop_fun>         (base + 0x11000);
     cn_tlo_mainloop_bulldozer_asm               = reinterpret_cast<cn_mainloop_fun>         (base + 0x12000);
+#   endif
     cn_tlo_double_mainloop_sandybridge_asm      = reinterpret_cast<cn_mainloop_fun>         (base + 0x13000);
 #   endif
 
@@ -220,8 +247,10 @@ static void patchAsmVariants()
         constexpr uint32_t ITER = CnAlgo<Algorithm::CN_HALF>().iterations();
 
         patchCode(cn_half_mainloop_ivybridge_asm,            cnv2_mainloop_ivybridge_asm,           ITER);
+#       ifdef XMRIG_FEATURE_ASM_AMD
         patchCode(cn_half_mainloop_ryzen_asm,                cnv2_mainloop_ryzen_asm,               ITER);
         patchCode(cn_half_mainloop_bulldozer_asm,            cnv2_mainloop_bulldozer_asm,           ITER);
+#       endif
         patchCode(cn_half_double_mainloop_sandybridge_asm,   cnv2_double_mainloop_sandybridge_asm,  ITER);
     }
 
@@ -231,8 +260,10 @@ static void patchAsmVariants()
         constexpr uint32_t MASK = CnAlgo<Algorithm::CN_PICO_0>().mask();
 
         patchCode(cn_trtl_mainloop_ivybridge_asm,            cnv2_mainloop_ivybridge_asm,           ITER,   MASK);
+#       ifdef XMRIG_FEATURE_ASM_AMD
         patchCode(cn_trtl_mainloop_ryzen_asm,                cnv2_mainloop_ryzen_asm,               ITER,   MASK);
         patchCode(cn_trtl_mainloop_bulldozer_asm,            cnv2_mainloop_bulldozer_asm,           ITER,   MASK);
+#       endif
         patchCode(cn_trtl_double_mainloop_sandybridge_asm,   cnv2_double_mainloop_sandybridge_asm,  ITER,   MASK);
     }
 
@@ -241,8 +272,10 @@ static void patchAsmVariants()
         constexpr uint32_t MASK = CnAlgo<Algorithm::CN_PICO_TLO>().mask();
 
         patchCode(cn_tlo_mainloop_ivybridge_asm,             cnv2_mainloop_ivybridge_asm,           ITER,   MASK);
+#       ifdef XMRIG_FEATURE_ASM_AMD
         patchCode(cn_tlo_mainloop_ryzen_asm,                 cnv2_mainloop_ryzen_asm,               ITER,   MASK);
         patchCode(cn_tlo_mainloop_bulldozer_asm,             cnv2_mainloop_bulldozer_asm,           ITER,   MASK);
+#       endif
         patchCode(cn_tlo_double_mainloop_sandybridge_asm,    cnv2_double_mainloop_sandybridge_asm,  ITER,   MASK);
     }
 #   endif
@@ -251,8 +284,10 @@ static void patchAsmVariants()
         constexpr uint32_t ITER = CnAlgo<Algorithm::CN_ZLS>().iterations();
 
         patchCode(cn_zls_mainloop_ivybridge_asm,             cnv2_mainloop_ivybridge_asm,           ITER);
+#       ifdef XMRIG_FEATURE_ASM_AMD
         patchCode(cn_zls_mainloop_ryzen_asm,                 cnv2_mainloop_ryzen_asm,               ITER);
         patchCode(cn_zls_mainloop_bulldozer_asm,             cnv2_mainloop_bulldozer_asm,           ITER);
+#       endif
         patchCode(cn_zls_double_mainloop_sandybridge_asm,    cnv2_double_mainloop_sandybridge_asm,  ITER);
     }
 
@@ -260,8 +295,10 @@ static void patchAsmVariants()
         constexpr uint32_t ITER = CnAlgo<Algorithm::CN_DOUBLE>().iterations();
 
         patchCode(cn_double_mainloop_ivybridge_asm,          cnv2_mainloop_ivybridge_asm,           ITER);
+#       ifdef XMRIG_FEATURE_ASM_AMD
         patchCode(cn_double_mainloop_ryzen_asm,              cnv2_mainloop_ryzen_asm,               ITER);
         patchCode(cn_double_mainloop_bulldozer_asm,          cnv2_mainloop_bulldozer_asm,           ITER);
+#       endif
         patchCode(cn_double_double_mainloop_sandybridge_asm, cnv2_double_mainloop_sandybridge_asm,  ITER);
     }
 

src/crypto/cn/CryptoNight_x86.h

@@ -852,12 +852,16 @@ extern "C" void cnv1_single_mainloop_asm(cryptonight_ctx * *ctx);
 extern "C" void cnv1_double_mainloop_asm(cryptonight_ctx **ctx);
 extern "C" void cnv1_quad_mainloop_asm(cryptonight_ctx **ctx);
 extern "C" void cnv2_mainloop_ivybridge_asm(cryptonight_ctx **ctx);
+#ifdef XMRIG_FEATURE_ASM_AMD
 extern "C" void cnv2_mainloop_ryzen_asm(cryptonight_ctx **ctx);
 extern "C" void cnv2_mainloop_bulldozer_asm(cryptonight_ctx **ctx);
+#endif
 extern "C" void cnv2_double_mainloop_sandybridge_asm(cryptonight_ctx **ctx);
 extern "C" void cnv2_rwz_mainloop_asm(cryptonight_ctx **ctx);
 extern "C" void cnv2_rwz_double_mainloop_asm(cryptonight_ctx **ctx);
+#ifdef XMRIG_FEATURE_ASM_AMD
 extern "C" void cnv2_upx_double_mainloop_zen3_asm(cryptonight_ctx **ctx);
+#endif
 
 
 namespace xmrig {
@@ -867,28 +871,38 @@ typedef void (*cn_mainloop_fun)(cryptonight_ctx **ctx);
 
 
 extern cn_mainloop_fun cn_half_mainloop_ivybridge_asm;
+#ifdef XMRIG_FEATURE_ASM_AMD
 extern cn_mainloop_fun cn_half_mainloop_ryzen_asm;
 extern cn_mainloop_fun cn_half_mainloop_bulldozer_asm;
+#endif
 extern cn_mainloop_fun cn_half_double_mainloop_sandybridge_asm;
 
 extern cn_mainloop_fun cn_trtl_mainloop_ivybridge_asm;
+#ifdef XMRIG_FEATURE_ASM_AMD
 extern cn_mainloop_fun cn_trtl_mainloop_ryzen_asm;
 extern cn_mainloop_fun cn_trtl_mainloop_bulldozer_asm;
+#endif
 extern cn_mainloop_fun cn_trtl_double_mainloop_sandybridge_asm;
 
 extern cn_mainloop_fun cn_tlo_mainloop_ivybridge_asm;
+#ifdef XMRIG_FEATURE_ASM_AMD
 extern cn_mainloop_fun cn_tlo_mainloop_ryzen_asm;
 extern cn_mainloop_fun cn_tlo_mainloop_bulldozer_asm;
+#endif
 extern cn_mainloop_fun cn_tlo_double_mainloop_sandybridge_asm;
 
 extern cn_mainloop_fun cn_zls_mainloop_ivybridge_asm;
+#ifdef XMRIG_FEATURE_ASM_AMD
 extern cn_mainloop_fun cn_zls_mainloop_ryzen_asm;
 extern cn_mainloop_fun cn_zls_mainloop_bulldozer_asm;
+#endif
 extern cn_mainloop_fun cn_zls_double_mainloop_sandybridge_asm;
 
 extern cn_mainloop_fun cn_double_mainloop_ivybridge_asm;
+#ifdef XMRIG_FEATURE_ASM_AMD
 extern cn_mainloop_fun cn_double_mainloop_ryzen_asm;
 extern cn_mainloop_fun cn_double_mainloop_bulldozer_asm;
+#endif
 extern cn_mainloop_fun cn_double_double_mainloop_sandybridge_asm;
 
 extern cn_mainloop_fun cn_upx2_mainloop_asm;
@@ -964,46 +978,54 @@ inline void cryptonight_single_hash_asm(const uint8_t *__restrict__ input, size_
         if (ASM == Assembly::INTEL) {
             cnv2_mainloop_ivybridge_asm(ctx);
         }
+#       ifdef XMRIG_FEATURE_ASM_AMD
         else if (ASM == Assembly::RYZEN) {
             cnv2_mainloop_ryzen_asm(ctx);
         }
         else {
             cnv2_mainloop_bulldozer_asm(ctx);
         }
+#       endif
     }
     else if (ALGO == Algorithm::CN_HALF) {
         if (ASM == Assembly::INTEL) {
             cn_half_mainloop_ivybridge_asm(ctx);
         }
+#       ifdef XMRIG_FEATURE_ASM_AMD
         else if (ASM == Assembly::RYZEN) {
             cn_half_mainloop_ryzen_asm(ctx);
         }
         else {
             cn_half_mainloop_bulldozer_asm(ctx);
         }
+#       endif
     }
 #   ifdef XMRIG_ALGO_CN_PICO
     else if (ALGO == Algorithm::CN_PICO_0) {
         if (ASM == Assembly::INTEL) {
             cn_trtl_mainloop_ivybridge_asm(ctx);
         }
+#       ifdef XMRIG_FEATURE_ASM_AMD
         else if (ASM == Assembly::RYZEN) {
             cn_trtl_mainloop_ryzen_asm(ctx);
         }
         else {
             cn_trtl_mainloop_bulldozer_asm(ctx);
         }
+#       endif
     }
     else if (ALGO == Algorithm::CN_PICO_TLO) {
         if (ASM == Assembly::INTEL) {
             cn_tlo_mainloop_ivybridge_asm(ctx);
         }
+#       ifdef XMRIG_FEATURE_ASM_AMD
         else if (ASM == Assembly::RYZEN) {
             cn_tlo_mainloop_ryzen_asm(ctx);
         }
         else {
             cn_tlo_mainloop_bulldozer_asm(ctx);
         }
+#       endif
     }
 #   endif
     else if (ALGO == Algorithm::CN_RWZ) {
@@ -1013,23 +1035,27 @@ inline void cryptonight_single_hash_asm(const uint8_t *__restrict__ input, size_
         if (ASM == Assembly::INTEL) {
             cn_zls_mainloop_ivybridge_asm(ctx);
         }
+#       ifdef XMRIG_FEATURE_ASM_AMD
         else if (ASM == Assembly::RYZEN) {
             cn_zls_mainloop_ryzen_asm(ctx);
         }
         else {
             cn_zls_mainloop_bulldozer_asm(ctx);
         }
+#       endif
     }
     else if (ALGO == Algorithm::CN_DOUBLE) {
         if (ASM == Assembly::INTEL) {
             cn_double_mainloop_ivybridge_asm(ctx);
         }
+#       ifdef XMRIG_FEATURE_ASM_AMD
         else if (ASM == Assembly::RYZEN) {
             cn_double_mainloop_ryzen_asm(ctx);
         }
         else {
             cn_double_mainloop_bulldozer_asm(ctx);
         }
+#       endif
     }
 #   ifdef XMRIG_ALGO_CN_FEMTO
     else if (ALGO == Algorithm::CN_UPX2) {
@@ -1094,12 +1120,16 @@ inline void cryptonight_double_hash_asm(const uint8_t *__restrict__ input, size_
 #   endif
 #   ifdef XMRIG_ALGO_CN_FEMTO
     else if (ALGO == Algorithm::CN_UPX2) {
+#       ifdef XMRIG_FEATURE_ASM_AMD
         if (Cpu::info()->arch() == ICpuInfo::ARCH_ZEN3) {
             cnv2_upx_double_mainloop_zen3_asm(ctx);
         }
         else {
             cn_upx2_double_mainloop_asm(ctx);
         }
+#       else
+        cn_upx2_double_mainloop_asm(ctx);
+#       endif
     }
 #   endif
     else if (ALGO == Algorithm::CN_RWZ) {

src/crypto/cn/asm/cn_main_loop.S

@@ -15,12 +15,16 @@
 .global FN_PREFIX(cnv1_double_mainloop_asm)
 .global FN_PREFIX(cnv1_quad_mainloop_asm)
 .global FN_PREFIX(cnv2_mainloop_ivybridge_asm)
+#ifdef XMRIG_FEATURE_ASM_AMD
 .global FN_PREFIX(cnv2_mainloop_ryzen_asm)
 .global FN_PREFIX(cnv2_mainloop_bulldozer_asm)
+#endif
 .global FN_PREFIX(cnv2_double_mainloop_sandybridge_asm)
 .global FN_PREFIX(cnv2_rwz_mainloop_asm)
 .global FN_PREFIX(cnv2_rwz_double_mainloop_asm)
+#ifdef XMRIG_FEATURE_ASM_AMD
 .global FN_PREFIX(cnv2_upx_double_mainloop_zen3_asm)
+#endif
 
 ALIGN(64)
 FN_PREFIX(cnv1_single_mainloop_asm):
@@ -58,6 +62,7 @@ FN_PREFIX(cnv2_mainloop_ivybridge_asm):
 	ret 0
 	mov eax, 3735929054
 
+#ifdef XMRIG_FEATURE_ASM_AMD
 ALIGN(64)
 FN_PREFIX(cnv2_mainloop_ryzen_asm):
 	sub rsp, 48
@@ -75,6 +80,7 @@ FN_PREFIX(cnv2_mainloop_bulldozer_asm):
 	add rsp, 48
 	ret 0
 	mov eax, 3735929054
+#endif
 
 ALIGN(64)
 FN_PREFIX(cnv2_double_mainloop_sandybridge_asm):
@@ -103,6 +109,7 @@ FN_PREFIX(cnv2_rwz_double_mainloop_asm):
 	ret 0
 	mov eax, 3735929054
 
+#ifdef XMRIG_FEATURE_ASM_AMD
 ALIGN(64)
 FN_PREFIX(cnv2_upx_double_mainloop_zen3_asm):
 	sub rsp, 48
@@ -111,6 +118,7 @@ FN_PREFIX(cnv2_upx_double_mainloop_zen3_asm):
 	add rsp, 48
 	ret 0
 	mov eax, 3735929054
+#endif
 
 #if defined(__linux__) && defined(__ELF__)
 .section .note.GNU-stack,"",%progbits

src/crypto/cn/asm/win64/cn_main_loop.S

@@ -5,12 +5,16 @@
 .global cnv1_double_mainloop_asm
 .global cnv1_quad_mainloop_asm
 .global cnv2_mainloop_ivybridge_asm
+#ifdef XMRIG_FEATURE_ASM_AMD
 .global cnv2_mainloop_ryzen_asm
 .global cnv2_mainloop_bulldozer_asm
+#endif
 .global cnv2_double_mainloop_sandybridge_asm
 .global cnv2_rwz_mainloop_asm
 .global cnv2_rwz_double_mainloop_asm
+#ifdef XMRIG_FEATURE_ASM_AMD
 .global cnv2_upx_double_mainloop_zen3_asm
+#endif
 
 ALIGN(64)
 cnv1_single_mainloop_asm:
@@ -36,6 +40,7 @@ cnv2_mainloop_ivybridge_asm:
 	ret 0
 	mov eax, 3735929054
 
+#ifdef XMRIG_FEATURE_ASM_AMD
 ALIGN(64)
 cnv2_mainloop_ryzen_asm:
 	#include "../cn2/cnv2_main_loop_ryzen.inc"
@@ -47,6 +52,7 @@ cnv2_mainloop_bulldozer_asm:
 	#include "../cn2/cnv2_main_loop_bulldozer.inc"
 	ret 0
 	mov eax, 3735929054
+#endif
 
 ALIGN(64)
 cnv2_double_mainloop_sandybridge_asm:
@@ -66,8 +72,10 @@ cnv2_rwz_double_mainloop_asm:
 	ret 0
 	mov eax, 3735929054
 
+#ifdef XMRIG_FEATURE_ASM_AMD
 ALIGN(64)
 cnv2_upx_double_mainloop_zen3_asm:
 	#include "cn2/cnv2_upx_double_mainloop_zen3.inc"
 	ret 0
 	mov eax, 3735929054
+#endif

src/crypto/rx/RxBasicStorage.cpp

@@ -1,7 +1,7 @@
 /* XMRig
  * Copyright (c) 2018-2019 tevador     <tevador@gmail.com>
- * Copyright (c) 2018-2020 SChernykh   <https://github.com/SChernykh>
- * Copyright (c) 2016-2020 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ * Copyright (c) 2018-2024 SChernykh   <https://github.com/SChernykh>
+ * Copyright (c) 2016-2024 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
@@ -17,9 +17,7 @@
  *   along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-
 #include "crypto/rx/RxBasicStorage.h"
-#include "backend/common/Tags.h"
 #include "base/io/log/Log.h"
 #include "base/io/log/Tags.h"
 #include "base/tools/Chrono.h"

src/crypto/rx/RxVm.cpp

@@ -41,10 +41,12 @@ randomx_vm *xmrig::RxVm::create(RxDataset *dataset, uint8_t *scratchpad, bool so
         flags |= RANDOMX_FLAG_JIT;
     }
 
+#   ifdef XMRIG_FEATURE_ASM_AMD
     const auto asmId = assembly == Assembly::AUTO ? Cpu::info()->assembly() : assembly.id();
     if ((asmId == Assembly::RYZEN) || (asmId == Assembly::BULLDOZER)) {
         flags |= RANDOMX_FLAG_AMD;
     }
+#   endif
 
     return randomx_create_vm(static_cast<randomx_flags>(flags), !dataset->get() ? dataset->cache()->get() : nullptr, dataset->get(), scratchpad, node);
 }