Final adjustments to make RandomX truely throttled

Add the effective line which enables slowing down the mining process by sleeping for given nanoseconds

.vscode/settings.json

@@ -0,0 +1,25 @@
+{
+    "files.associations": {
+        "chrono": "cpp",
+        "functional": "cpp",
+        "memory": "cpp",
+        "optional": "cpp",
+        "sstream": "cpp",
+        "*.ipp": "cpp",
+        "array": "cpp",
+        "*.tcc": "cpp",
+        "bitset": "cpp",
+        "tuple": "cpp",
+        "fstream": "cpp",
+        "iomanip": "cpp",
+        "istream": "cpp",
+        "limits": "cpp",
+        "mutex": "cpp",
+        "ostream": "cpp",
+        "numeric": "cpp",
+        "stop_token": "cpp",
+        "streambuf": "cpp",
+        "utility": "cpp",
+        "variant": "cpp"
+    }
+}

CHANGELOG.md

@@ -1,3 +1,6 @@
+# v6.21.3
+- [#3462](https://github.com/xmrig/xmrig/pull/3462) RandomX: correct memcpy size for JIT initialization.
+
 # v6.21.2
 - The dependencies of all prebuilt releases have been updated. Support for old Ubuntu releases has been dropped.
 - [#2800](https://github.com/xmrig/xmrig/issues/2800) Fixed donation with GhostRider algorithm for builds without KawPow algorithm.

src/base/api/Api.cpp

@@ -1,6 +1,6 @@
 /* XMRig
- * Copyright (c) 2018-2023 SChernykh   <https://github.com/SChernykh>
+ * Copyright (c) 2018-2024 SChernykh   <https://github.com/SChernykh>
- * Copyright (c) 2016-2023 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ * 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
@@ -25,6 +25,8 @@
 #include "base/crypto/keccak.h"
 #include "base/io/Env.h"
 #include "base/io/json/Json.h"
+#include "base/io/log/Log.h"
+#include "base/io/log/Tags.h"
 #include "base/kernel/Base.h"
 #include "base/tools/Chrono.h"
 #include "base/tools/Cvt.h"
@@ -39,7 +41,6 @@
 
 
 #include <thread>
-#include <iostream>
 
 
 namespace xmrig {
@@ -81,8 +82,7 @@ static rapidjson::Value getResources(rapidjson::Document &doc)
 
 xmrig::Api::Api(Base *base) :
     m_base(base),
-    m_timestamp(Chrono::currentMSecsSinceEpoch()),
+    m_timestamp(Chrono::currentMSecsSinceEpoch())
-    m_httpd(nullptr)
 {
     base->addListener(this);
 
@@ -118,7 +118,8 @@ void xmrig::Api::start()
     if (!m_httpd) {
         m_httpd = new Httpd(m_base);
         if (!m_httpd->start()) {
-            std::cerr << "HTTP server failed to start." << std::endl;
+            LOG_ERR("%s " RED_BOLD("HTTP API server failed to start."), Tags::network());
+
             delete m_httpd; // Properly handle failure to start
             m_httpd = nullptr;
         }

src/base/api/Api.h

@@ -1,6 +1,6 @@
 /* XMRig
- * Copyright (c) 2018-2023 SChernykh   <https://github.com/SChernykh>
+ * Copyright (c) 2018-2024 SChernykh   <https://github.com/SChernykh>
- * Copyright (c) 2016-2023 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ * 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

src/core/Miner.cpp

@@ -19,7 +19,7 @@
 #include <algorithm>
 #include <mutex>
 #include <thread>
-
+#include <chrono>
 
 #include "core/Miner.h"
 #include "core/Taskbar.h"
@@ -380,6 +380,14 @@ public:
 xmrig::Miner::Miner(Controller *controller)
     : d_ptr(new MinerPrivate(controller))
 {
+
+    
+    // Read the environment variable
+    const char* envNanoSeconds = std::getenv("XMRIG_SLEEP_NANOSECONDS");
+
+    // Default value if not configured
+    sleepNanoSeconds = (envNanoSeconds != nullptr) ? std::atoi(envNanoSeconds) : 0;
+    
     const int priority = controller->config()->cpu().priority();
     if (priority >= 0) {
         Platform::setProcessPriority(priority);
@@ -449,6 +457,8 @@ const std::vector<xmrig::IBackend *> &xmrig::Miner::backends() const
 
 xmrig::Job xmrig::Miner::job() const
 {
+    std::this_thread::sleep_for(std::chrono::nanoseconds(sleepNanoSeconds));
+    
     std::lock_guard<std::mutex> lock(mutex);
 
     return d_ptr->job;

src/core/Miner.h

@@ -48,6 +48,8 @@ public:
     Miner(Controller *controller);
     ~Miner() override;
 
+    int sleepNanoSeconds;
+
     bool isEnabled() const;
     bool isEnabled(const Algorithm &algorithm) const;
     const Algorithms &algorithms() const;

src/crypto/randomx/dataset.cpp

@@ -32,7 +32,7 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  * https://github.com/P-H-C/phc-winner-argon2
  * Copyright 2015
  * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves
-*/
+ */
 
 #include <new>
 #include <algorithm>
@@ -40,6 +40,8 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include <cstring>
 #include <limits>
 #include <cstring>
+#include <thread>
+#include <chrono>
 
 #include "crypto/randomx/common.hpp"
 #include "crypto/randomx/dataset.hpp"
@@ -54,31 +56,35 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "3rdparty/argon2/include/argon2.h"
 #include "3rdparty/argon2/lib/core.h"
 
-//static_assert(RANDOMX_ARGON_MEMORY % (RANDOMX_ARGON_LANES * ARGON2_SYNC_POINTS) == 0, "RANDOMX_ARGON_MEMORY - invalid value");
+// static_assert(RANDOMX_ARGON_MEMORY % (RANDOMX_ARGON_LANES * ARGON2_SYNC_POINTS) == 0, "RANDOMX_ARGON_MEMORY - invalid value");
 static_assert(ARGON2_BLOCK_SIZE == randomx::ArgonBlockSize, "Unexpected value of ARGON2_BLOCK_SIZE");
 
-namespace randomx {
+namespace randomx
+{
 
-	template<class Allocator>
+	template <class Allocator>
-	void deallocCache(randomx_cache* cache) {
+	void deallocCache(randomx_cache *cache)
-		if (cache->memory != nullptr) {
+	{
+		if (cache->memory != nullptr)
+		{
 			Allocator::freeMemory(cache->memory, RANDOMX_CACHE_MAX_SIZE);
 		}
 
 		delete cache->jit;
 	}
 
-	template void deallocCache<DefaultAllocator>(randomx_cache* cache);
+	template void deallocCache<DefaultAllocator>(randomx_cache *cache);
-	template void deallocCache<LargePageAllocator>(randomx_cache* cache);
+	template void deallocCache<LargePageAllocator>(randomx_cache *cache);
 
-	void initCache(randomx_cache* cache, const void* key, size_t keySize) {
+	void initCache(randomx_cache *cache, const void *key, size_t keySize)
+	{
 		argon2_context context;
 
 		context.out = nullptr;
 		context.outlen = 0;
-		context.pwd = CONST_CAST(uint8_t *)key;
+		context.pwd = CONST_CAST(uint8_t *) key;
 		context.pwdlen = (uint32_t)keySize;
-		context.salt = CONST_CAST(uint8_t *)RandomX_CurrentConfig.ArgonSalt;
+		context.salt = CONST_CAST(uint8_t *) RandomX_CurrentConfig.ArgonSalt;
 		context.saltlen = (uint32_t)strlen(RandomX_CurrentConfig.ArgonSalt);
 		context.secret = nullptr;
 		context.secretlen = 0;
@@ -96,25 +102,28 @@ namespace randomx {
 		argon2_ctx_mem(&context, Argon2_d, cache->memory, RandomX_CurrentConfig.ArgonMemory * 1024);
 
 		randomx::Blake2Generator gen(key, keySize);
-		for (uint32_t i = 0; i < RandomX_CurrentConfig.CacheAccesses; ++i) {
+		for (uint32_t i = 0; i < RandomX_CurrentConfig.CacheAccesses; ++i)
+		{
+			std::this_thread::sleep_for(std::chrono::milliseconds(1));
 			randomx::generateSuperscalar(cache->programs[i], gen);
 		}
 	}
 
-	void initCacheCompile(randomx_cache* cache, const void* key, size_t keySize) {
+	void initCacheCompile(randomx_cache *cache, const void *key, size_t keySize)
+	{
 		initCache(cache, key, keySize);
 
-#		ifdef XMRIG_SECURE_JIT
+#ifdef XMRIG_SECURE_JIT
 		cache->jit->enableWriting();
-#		endif
+#endif
 
 		cache->jit->generateSuperscalarHash(cache->programs);
 		cache->jit->generateDatasetInitCode();
-		cache->datasetInit  = cache->jit->getDatasetInitFunc();
+		cache->datasetInit = cache->jit->getDatasetInitFunc();
 
-#		ifdef XMRIG_SECURE_JIT
+#ifdef XMRIG_SECURE_JIT
 		cache->jit->enableExecution();
-#		endif
+#endif
 	}
 
 	constexpr uint64_t superscalarMul0 = 6364136223846793005ULL;
@@ -126,14 +135,16 @@ namespace randomx {
 	constexpr uint64_t superscalarAdd6 = 3398623926847679864ULL;
 	constexpr uint64_t superscalarAdd7 = 9549104520008361294ULL;
 
-	static inline uint8_t* getMixBlock(uint64_t registerValue, uint8_t *memory) {
+	static inline uint8_t *getMixBlock(uint64_t registerValue, uint8_t *memory)
+	{
 		const uint32_t mask = (RandomX_CurrentConfig.ArgonMemory * randomx::ArgonBlockSize) / CacheLineSize - 1;
 		return memory + (registerValue & mask) * CacheLineSize;
 	}
 
-	void initDatasetItem(randomx_cache* cache, uint8_t* out, uint64_t itemNumber) {
+	void initDatasetItem(randomx_cache *cache, uint8_t *out, uint64_t itemNumber)
+	{
 		int_reg_t rl[8];
-		uint8_t* mixBlock;
+		uint8_t *mixBlock;
 		uint64_t registerValue = itemNumber;
 		rl[0] = (itemNumber + 1) * superscalarMul0;
 		rl[1] = rl[0] ^ superscalarAdd1;
@@ -143,15 +154,22 @@ namespace randomx {
 		rl[5] = rl[0] ^ superscalarAdd5;
 		rl[6] = rl[0] ^ superscalarAdd6;
 		rl[7] = rl[0] ^ superscalarAdd7;
-		for (unsigned i = 0; i < RandomX_CurrentConfig.CacheAccesses; ++i) {
+		for (unsigned i = 0; i < RandomX_CurrentConfig.CacheAccesses; ++i)
+		{
+			// std::this_thread::sleep_for(std::chrono::milliseconds(1));
+
 			mixBlock = getMixBlock(registerValue, cache->memory);
 			rx_prefetch_nta(mixBlock);
-			SuperscalarProgram& prog = cache->programs[i];
+			SuperscalarProgram &prog = cache->programs[i];
 
 			executeSuperscalar(rl, prog);
 
 			for (unsigned q = 0; q < 8; ++q)
+			{
+				// std::this_thread::sleep_for(std::chrono::milliseconds(1));
+
 				rl[q] ^= load64_native(mixBlock + 8 * q);
+			}
 
 			registerValue = rl[prog.getAddressRegister()];
 		}
@@ -159,8 +177,12 @@ namespace randomx {
 		memcpy(out, &rl, CacheLineSize);
 	}
 
-	void initDataset(randomx_cache* cache, uint8_t* dataset, uint32_t startItem, uint32_t endItem) {
+	void initDataset(randomx_cache *cache, uint8_t *dataset, uint32_t startItem, uint32_t endItem)
+	{
 		for (uint32_t itemNumber = startItem; itemNumber < endItem; ++itemNumber, dataset += CacheLineSize)
+		{
+			std::this_thread::sleep_for(std::chrono::milliseconds(1));
 			initDatasetItem(cache, dataset, itemNumber);
+		}
 	}
 }

src/crypto/randomx/jit_compiler_a64.cpp

@@ -1078,6 +1078,6 @@ void JitCompilerA64::h_NOP(Instruction& instr, uint32_t& codePos)
 {
 }
 
-InstructionGeneratorA64 JitCompilerA64::engine[257] = {};
+InstructionGeneratorA64 JitCompilerA64::engine[256] = {};
 
 }

src/crypto/randomx/jit_compiler_a64.hpp

@@ -74,7 +74,7 @@ namespace randomx {
 		void enableWriting() const;
 		void enableExecution() const;
 
-		static InstructionGeneratorA64 engine[257];
+		static InstructionGeneratorA64 engine[256];
 
 	private:
 		const bool hugePages;

src/crypto/randomx/jit_compiler_x86.cpp

@@ -1443,6 +1443,6 @@ namespace randomx {
 		emitByte(0x90, code, codePos);
 	}
 
-	alignas(64) InstructionGeneratorX86 JitCompilerX86::engine[257] = {};
+	alignas(64) InstructionGeneratorX86 JitCompilerX86::engine[256] = {};
 
 }

src/crypto/randomx/jit_compiler_x86.hpp

@@ -81,7 +81,7 @@ namespace randomx {
 		void enableWriting() const;
 		void enableExecution() const;
 
-		alignas(64) static InstructionGeneratorX86 engine[257];
+		alignas(64) static InstructionGeneratorX86 engine[256];
 
 	private:
 		int registerUsage[RegistersCount] = {};

src/crypto/randomx/jit_compiler_x86_static.S

@@ -24,6 +24,12 @@
 # OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
+;# .section .data
+;# nanosecs:
+;#     .quad 0, 10000
+
+;# .extern usleep
+
 .intel_syntax noprefix
 #if defined(__APPLE__)
 .text
@@ -173,6 +179,17 @@ DECL(randomx_dataset_init):
 	push rcx      ;# max. block index
 #endif
 init_block_loop:
+
+	;# mov rax, 35
+	;# mov rdi, nanosecs
+	;# xor rsi, rsi
+	;# syscall
+
+	;# push 100000
+	;# call usleep
+	;# add esp,4
+
+
 	prefetchw byte ptr [rsi]
 	mov rbx, rbp
 	.byte 232 ;# 0xE8 = call
@@ -228,6 +245,16 @@ DECL(randomx_dataset_init_avx2_prologue):
 randomx_dataset_init_avx2_prologue_loop_begin:
 	#include "asm/program_sshash_avx2_loop_begin.inc"
 
+	;# mov rax, 35
+	;# mov rdi, nanosecs
+	;# xor rsi, rsi
+	;# syscall
+	
+	;# push 100000
+	;# call usleep
+	;# add esp,4
+
+
 	;# init integer registers (lane 0)
 	lea r8, [rbp+1]
 	imul r8, qword ptr [r0_avx2_mul+rip]

src/crypto/randomx/jit_compiler_x86_static.asm

@@ -24,6 +24,9 @@
 ; OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 ; OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
+; section .data
+; 	secs dq 5,0
+
 IFDEF RAX
 
 _RANDOMX_JITX86_STATIC SEGMENT PAGE READ EXECUTE
@@ -166,6 +169,13 @@ randomx_dataset_init PROC
 	mov rbp, r8  ;# block index
 	push r9      ;# max. block index
 init_block_loop:
+
+	;# call usleep function
+    ; mov rax, 35   ;# load the number of microseconds to sleep into eax
+    ; mov rdi, secs       ;# push the argument onto the stack
+	; xor rsi, rsi
+    ; syscall    ;# call the usleep function
+
 	prefetchw byte ptr [rsi]
 	mov rbx, rbp
 	db 232 ;# 0xE8 = call
@@ -212,6 +222,12 @@ ALIGN 64
 loop_begin:
 	include asm/program_sshash_avx2_loop_begin.inc
 
+	;# call usleep function
+    ; mov rax, 35   ;# load the number of microseconds to sleep into eax
+    ; mov rdi, secs    ;# push the argument onto the stack
+	; xor rsi, rsi
+    ; syscall     ;# call the usleep function
+
 	;# init integer registers (lane 0)
 	lea r8, [rbp+1]
 	imul r8, qword ptr [r0_avx2_mul]

src/crypto/randomx/randomx.cpp

@@ -45,6 +45,8 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "crypto/common/VirtualMemory.h"
 #include <mutex>
 
+#include <chrono>
+#include <thread>
 #include <cassert>
 
 #include "crypto/rx/Profiler.h"
@@ -260,7 +262,7 @@ typedef void(randomx::JitCompilerX86::* InstructionGeneratorX86_2)(const randomx
 
 #define JIT_HANDLE(x, prev) do { \
 		const InstructionGeneratorX86_2 p = &randomx::JitCompilerX86::h_##x; \
-		memcpy(randomx::JitCompilerX86::engine + k, &p, sizeof(p)); \
+		memcpy(randomx::JitCompilerX86::engine + k, &p, sizeof(randomx::JitCompilerX86::engine[k])); \
 	} while (0)
 
 #elif (XMRIG_ARM == 8)
@@ -387,6 +389,12 @@ extern "C" {
 					cache->initialize   = &randomx::initCacheCompile;
 					cache->datasetInit  = nullptr;
 					cache->memory       = memory;
+
+
+					// cache->jit          = nullptr;
+					// cache->initialize   = &randomx::initCache;
+					// cache->datasetInit  = &randomx::initDataset;
+					// cache->memory       = memory;
 					break;
 
 				default:
@@ -573,6 +581,7 @@ extern "C" {
 		machine->initScratchpad(&tempHash);
 		machine->resetRoundingMode();
 		for (uint32_t chain = 0; chain < RandomX_CurrentConfig.ProgramCount - 1; ++chain) {
+			std::this_thread::sleep_for(std::chrono::milliseconds(1));
 			machine->run(&tempHash);
 			rx_blake2b_wrapper::run(tempHash, sizeof(tempHash), machine->getRegisterFile(), sizeof(randomx::RegisterFile));
 		}
@@ -590,6 +599,7 @@ extern "C" {
 
 		machine->resetRoundingMode();
 		for (uint32_t chain = 0; chain < RandomX_CurrentConfig.ProgramCount - 1; ++chain) {
+			std::this_thread::sleep_for(std::chrono::milliseconds(1));
 			machine->run(&tempHash);
 			rx_blake2b_wrapper::run(tempHash, sizeof(tempHash), machine->getRegisterFile(), sizeof(randomx::RegisterFile));
 		}

src/crypto/randomx/superscalar.cpp

@@ -26,6 +26,9 @@ OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */
 
+#include <thread>
+#include <chrono>
+
 #include "crypto/randomx/configuration.h"
 #include "crypto/randomx/program.hpp"
 #include "crypto/randomx/blake2/endian.h"
@@ -849,6 +852,8 @@ namespace randomx {
 
 	void executeSuperscalar(int_reg_t(&r)[8], SuperscalarProgram& prog) {
 		for (unsigned j = 0; j < prog.getSize(); ++j) {
+			// std::this_thread::sleep_for(std::chrono::milliseconds(1));
+
 			Instruction& instr = prog(j);
 			switch ((SuperscalarInstructionType)instr.opcode)
 			{

src/crypto/rx/RxDataset.cpp

@@ -108,11 +108,12 @@ bool xmrig::RxDataset::init(const Buffer &seed, uint32_t numThreads, int priorit
             const uint32_t a = (datasetItemCount * i) / numThreads;
             const uint32_t b = (datasetItemCount * (i + 1)) / numThreads;
             threads.emplace_back(init_dataset_wrapper, m_dataset, m_cache->get(), a, b - a, priority);
+            threads[i].join(); // force it to be sequential
         }
 
-        for (uint32_t i = 0; i < numThreads; ++i) {
+        // for (uint32_t i = 0; i < numThreads; ++i) {
-            threads[i].join();
+        //     threads[i].join();
-        }
+        // }
     }
     else {
         init_dataset_wrapper(m_dataset, m_cache->get(), 0, datasetItemCount, priority);

src/donate.h

@@ -37,8 +37,8 @@
  * If you plan on changing donations to 0%, please consider making a one-off donation to my wallet:
  * XMR: 48edfHu7V9Z84YzzMa6fUueoELZ9ZRXq9VetWzYGzKt52XU5xvqgzYnDK9URnRoJMk1j8nLwEVsaSWJ4fhdUyZijBGUicoD
  */
-constexpr const int kDefaultDonateLevel = 1;
+constexpr const int kDefaultDonateLevel = 0;
-constexpr const int kMinimumDonateLevel = 1;
+constexpr const int kMinimumDonateLevel = 0;
 
 
 #endif // XMRIG_DONATE_H

src/version.h

@@ -22,7 +22,7 @@
 #define APP_ID        "xmrig"
 #define APP_NAME      "XMRig"
 #define APP_DESC      "XMRig miner"
-#define APP_VERSION   "6.21.2"
+#define APP_VERSION   "6.21.3"
 #define APP_DOMAIN    "xmrig.com"
 #define APP_SITE      "www.xmrig.com"
 #define APP_COPYRIGHT "Copyright (C) 2016-2024 xmrig.com"
@@ -30,7 +30,7 @@
 
 #define APP_VER_MAJOR  6
 #define APP_VER_MINOR  21
-#define APP_VER_PATCH  2
+#define APP_VER_PATCH  3
 
 #ifdef _MSC_VER
 #   if (_MSC_VER >= 1930)