fzorb/REDACTED-rig
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Merge https://github.com/xmrig/xmrig

Browse source
This commit is contained in:
MoneroOcean 2019-11-13 10:40:40 -08:00
commit da8f0fe394
14 changed files with 234 additions and 165 deletions
Download patch file Download diff file Expand all files Collapse all files

98
CHANGELOG.md
View file

@ -1,111 +1,25 @@
# v4.6.1-beta # v5.0.0
This version is first stable unified 3 in 1 GPU+CPU release, OpenCL support built in in miner and not require additional external dependencies on compile time, NVIDIA CUDA available as external [CUDA plugin](https://github.com/xmrig/xmrig-cuda), for convenient, 3 in 1 downloads with recent CUDA version also provided.
This release based on 4.x.x series and include all features from v4.6.2-beta, changelog below include only the most important changes, [full changelog](doc/CHANGELOG_OLD.md) available separately.
- [#1272](https://github.com/xmrig/xmrig/pull/1272) Optimized hashrate calculation. - [#1272](https://github.com/xmrig/xmrig/pull/1272) Optimized hashrate calculation.
- [#1273](https://github.com/xmrig/xmrig/issues/1273) Fixed crash when use `GET /2/backends` API endpoint with disabled CUDA.
# v4.6.0-beta
- [#1263](https://github.com/xmrig/xmrig/pull/1263) Added new option `dataset_host` for NVIDIA GPUs with less than 4 GB memory (RandomX only). - [#1263](https://github.com/xmrig/xmrig/pull/1263) Added new option `dataset_host` for NVIDIA GPUs with less than 4 GB memory (RandomX only).
# v4.5.0-beta
- Added NVIDIA CUDA support via external [CUDA plugun](https://github.com/xmrig/xmrig-cuda). XMRig now is unified 3 in 1 miner.
# v4.4.0-beta
- [#1068](https://github.com/xmrig/xmrig/pull/1068) Added support for `self-select` stratum protocol extension. - [#1068](https://github.com/xmrig/xmrig/pull/1068) Added support for `self-select` stratum protocol extension.
- [#1240](https://github.com/xmrig/xmrig/pull/1240) Sync with the latest RandomX code.
- [#1241](https://github.com/xmrig/xmrig/issues/1241) Fixed regression with colors on old Windows systems.
- [#1243](https://github.com/xmrig/xmrig/pull/1243) Fixed incorrect OpenCL memory size detection in some cases.
- [#1247](https://github.com/xmrig/xmrig/pull/1247) Fixed ARM64 RandomX code alignment.
- [#1248](https://github.com/xmrig/xmrig/pull/1248) Fixed RandomX code cache cleanup on iOS/Darwin.
# v4.3.1-beta
- Fixed regression in v4.3.0, miner didn't create `cn` mining profile with default config example.
# v4.3.0-beta
- [#1227](https://github.com/xmrig/xmrig/pull/1227) Added new algorithm `rx/arq`, RandomX variant for upcoming ArQmA fork. - [#1227](https://github.com/xmrig/xmrig/pull/1227) Added new algorithm `rx/arq`, RandomX variant for upcoming ArQmA fork.
- [#808](https://github.com/xmrig/xmrig/issues/808#issuecomment-539297156) Added experimental support for persistent memory for CPU mining threads. - [#808](https://github.com/xmrig/xmrig/issues/808#issuecomment-539297156) Added experimental support for persistent memory for CPU mining threads.
- [#1221](https://github.com/xmrig/xmrig/issues/1221) Improved RandomX dataset memory usage and initialization speed for NUMA machines. - [#1221](https://github.com/xmrig/xmrig/issues/1221) Improved RandomX dataset memory usage and initialization speed for NUMA machines.
# v4.2.1-beta
- [#1150](https://github.com/xmrig/xmrig/issues/1150) Fixed build on FreeBSD.
- [#1175](https://github.com/xmrig/xmrig/issues/1175) Fixed support for systems where total count of NUMA nodes not equal usable nodes count. - [#1175](https://github.com/xmrig/xmrig/issues/1175) Fixed support for systems where total count of NUMA nodes not equal usable nodes count.
- [#1199](https://github.com/xmrig/xmrig/issues/1199) Fixed excessive memory allocation for OpenCL threads with low intensity.
- [#1212](https://github.com/xmrig/xmrig/issues/1212) Fixed low RandomX performance after fast algorithm switching.
# v4.2.0-beta
- [#1202](https://github.com/xmrig/xmrig/issues/1202) Fixed algorithm verification in donate strategy.
- Added per pool option `coin` with single possible value `monero` for pools without algorithm negotiation, for upcoming Monero fork.
- Added config option `cpu/max-threads-hint` and command line option `--cpu-max-threads-hint`. - Added config option `cpu/max-threads-hint` and command line option `--cpu-max-threads-hint`.
# v4.1.0-beta
- **OpenCL backend disabled by default.**.
- [#1183](https://github.com/xmrig/xmrig/issues/1183) Fixed compatibility with systemd.
- [#1185](https://github.com/xmrig/xmrig/pull/1185) Added JIT compiler for RandomX on ARMv8. - [#1185](https://github.com/xmrig/xmrig/pull/1185) Added JIT compiler for RandomX on ARMv8.
- Improved API endpoint `GET /2/backends` and added support for this endpoint to [workers.xmrig.info](http://workers.xmrig.info). - Improved API endpoint `GET /2/backends` and added support for this endpoint to [workers.xmrig.info](http://workers.xmrig.info).
- Added command line option `--no-cpu` to disable CPU backend. - Added command line option `--no-cpu` to disable CPU backend.
- Added OpenCL specific command line options: `--opencl`, `--opencl-devices`, `--opencl-platform`, `--opencl-loader` and `--opencl-no-cache`. - Added OpenCL specific command line options: `--opencl`, `--opencl-devices`, `--opencl-platform`, `--opencl-loader` and `--opencl-no-cache`.
- Added CUDA specific command line options: `--cuda`, `--cuda-loader` and `--no-nvml`.
- Removed command line option `--http-enabled`, HTTP API enabled automatically if any other `--http-*` option provided. - Removed command line option `--http-enabled`, HTTP API enabled automatically if any other `--http-*` option provided.
# v4.0.1-beta
- [#1177](https://github.com/xmrig/xmrig/issues/1177) Fixed compatibility with old AMD drivers.
- [#1180](https://github.com/xmrig/xmrig/issues/1180) Fixed possible duplicated shares after algorithm switching.
- Added support for case if not all backend threads successfully started.
- Fixed wrong config file permissions after write (only gcc builds on recent Windows 10 affected).
# v4.0.0-beta
- [#1172](https://github.com/xmrig/xmrig/issues/1172) **Added OpenCL mining backend.** - [#1172](https://github.com/xmrig/xmrig/issues/1172) **Added OpenCL mining backend.**
- [#268](https://github.com/xmrig/xmrig-amd/pull/268) [#270](https://github.com/xmrig/xmrig-amd/pull/270) [#271](https://github.com/xmrig/xmrig-amd/pull/271) [#273](https://github.com/xmrig/xmrig-amd/pull/273) [#274](https://github.com/xmrig/xmrig-amd/pull/274) [#1171](https://github.com/xmrig/xmrig/pull/1171) Added RandomX support for OpenCL, thanks [@SChernykh](https://github.com/SChernykh). - [#268](https://github.com/xmrig/xmrig-amd/pull/268) [#270](https://github.com/xmrig/xmrig-amd/pull/270) [#271](https://github.com/xmrig/xmrig-amd/pull/271) [#273](https://github.com/xmrig/xmrig-amd/pull/273) [#274](https://github.com/xmrig/xmrig-amd/pull/274) [#1171](https://github.com/xmrig/xmrig/pull/1171) Added RandomX support for OpenCL, thanks [@SChernykh](https://github.com/SChernykh).
- Algorithm `cn/wow` removed, as no longer alive. - Algorithm `cn/wow` removed, as no longer alive.
# v3.2.0
- Added per pool option `coin` with single possible value `monero` for pools without algorithm negotiation, for upcoming Monero fork.
- [#1183](https://github.com/xmrig/xmrig/issues/1183) Fixed compatibility with systemd.
# v3.1.3
- [#1180](https://github.com/xmrig/xmrig/issues/1180) Fixed possible duplicated shares after algorithm switching.
- Fixed wrong config file permissions after write (only gcc builds on recent Windows 10 affected).
# v3.1.2
- Many RandomX optimizations and fixes.
- [#1132](https://github.com/xmrig/xmrig/issues/1132) Fixed build on CentOS 7.
- [#1163](https://github.com/xmrig/xmrig/pull/1163) Optimized soft AES code, up to +30% hashrate on CPU without AES support and other optimizations.
- [#1166](https://github.com/xmrig/xmrig/pull/1166) Fixed crash when initialize dataset with big threads count (eg 272).
- [#1168](https://github.com/xmrig/xmrig/pull/1168) Optimized loading from scratchpad.
- [#1128](https://github.com/xmrig/xmrig/issues/1128) Fixed CMake 2.8 compatibility.
# v3.1.1
- [#1133](https://github.com/xmrig/xmrig/issues/1133) Fixed syslog regression.
- [#1138](https://github.com/xmrig/xmrig/issues/1138) Fixed multiple network bugs.
- [#1141](https://github.com/xmrig/xmrig/issues/1141) Fixed log in background mode.
- [#1142](https://github.com/xmrig/xmrig/pull/1142) RandomX hashrate improved by 0.5-1.5% depending on variant and CPU.
- [#1146](https://github.com/xmrig/xmrig/pull/1146) Fixed race condition in RandomX thread init.
- [#1148](https://github.com/xmrig/xmrig/pull/1148) Fixed, on Linux linker marking entire executable as having an executable stack.
- Fixed, for Argon2 algorithms command line options like `--threads` was ignored.
- Fixed command line options for single pool, free order allowed again.
# v3.1.0
- [#1107](https://github.com/xmrig/xmrig/issues/1107#issuecomment-522235892) Added Argon2 algorithm family: `argon2/chukwa` and `argon2/wrkz`.
# v3.0.0
- **[#1111](https://github.com/xmrig/xmrig/pull/1111) Added RandomX (`rx/test`) algorithm for testing and benchmarking.**
- **[#1036](https://github.com/xmrig/xmrig/pull/1036) Added RandomWOW (`rx/wow`) algorithm for [Wownero](http://wownero.org/).**
- **[#1050](https://github.com/xmrig/xmrig/pull/1050) Added RandomXL (`rx/loki`) algorithm for [Loki](https://loki.network/).**
- **[#1077](https://github.com/xmrig/xmrig/issues/1077) Added NUMA support via hwloc**.
- **Added flexible [multi algorithm](doc/CPU.md) configuration.**
- **Added unlimited switching between incompatible algorithms, all mining options can be changed in runtime.**
- [#257](https://github.com/xmrig/xmrig-nvidia/pull/257) New logging subsystem, file and syslog now always without colors.
- [#314](https://github.com/xmrig/xmrig-proxy/issues/314) Added donate over proxy feature.
- [#1007](https://github.com/xmrig/xmrig/issues/1007) Old HTTP API backend based on libmicrohttpd, replaced to custom HTTP server (libuv + http_parser).
- [#1010](https://github.com/xmrig/xmrig/pull/1010#issuecomment-482632107) Added daemon support (solo mining).
- [#1066](https://github.com/xmrig/xmrig/issues/1066#issuecomment-518080529) Added error message if pool not ready for RandomX.
- [#1105](https://github.com/xmrig/xmrig/issues/1105) Improved auto configuration for `cn-pico` algorithm.
- Added commands `pause` and `resume` via JSON RPC 2.0 API (`POST /json_rpc`).
- Added command line option `--export-topology` for export hwloc topology to a XML file.
- Breaked backward compatibility with previous configs and command line, `variant` option replaced to `algo`, global option `algo` removed, all CPU related settings moved to `cpu` object.
- Options `av`, `safe` and `max-cpu-usage` removed.
- Algorithm `cn/msr` renamed to `cn/fast`.
- Algorithm `cn/xtl` removed.
- API endpoint `GET /1/threads` replaced to `GET /2/backends`.
- Added global uptime and extended connection information in API.
- API now return current algorithm.
# Previous versions # Previous versions
[doc/CHANGELOG_OLD.md](doc/CHANGELOG_OLD.md) [doc/CHANGELOG_OLD.md](doc/CHANGELOG_OLD.md)

3
README.md
View file

@ -74,7 +74,7 @@ API:
OpenCL backend: OpenCL backend:
--opencl enable OpenCL mining backend --opencl enable OpenCL mining backend
--opencl-devices=N list of OpenCL devices to use --opencl-devices=N comma separated list of OpenCL devices to use
--opencl-platform=N OpenCL platform index or name --opencl-platform=N OpenCL platform index or name
--opencl-loader=PATH path to OpenCL-ICD-Loader (OpenCL.dll or libOpenCL.so) --opencl-loader=PATH path to OpenCL-ICD-Loader (OpenCL.dll or libOpenCL.so)
--opencl-no-cache disable OpenCL cache --opencl-no-cache disable OpenCL cache
@ -83,6 +83,7 @@ OpenCL backend:
CUDA backend: CUDA backend:
--cuda enable CUDA mining backend --cuda enable CUDA mining backend
--cuda-loader=PATH path to CUDA plugin (xmrig-cuda.dll or libxmrig-cuda.so) --cuda-loader=PATH path to CUDA plugin (xmrig-cuda.dll or libxmrig-cuda.so)
--cuda-devices=N comma separated list of CUDA devices to use
--no-nvml disable NVML (NVIDIA Management Library) support --no-nvml disable NVML (NVIDIA Management Library) support
Logging: Logging:

113
doc/CHANGELOG_OLD.md
View file

@ -1,3 +1,116 @@
# v4.6.2-beta
- [#1274](https://github.com/xmrig/xmrig/issues/1274) Added `--cuda-devices` command line option.
- [#1277](https://github.com/xmrig/xmrig/pull/1277) Fixed function names for clang on Apple.
# v4.6.1-beta
- [#1272](https://github.com/xmrig/xmrig/pull/1272) Optimized hashrate calculation.
- [#1273](https://github.com/xmrig/xmrig/issues/1273) Fixed crash when use `GET /2/backends` API endpoint with disabled CUDA.
# v4.6.0-beta
- [#1263](https://github.com/xmrig/xmrig/pull/1263) Added new option `dataset_host` for NVIDIA GPUs with less than 4 GB memory (RandomX only).
# v4.5.0-beta
- Added NVIDIA CUDA support via external [CUDA plugun](https://github.com/xmrig/xmrig-cuda). XMRig now is unified 3 in 1 miner.
# v4.4.0-beta
- [#1068](https://github.com/xmrig/xmrig/pull/1068) Added support for `self-select` stratum protocol extension.
- [#1240](https://github.com/xmrig/xmrig/pull/1240) Sync with the latest RandomX code.
- [#1241](https://github.com/xmrig/xmrig/issues/1241) Fixed regression with colors on old Windows systems.
- [#1243](https://github.com/xmrig/xmrig/pull/1243) Fixed incorrect OpenCL memory size detection in some cases.
- [#1247](https://github.com/xmrig/xmrig/pull/1247) Fixed ARM64 RandomX code alignment.
- [#1248](https://github.com/xmrig/xmrig/pull/1248) Fixed RandomX code cache cleanup on iOS/Darwin.
# v4.3.1-beta
- Fixed regression in v4.3.0, miner didn't create `cn` mining profile with default config example.
# v4.3.0-beta
- [#1227](https://github.com/xmrig/xmrig/pull/1227) Added new algorithm `rx/arq`, RandomX variant for upcoming ArQmA fork.
- [#808](https://github.com/xmrig/xmrig/issues/808#issuecomment-539297156) Added experimental support for persistent memory for CPU mining threads.
- [#1221](https://github.com/xmrig/xmrig/issues/1221) Improved RandomX dataset memory usage and initialization speed for NUMA machines.
# v4.2.1-beta
- [#1150](https://github.com/xmrig/xmrig/issues/1150) Fixed build on FreeBSD.
- [#1175](https://github.com/xmrig/xmrig/issues/1175) Fixed support for systems where total count of NUMA nodes not equal usable nodes count.
- [#1199](https://github.com/xmrig/xmrig/issues/1199) Fixed excessive memory allocation for OpenCL threads with low intensity.
- [#1212](https://github.com/xmrig/xmrig/issues/1212) Fixed low RandomX performance after fast algorithm switching.
# v4.2.0-beta
- [#1202](https://github.com/xmrig/xmrig/issues/1202) Fixed algorithm verification in donate strategy.
- Added per pool option `coin` with single possible value `monero` for pools without algorithm negotiation, for upcoming Monero fork.
- Added config option `cpu/max-threads-hint` and command line option `--cpu-max-threads-hint`.
# v4.1.0-beta
- **OpenCL backend disabled by default.**.
- [#1183](https://github.com/xmrig/xmrig/issues/1183) Fixed compatibility with systemd.
- [#1185](https://github.com/xmrig/xmrig/pull/1185) Added JIT compiler for RandomX on ARMv8.
- Improved API endpoint `GET /2/backends` and added support for this endpoint to [workers.xmrig.info](http://workers.xmrig.info).
- Added command line option `--no-cpu` to disable CPU backend.
- Added OpenCL specific command line options: `--opencl`, `--opencl-devices`, `--opencl-platform`, `--opencl-loader` and `--opencl-no-cache`.
- Removed command line option `--http-enabled`, HTTP API enabled automatically if any other `--http-*` option provided.
# v4.0.1-beta
- [#1177](https://github.com/xmrig/xmrig/issues/1177) Fixed compatibility with old AMD drivers.
- [#1180](https://github.com/xmrig/xmrig/issues/1180) Fixed possible duplicated shares after algorithm switching.
- Added support for case if not all backend threads successfully started.
- Fixed wrong config file permissions after write (only gcc builds on recent Windows 10 affected).
# v4.0.0-beta
- [#1172](https://github.com/xmrig/xmrig/issues/1172) **Added OpenCL mining backend.**
- [#268](https://github.com/xmrig/xmrig-amd/pull/268) [#270](https://github.com/xmrig/xmrig-amd/pull/270) [#271](https://github.com/xmrig/xmrig-amd/pull/271) [#273](https://github.com/xmrig/xmrig-amd/pull/273) [#274](https://github.com/xmrig/xmrig-amd/pull/274) [#1171](https://github.com/xmrig/xmrig/pull/1171) Added RandomX support for OpenCL, thanks [@SChernykh](https://github.com/SChernykh).
- Algorithm `cn/wow` removed, as no longer alive.
# v3.2.0
- Added per pool option `coin` with single possible value `monero` for pools without algorithm negotiation, for upcoming Monero fork.
- [#1183](https://github.com/xmrig/xmrig/issues/1183) Fixed compatibility with systemd.
# v3.1.3
- [#1180](https://github.com/xmrig/xmrig/issues/1180) Fixed possible duplicated shares after algorithm switching.
- Fixed wrong config file permissions after write (only gcc builds on recent Windows 10 affected).
# v3.1.2
- Many RandomX optimizations and fixes.
- [#1132](https://github.com/xmrig/xmrig/issues/1132) Fixed build on CentOS 7.
- [#1163](https://github.com/xmrig/xmrig/pull/1163) Optimized soft AES code, up to +30% hashrate on CPU without AES support and other optimizations.
- [#1166](https://github.com/xmrig/xmrig/pull/1166) Fixed crash when initialize dataset with big threads count (eg 272).
- [#1168](https://github.com/xmrig/xmrig/pull/1168) Optimized loading from scratchpad.
- [#1128](https://github.com/xmrig/xmrig/issues/1128) Fixed CMake 2.8 compatibility.
# v3.1.1
- [#1133](https://github.com/xmrig/xmrig/issues/1133) Fixed syslog regression.
- [#1138](https://github.com/xmrig/xmrig/issues/1138) Fixed multiple network bugs.
- [#1141](https://github.com/xmrig/xmrig/issues/1141) Fixed log in background mode.
- [#1142](https://github.com/xmrig/xmrig/pull/1142) RandomX hashrate improved by 0.5-1.5% depending on variant and CPU.
- [#1146](https://github.com/xmrig/xmrig/pull/1146) Fixed race condition in RandomX thread init.
- [#1148](https://github.com/xmrig/xmrig/pull/1148) Fixed, on Linux linker marking entire executable as having an executable stack.
- Fixed, for Argon2 algorithms command line options like `--threads` was ignored.
- Fixed command line options for single pool, free order allowed again.
# v3.1.0
- [#1107](https://github.com/xmrig/xmrig/issues/1107#issuecomment-522235892) Added Argon2 algorithm family: `argon2/chukwa` and `argon2/wrkz`.
# v3.0.0
- **[#1111](https://github.com/xmrig/xmrig/pull/1111) Added RandomX (`rx/test`) algorithm for testing and benchmarking.**
- **[#1036](https://github.com/xmrig/xmrig/pull/1036) Added RandomWOW (`rx/wow`) algorithm for [Wownero](http://wownero.org/).**
- **[#1050](https://github.com/xmrig/xmrig/pull/1050) Added RandomXL (`rx/loki`) algorithm for [Loki](https://loki.network/).**
- **[#1077](https://github.com/xmrig/xmrig/issues/1077) Added NUMA support via hwloc**.
- **Added flexible [multi algorithm](doc/CPU.md) configuration.**
- **Added unlimited switching between incompatible algorithms, all mining options can be changed in runtime.**
- [#257](https://github.com/xmrig/xmrig-nvidia/pull/257) New logging subsystem, file and syslog now always without colors.
- [#314](https://github.com/xmrig/xmrig-proxy/issues/314) Added donate over proxy feature.
- [#1007](https://github.com/xmrig/xmrig/issues/1007) Old HTTP API backend based on libmicrohttpd, replaced to custom HTTP server (libuv + http_parser).
- [#1010](https://github.com/xmrig/xmrig/pull/1010#issuecomment-482632107) Added daemon support (solo mining).
- [#1066](https://github.com/xmrig/xmrig/issues/1066#issuecomment-518080529) Added error message if pool not ready for RandomX.
- [#1105](https://github.com/xmrig/xmrig/issues/1105) Improved auto configuration for `cn-pico` algorithm.
- Added commands `pause` and `resume` via JSON RPC 2.0 API (`POST /json_rpc`).
- Added command line option `--export-topology` for export hwloc topology to a XML file.
- Breaked backward compatibility with previous configs and command line, `variant` option replaced to `algo`, global option `algo` removed, all CPU related settings moved to `cpu` object.
- Options `av`, `safe` and `max-cpu-usage` removed.
- Algorithm `cn/msr` renamed to `cn/fast`.
- Algorithm `cn/xtl` removed.
- API endpoint `GET /1/threads` replaced to `GET /2/backends`.
- Added global uptime and extended connection information in API.
- API now return current algorithm.
# v2.99.6-beta # v2.99.6-beta
- Added commands `pause` and `resume` via JSON RPC 2.0 API (`POST /json_rpc`). - Added commands `pause` and `resume` via JSON RPC 2.0 API (`POST /json_rpc`).
- Fixed autoconfig regression (since 2.99.5), mostly `rx/wow` was affected by this bug. - Fixed autoconfig regression (since 2.99.5), mostly `rx/wow` was affected by this bug.

9
src/backend/cuda/CudaBackend.cpp
View file

@ -155,11 +155,14 @@ public:
return; return;
} }
devices = CudaLib::devices(cuda.bfactor(), cuda.bsleep(), cuda.devicesHint());
if (devices.empty()) {
return printDisabled(kLabel, RED_S " (no devices)");
}
Log::print(GREEN_BOLD(" * ") WHITE_BOLD("%-13s") WHITE_BOLD("%s") "/" WHITE_BOLD("%s") BLACK_BOLD("/%s"), kLabel, Log::print(GREEN_BOLD(" * ") WHITE_BOLD("%-13s") WHITE_BOLD("%s") "/" WHITE_BOLD("%s") BLACK_BOLD("/%s"), kLabel,
CudaLib::version(runtimeVersion).c_str(), CudaLib::version(driverVersion).c_str(), CudaLib::pluginVersion()); CudaLib::version(runtimeVersion).c_str(), CudaLib::version(driverVersion).c_str(), CudaLib::pluginVersion());
devices = CudaLib::devices(cuda.bfactor(), cuda.bsleep());
# ifdef XMRIG_FEATURE_NVML # ifdef XMRIG_FEATURE_NVML
if (cuda.isNvmlEnabled()) { if (cuda.isNvmlEnabled()) {
if (NvmlLib::init(cuda.nvmlLoader())) { if (NvmlLib::init(cuda.nvmlLoader())) {
@ -172,7 +175,7 @@ public:
); );
} }
else { else {
printDisabled(kLabel, RED_S " (failed to load NVML)"); printDisabled(kNvmlLabel, RED_S " (failed to load NVML)");
} }
} }
else { else {

19
src/backend/cuda/CudaConfig.cpp
View file

@ -78,6 +78,16 @@ rapidjson::Value xmrig::CudaConfig::toJSON(rapidjson::Document &doc) const
std::vector<xmrig::CudaLaunchData> xmrig::CudaConfig::get(const Miner *miner, const Algorithm &algorithm, const std::vector<CudaDevice> &devices) const std::vector<xmrig::CudaLaunchData> xmrig::CudaConfig::get(const Miner *miner, const Algorithm &algorithm, const std::vector<CudaDevice> &devices) const
{ {
auto deviceIndex = [&devices](uint32_t index) -> int {
for (uint32_t i = 0; i < devices.size(); ++i) {
if (devices[i].index() == index) {
return i;
}
}
return -1;
};
std::vector<CudaLaunchData> out; std::vector<CudaLaunchData> out;
const auto &threads = m_threads.get(algorithm); const auto &threads = m_threads.get(algorithm);
@ -85,15 +95,16 @@ std::vector<xmrig::CudaLaunchData> xmrig::CudaConfig::get(const Miner *miner, co
return out; return out;
} }
out.reserve(threads.count() * 2); out.reserve(threads.count());
for (const auto &thread : threads.data()) { for (const auto &thread : threads.data()) {
if (thread.index() >= devices.size()) { const int index = deviceIndex(thread.index());
if (index == -1) {
LOG_INFO("%s" YELLOW(" skip non-existing device with index ") YELLOW_BOLD("%u"), cuda_tag(), thread.index()); LOG_INFO("%s" YELLOW(" skip non-existing device with index ") YELLOW_BOLD("%u"), cuda_tag(), thread.index());
continue; continue;
} }
out.emplace_back(miner, algorithm, thread, devices[thread.index()]); out.emplace_back(miner, algorithm, thread, devices[static_cast<size_t>(index)]);
} }
return out; return out;
@ -153,7 +164,7 @@ void xmrig::CudaConfig::generate()
return; return;
} }
const auto devices = CudaLib::devices(bfactor(), bsleep()); const auto devices = CudaLib::devices(bfactor(), bsleep(), m_devicesHint);
if (devices.empty()) { if (devices.empty()) {
return; return;
} }

17
src/backend/cuda/CudaConfig.h
View file

@ -43,16 +43,17 @@ public:
std::vector<CudaLaunchData> get(const Miner *miner, const Algorithm &algorithm, const std::vector<CudaDevice> &devices) const; std::vector<CudaLaunchData> get(const Miner *miner, const Algorithm &algorithm, const std::vector<CudaDevice> &devices) const;
void read(const rapidjson::Value &value); void read(const rapidjson::Value &value);
inline bool isEnabled() const { return m_enabled; } inline bool isEnabled() const { return m_enabled; }
inline bool isShouldSave() const { return m_shouldSave; } inline bool isShouldSave() const { return m_shouldSave; }
inline const String &loader() const { return m_loader; } inline const std::vector<uint32_t> &devicesHint() const { return m_devicesHint; }
inline const Threads<CudaThreads> &threads() const { return m_threads; } inline const String &loader() const { return m_loader; }
inline int32_t bfactor() const { return m_bfactor; } inline const Threads<CudaThreads> &threads() const { return m_threads; }
inline int32_t bsleep() const { return m_bsleep; } inline int32_t bfactor() const { return m_bfactor; }
inline int32_t bsleep() const { return m_bsleep; }
# ifdef XMRIG_FEATURE_NVML # ifdef XMRIG_FEATURE_NVML
inline bool isNvmlEnabled() const { return m_nvml; } inline bool isNvmlEnabled() const { return m_nvml; }
inline const String &nvmlLoader() const { return m_nvmlLoader; } inline const String &nvmlLoader() const { return m_nvmlLoader; }
# endif # endif
private: private:

24
src/backend/cuda/wrappers/CudaLib.cpp
View file

@ -209,7 +209,7 @@ std::string xmrig::CudaLib::version(uint32_t version)
} }
std::vector<xmrig::CudaDevice> xmrig::CudaLib::devices(int32_t bfactor, int32_t bsleep) noexcept std::vector<xmrig::CudaDevice> xmrig::CudaLib::devices(int32_t bfactor, int32_t bsleep, const std::vector<uint32_t> &hints) noexcept
{ {
const uint32_t count = deviceCount(); const uint32_t count = deviceCount();
if (!count) { if (!count) {
@ -219,10 +219,24 @@ std::vector<xmrig::CudaDevice> xmrig::CudaLib::devices(int32_t bfactor, int32_t
std::vector<CudaDevice> out; std::vector<CudaDevice> out;
out.reserve(count); out.reserve(count);
for (uint32_t i = 0; i < count; ++i) { if (hints.empty()) {
CudaDevice device(i, bfactor, bsleep); for (uint32_t i = 0; i < count; ++i) {
if (device.isValid()) { CudaDevice device(i, bfactor, bsleep);
out.emplace_back(std::move(device)); if (device.isValid()) {
out.emplace_back(std::move(device));
}
}
}
else {
for (const uint32_t i : hints) {
if (i >= count) {
continue;
}
CudaDevice device(i, bfactor, bsleep);
if (device.isValid()) {
out.emplace_back(std::move(device));
}
} }
} }

2
src/backend/cuda/wrappers/CudaLib.h
View file

@ -85,7 +85,7 @@ public:
static int32_t deviceInt(nvid_ctx *ctx, DeviceProperty property) noexcept; static int32_t deviceInt(nvid_ctx *ctx, DeviceProperty property) noexcept;
static nvid_ctx *alloc(uint32_t id, int32_t bfactor, int32_t bsleep) noexcept; static nvid_ctx *alloc(uint32_t id, int32_t bfactor, int32_t bsleep) noexcept;
static std::string version(uint32_t version); static std::string version(uint32_t version);
static std::vector<CudaDevice> devices(int32_t bfactor, int32_t bsleep) noexcept; static std::vector<CudaDevice> devices(int32_t bfactor, int32_t bsleep, const std::vector<uint32_t> &hints) noexcept;
static uint32_t deviceCount() noexcept; static uint32_t deviceCount() noexcept;
static uint32_t deviceUint(nvid_ctx *ctx, DeviceProperty property) noexcept; static uint32_t deviceUint(nvid_ctx *ctx, DeviceProperty property) noexcept;
static uint32_t driverVersion() noexcept; static uint32_t driverVersion() noexcept;

2
src/base/io/json/Json.cpp
View file

@ -50,7 +50,7 @@ bool xmrig::Json::getBool(const rapidjson::Value &obj, const char *key, bool def
} }
const char *xmrig::Json::getString(const rapidjson::Value &obj, const char *key, const char *defaultValue) const char *xmrig::Json::getString(const rapidjson::Value &obj, const char *key, const char *defaultValue)
{ {
assert(obj.IsObject()); assert(obj.IsObject());

6
src/core/config/ConfigTransform.cpp
View file

@ -187,10 +187,14 @@ void xmrig::ConfigTransform::transform(rapidjson::Document &doc, int key, const
# ifdef XMRIG_FEATURE_CUDA # ifdef XMRIG_FEATURE_CUDA
case IConfig::CudaKey: /* --cuda */ case IConfig::CudaKey: /* --cuda */
return set(doc, kCuda, "enabled", true); return set(doc, kCuda, kEnabled, true);
case IConfig::CudaLoaderKey: /* --cuda-loader */ case IConfig::CudaLoaderKey: /* --cuda-loader */
return set(doc, kCuda, "loader", arg); return set(doc, kCuda, "loader", arg);
case IConfig::CudaDevicesKey: /* --cuda-devices */
set(doc, kCuda, kEnabled, true);
return set(doc, kCuda, "devices-hint", arg);
# endif # endif
# ifdef XMRIG_FEATURE_NVML # ifdef XMRIG_FEATURE_NVML

1
src/core/config/Config_platform.h
View file

@ -109,6 +109,7 @@ static const option options[] = {
# ifdef XMRIG_FEATURE_CUDA # ifdef XMRIG_FEATURE_CUDA
{ "cuda", 0, nullptr, IConfig::CudaKey }, { "cuda", 0, nullptr, IConfig::CudaKey },
{ "cuda-loader", 1, nullptr, IConfig::CudaLoaderKey }, { "cuda-loader", 1, nullptr, IConfig::CudaLoaderKey },
{ "cuda-devices", 1, nullptr, IConfig::CudaDevicesKey },
# endif # endif
# ifdef XMRIG_FEATURE_NVML # ifdef XMRIG_FEATURE_NVML
{ "no-nvml", 0, nullptr, IConfig::NvmlKey }, { "no-nvml", 0, nullptr, IConfig::NvmlKey },

3
src/core/config/usage.h
View file

@ -101,7 +101,7 @@ static inline const std::string &usage()
# ifdef XMRIG_FEATURE_OPENCL # ifdef XMRIG_FEATURE_OPENCL
u += "\nOpenCL backend:\n"; u += "\nOpenCL backend:\n";
u += " --opencl enable OpenCL mining backend\n"; u += " --opencl enable OpenCL mining backend\n";
u += " --opencl-devices=N list of OpenCL devices to use\n"; u += " --opencl-devices=N comma separated list of OpenCL devices to use\n";
u += " --opencl-platform=N OpenCL platform index or name\n"; u += " --opencl-platform=N OpenCL platform index or name\n";
u += " --opencl-loader=PATH path to OpenCL-ICD-Loader (OpenCL.dll or libOpenCL.so)\n"; u += " --opencl-loader=PATH path to OpenCL-ICD-Loader (OpenCL.dll or libOpenCL.so)\n";
u += " --opencl-no-cache disable OpenCL cache\n"; u += " --opencl-no-cache disable OpenCL cache\n";
@ -112,6 +112,7 @@ static inline const std::string &usage()
u += "\nCUDA backend:\n"; u += "\nCUDA backend:\n";
u += " --cuda enable CUDA mining backend\n"; u += " --cuda enable CUDA mining backend\n";
u += " --cuda-loader=PATH path to CUDA plugin (xmrig-cuda.dll or libxmrig-cuda.so)\n"; u += " --cuda-loader=PATH path to CUDA plugin (xmrig-cuda.dll or libxmrig-cuda.so)\n";
u += " --cuda-devices=N comma separated list of CUDA devices to use\n";
# endif # endif
# ifdef XMRIG_FEATURE_NVML # ifdef XMRIG_FEATURE_NVML
u += " --no-nvml disable NVML (NVIDIA Management Library) support\n"; u += " --no-nvml disable NVML (NVIDIA Management Library) support\n";

94
src/crypto/randomx/jit_compiler_a64_static.S
View file

@ -25,26 +25,32 @@
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE # 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. # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#if defined(__APPLE__)
#define DECL(x) _##x
#else
#define DECL(x) x
#endif
.arch armv8-a .arch armv8-a
.text .text
.global randomx_program_aarch64 .global DECL(randomx_program_aarch64)
.global randomx_program_aarch64_main_loop .global DECL(randomx_program_aarch64_main_loop)
.global randomx_program_aarch64_vm_instructions .global DECL(randomx_program_aarch64_vm_instructions)
.global randomx_program_aarch64_imul_rcp_literals_end .global DECL(randomx_program_aarch64_imul_rcp_literals_end)
.global randomx_program_aarch64_vm_instructions_end .global DECL(randomx_program_aarch64_vm_instructions_end)
.global randomx_program_aarch64_cacheline_align_mask1 .global DECL(randomx_program_aarch64_cacheline_align_mask1)
.global randomx_program_aarch64_cacheline_align_mask2 .global DECL(randomx_program_aarch64_cacheline_align_mask2)
.global randomx_program_aarch64_update_spMix1 .global DECL(randomx_program_aarch64_update_spMix1)
.global randomx_program_aarch64_vm_instructions_end_light .global DECL(randomx_program_aarch64_vm_instructions_end_light)
.global randomx_program_aarch64_light_cacheline_align_mask .global DECL(randomx_program_aarch64_light_cacheline_align_mask)
.global randomx_program_aarch64_light_dataset_offset .global DECL(randomx_program_aarch64_light_dataset_offset)
.global randomx_init_dataset_aarch64 .global DECL(randomx_init_dataset_aarch64)
.global randomx_init_dataset_aarch64_end .global DECL(randomx_init_dataset_aarch64_end)
.global randomx_calc_dataset_item_aarch64 .global DECL(randomx_calc_dataset_item_aarch64)
.global randomx_calc_dataset_item_aarch64_prefetch .global DECL(randomx_calc_dataset_item_aarch64_prefetch)
.global randomx_calc_dataset_item_aarch64_mix .global DECL(randomx_calc_dataset_item_aarch64_mix)
.global randomx_calc_dataset_item_aarch64_store_result .global DECL(randomx_calc_dataset_item_aarch64_store_result)
.global randomx_calc_dataset_item_aarch64_end .global DECL(randomx_calc_dataset_item_aarch64_end)
# Register allocation # Register allocation
@ -99,7 +105,7 @@
# v31 -> scale mask = 0x81f000000000000081f0000000000000 # v31 -> scale mask = 0x81f000000000000081f0000000000000
.balign 4 .balign 4
randomx_program_aarch64: DECL(randomx_program_aarch64):
# Save callee-saved registers # Save callee-saved registers
sub sp, sp, 192 sub sp, sp, 192
stp x16, x17, [sp] stp x16, x17, [sp]
@ -187,7 +193,7 @@ randomx_program_aarch64:
ldr q14, literal_v14 ldr q14, literal_v14
ldr q15, literal_v15 ldr q15, literal_v15
randomx_program_aarch64_main_loop: DECL(randomx_program_aarch64_main_loop):
# spAddr0 = spMix1 & ScratchpadL3Mask64; # spAddr0 = spMix1 & ScratchpadL3Mask64;
# spAddr1 = (spMix1 >> 32) & ScratchpadL3Mask64; # spAddr1 = (spMix1 >> 32) & ScratchpadL3Mask64;
lsr x18, x10, 32 lsr x18, x10, 32
@ -260,7 +266,7 @@ randomx_program_aarch64_main_loop:
orr v23.16b, v23.16b, v30.16b orr v23.16b, v23.16b, v30.16b
# Execute VM instructions # Execute VM instructions
randomx_program_aarch64_vm_instructions: DECL(randomx_program_aarch64_vm_instructions):
# 16 KB buffer for generated instructions # 16 KB buffer for generated instructions
.fill 4096,4,0 .fill 4096,4,0
@ -278,7 +284,7 @@ literal_x27: .fill 1,8,0
literal_x28: .fill 1,8,0 literal_x28: .fill 1,8,0
literal_x29: .fill 1,8,0 literal_x29: .fill 1,8,0
literal_x30: .fill 1,8,0 literal_x30: .fill 1,8,0
randomx_program_aarch64_imul_rcp_literals_end: DECL(randomx_program_aarch64_imul_rcp_literals_end):
literal_v0: .fill 2,8,0 literal_v0: .fill 2,8,0
literal_v1: .fill 2,8,0 literal_v1: .fill 2,8,0
@ -297,14 +303,14 @@ literal_v13: .fill 2,8,0
literal_v14: .fill 2,8,0 literal_v14: .fill 2,8,0
literal_v15: .fill 2,8,0 literal_v15: .fill 2,8,0
randomx_program_aarch64_vm_instructions_end: DECL(randomx_program_aarch64_vm_instructions_end):
# mx ^= r[readReg2] ^ r[readReg3]; # mx ^= r[readReg2] ^ r[readReg3];
eor x9, x9, x18 eor x9, x9, x18
# Calculate dataset pointer for dataset prefetch # Calculate dataset pointer for dataset prefetch
mov w18, w9 mov w18, w9
randomx_program_aarch64_cacheline_align_mask1: DECL(randomx_program_aarch64_cacheline_align_mask1):
# Actual mask will be inserted by JIT compiler # Actual mask will be inserted by JIT compiler
and x18, x18, 1 and x18, x18, 1
add x18, x18, x1 add x18, x18, x1
@ -317,12 +323,12 @@ randomx_program_aarch64_cacheline_align_mask1:
# Calculate dataset pointer for dataset read # Calculate dataset pointer for dataset read
mov w10, w9 mov w10, w9
randomx_program_aarch64_cacheline_align_mask2: DECL(randomx_program_aarch64_cacheline_align_mask2):
# Actual mask will be inserted by JIT compiler # Actual mask will be inserted by JIT compiler
and x10, x10, 1 and x10, x10, 1
add x10, x10, x1 add x10, x10, x1
randomx_program_aarch64_xor_with_dataset_line: DECL(randomx_program_aarch64_xor_with_dataset_line):
# xor integer registers with dataset data # xor integer registers with dataset data
ldp x18, x19, [x10] ldp x18, x19, [x10]
eor x4, x4, x18 eor x4, x4, x18
@ -337,7 +343,7 @@ randomx_program_aarch64_xor_with_dataset_line:
eor x14, x14, x18 eor x14, x14, x18
eor x15, x15, x19 eor x15, x15, x19
randomx_program_aarch64_update_spMix1: DECL(randomx_program_aarch64_update_spMix1):
# JIT compiler will replace it with "eor x10, config.readReg0, config.readReg1" # JIT compiler will replace it with "eor x10, config.readReg0, config.readReg1"
eor x10, x0, x0 eor x10, x0, x0
@ -358,7 +364,7 @@ randomx_program_aarch64_update_spMix1:
stp q18, q19, [x16, 32] stp q18, q19, [x16, 32]
subs x3, x3, 1 subs x3, x3, 1
bne randomx_program_aarch64_main_loop bne DECL(randomx_program_aarch64_main_loop)
# Restore x0 # Restore x0
ldr x0, [sp], 16 ldr x0, [sp], 16
@ -392,7 +398,7 @@ randomx_program_aarch64_update_spMix1:
ret ret
randomx_program_aarch64_vm_instructions_end_light: DECL(randomx_program_aarch64_vm_instructions_end_light):
sub sp, sp, 96 sub sp, sp, 96
stp x0, x1, [sp, 64] stp x0, x1, [sp, 64]
stp x2, x30, [sp, 80] stp x2, x30, [sp, 80]
@ -409,26 +415,26 @@ randomx_program_aarch64_vm_instructions_end_light:
# x1 -> pointer to output # x1 -> pointer to output
mov x1, sp mov x1, sp
randomx_program_aarch64_light_cacheline_align_mask: DECL(randomx_program_aarch64_light_cacheline_align_mask):
# Actual mask will be inserted by JIT compiler # Actual mask will be inserted by JIT compiler
and w2, w9, 1 and w2, w9, 1
# x2 -> item number # x2 -> item number
lsr x2, x2, 6 lsr x2, x2, 6
randomx_program_aarch64_light_dataset_offset: DECL(randomx_program_aarch64_light_dataset_offset):
# Apply dataset offset (filled in by JIT compiler) # Apply dataset offset (filled in by JIT compiler)
add x2, x2, 0 add x2, x2, 0
add x2, x2, 0 add x2, x2, 0
bl randomx_calc_dataset_item_aarch64 bl DECL(randomx_calc_dataset_item_aarch64)
mov x10, sp mov x10, sp
ldp x0, x1, [sp, 64] ldp x0, x1, [sp, 64]
ldp x2, x30, [sp, 80] ldp x2, x30, [sp, 80]
add sp, sp, 96 add sp, sp, 96
b randomx_program_aarch64_xor_with_dataset_line b DECL(randomx_program_aarch64_xor_with_dataset_line)
@ -439,26 +445,26 @@ randomx_program_aarch64_light_dataset_offset:
# x2 -> start item # x2 -> start item
# x3 -> end item # x3 -> end item
randomx_init_dataset_aarch64: DECL(randomx_init_dataset_aarch64):
# Save x30 (return address) # Save x30 (return address)
str x30, [sp, -16]! str x30, [sp, -16]!
# Load pointer to cache memory # Load pointer to cache memory
ldr x0, [x0] ldr x0, [x0]
randomx_init_dataset_aarch64_main_loop: DECL(randomx_init_dataset_aarch64_main_loop):
bl randomx_calc_dataset_item_aarch64 bl DECL(randomx_calc_dataset_item_aarch64)
add x1, x1, 64 add x1, x1, 64
add x2, x2, 1 add x2, x2, 1
cmp x2, x3 cmp x2, x3
bne randomx_init_dataset_aarch64_main_loop bne DECL(randomx_init_dataset_aarch64_main_loop)
# Restore x30 (return address) # Restore x30 (return address)
ldr x30, [sp], 16 ldr x30, [sp], 16
ret ret
randomx_init_dataset_aarch64_end: DECL(randomx_init_dataset_aarch64_end):
# Input parameters # Input parameters
# #
@ -476,7 +482,7 @@ randomx_init_dataset_aarch64_end:
# x12 -> temporary # x12 -> temporary
# x13 -> temporary # x13 -> temporary
randomx_calc_dataset_item_aarch64: DECL(randomx_calc_dataset_item_aarch64):
sub sp, sp, 112 sub sp, sp, 112
stp x0, x1, [sp] stp x0, x1, [sp]
stp x2, x3, [sp, 16] stp x2, x3, [sp, 16]
@ -523,7 +529,7 @@ randomx_calc_dataset_item_aarch64:
ldr x12, superscalarAdd7 ldr x12, superscalarAdd7
eor x7, x0, x12 eor x7, x0, x12
b randomx_calc_dataset_item_aarch64_prefetch b DECL(randomx_calc_dataset_item_aarch64_prefetch)
superscalarMul0: .quad 6364136223846793005 superscalarMul0: .quad 6364136223846793005
superscalarAdd1: .quad 9298411001130361340 superscalarAdd1: .quad 9298411001130361340
@ -536,7 +542,7 @@ superscalarAdd7: .quad 9549104520008361294
# Prefetch -> SuperScalar hash -> Mix will be repeated N times # Prefetch -> SuperScalar hash -> Mix will be repeated N times
randomx_calc_dataset_item_aarch64_prefetch: DECL(randomx_calc_dataset_item_aarch64_prefetch):
# Actual mask will be inserted by JIT compiler # Actual mask will be inserted by JIT compiler
and x11, x10, 1 and x11, x10, 1
add x11, x8, x11, lsl 6 add x11, x8, x11, lsl 6
@ -544,7 +550,7 @@ randomx_calc_dataset_item_aarch64_prefetch:
# Generated SuperScalar hash program goes here # Generated SuperScalar hash program goes here
randomx_calc_dataset_item_aarch64_mix: DECL(randomx_calc_dataset_item_aarch64_mix):
ldp x12, x13, [x11] ldp x12, x13, [x11]
eor x0, x0, x12 eor x0, x0, x12
eor x1, x1, x13 eor x1, x1, x13
@ -558,7 +564,7 @@ randomx_calc_dataset_item_aarch64_mix:
eor x6, x6, x12 eor x6, x6, x12
eor x7, x7, x13 eor x7, x7, x13
randomx_calc_dataset_item_aarch64_store_result: DECL(randomx_calc_dataset_item_aarch64_store_result):
stp x0, x1, [x9] stp x0, x1, [x9]
stp x2, x3, [x9, 16] stp x2, x3, [x9, 16]
stp x4, x5, [x9, 32] stp x4, x5, [x9, 32]
@ -575,4 +581,4 @@ randomx_calc_dataset_item_aarch64_store_result:
ret ret
randomx_calc_dataset_item_aarch64_end: DECL(randomx_calc_dataset_item_aarch64_end):

8
src/version.h
View file

@ -28,15 +28,15 @@
#define APP_ID "xmrig" #define APP_ID "xmrig"
#define APP_NAME "XMRig" #define APP_NAME "XMRig"
#define APP_DESC "XMRig miner" #define APP_DESC "XMRig miner"
#define APP_VERSION "4.6.1-beta-mo1" #define APP_VERSION "5.0.0-mo1"
#define APP_DOMAIN "xmrig.com" #define APP_DOMAIN "xmrig.com"
#define APP_SITE "www.xmrig.com" #define APP_SITE "www.xmrig.com"
#define APP_COPYRIGHT "Copyright (C) 2016-2019 xmrig.com" #define APP_COPYRIGHT "Copyright (C) 2016-2019 xmrig.com"
#define APP_KIND "miner" #define APP_KIND "miner"
#define APP_VER_MAJOR 4 #define APP_VER_MAJOR 5
#define APP_VER_MINOR 6 #define APP_VER_MINOR 0
#define APP_VER_PATCH 1 #define APP_VER_PATCH 0
#ifdef _MSC_VER #ifdef _MSC_VER
# if (_MSC_VER >= 1920) # if (_MSC_VER >= 1920)