Skip to content

Commit

Permalink
extract compileKernel()
Browse files Browse the repository at this point in the history 
 * check period_seed and call compileKernel()
  • Loading branch information
@hackmod
hackmod committed Dec 18, 2018
1 parent cd3b62b commit 946d9c0
Show file tree
Hide file tree
Showing 2 changed files with 95 additions and 77 deletions.
168 changes: 91 additions & 77 deletions libethash-cl/CLMiner.cpp
View file
Expand Up @@ -299,6 +299,7 @@ void CLMiner::workLoop()
// The work package currently processed by GPU.
WorkPackage current;
current.header = h256();
uint64_t old_period_seed = -1;

if (!initDevice())
return;
Expand Down Expand Up @@ -332,6 +333,7 @@ void CLMiner::workLoop()

// Wait for work or 3 seconds (whichever the first)
const WorkPackage w = work();
uint64_t period_seed = w.block / PROGPOW_PERIOD;
if (!w)
{
boost::system_time const timeout =
Expand All @@ -353,6 +355,11 @@ void CLMiner::workLoop()

m_abortqueue.push_back(cl::CommandQueue(m_context[0], m_device));
}
else if (old_period_seed != period_seed)
{
compileKernel(period_seed);
}
old_period_seed = period_seed;

// Upper 64 bits of the boundary.
const uint64_t target = (uint64_t)(u64)((u256)w.boundary >> 192);
Expand Down Expand Up @@ -698,18 +705,15 @@ bool CLMiner::initEpoch_internal(uint64_t block_number)

try
{

char options[256] = {0};
int computeCapability = 0;
#ifndef __clang__

// Nvidia
if (!m_deviceDescriptor.clNvCompute.empty())
{
computeCapability =
m_computeCapability =
m_deviceDescriptor.clNvComputeMajor * 10 + m_deviceDescriptor.clNvComputeMinor;
int maxregs = computeCapability >= 35 ? 72 : 63;
sprintf(options, "-cl-nv-maxrregcount=%d", maxregs);
int maxregs = m_computeCapability >= 35 ? 72 : 63;
sprintf(m_options, "-cl-nv-maxrregcount=%d", maxregs);
}

#endif
Expand All @@ -722,63 +726,9 @@ bool CLMiner::initEpoch_internal(uint64_t block_number)

m_dagItems = m_epochContext.dagNumItems;

// patch source code
// note: The kernels here are simply compiled version of the respective .cl kernels
// into a byte array by bin2h.cmake. There is no need to load the file by hand in runtime
// See libethash-cl/CMakeLists.txt: add_custom_command()
// TODO: Just use C++ raw string literal.

cl::Program binaryProgram;
uint64_t period_seed = block_number / PROGPOW_PERIOD;
std::string code = ProgPow::getKern(period_seed, ProgPow::KERNEL_CL);
code += string(CLMiner_kernel, sizeof(CLMiner_kernel));

addDefinition(code, "GROUP_SIZE", m_settings.localWorkSize);
addDefinition(code, "ACCESSES", 64);
addDefinition(code, "LIGHT_WORDS", m_epochContext.lightNumItems);
addDefinition(code, "PROGPOW_DAG_BYTES", m_epochContext.dagSize);
addDefinition(code, "PROGPOW_DAG_ELEMENTS", m_dagItems / 2);

addDefinition(code, "MAX_OUTPUTS", c_maxSearchResults);
int platform = 0;
switch (m_deviceDescriptor.clPlatformType) {
case ClPlatformTypeEnum::Nvidia:
platform = 1;
break;
case ClPlatformTypeEnum::Amd:
platform = 2;
break;
case ClPlatformTypeEnum::Clover:
platform = 3;
break;
default:
break;
}
addDefinition(code, "PLATFORM", platform);
addDefinition(code, "COMPUTE", computeCapability);

if (m_deviceDescriptor.clPlatformType == ClPlatformTypeEnum::Clover)
addDefinition(code, "LEGACY", 1);

ofstream out;
out.open("kernel.cl");
out << code;
out.close();

// create miner OpenCL program
cl::Program::Sources sources{{code.data(), code.size()}};
cl::Program program(m_context[0], sources), binaryProgram;
try
{
program.build({m_device}, options);
}
catch (cl::BuildError const& buildErr)
{
cwarn << "OpenCL kernel build log:\n"
<< program.getBuildInfo<CL_PROGRAM_BUILD_LOG>(m_device);
cwarn << "OpenCL kernel build error (" << buildErr.err() << "):\n" << buildErr.what();
pause(MinerPauseEnum::PauseDueToInitEpochError);
return true;
}
compileKernel(period_seed);

/* If we have a binary kernel, we load it in tandem with the opencl,
that way, we can use the dag generate opencl code and fall back on
Expand Down Expand Up @@ -819,7 +769,7 @@ bool CLMiner::initEpoch_internal(uint64_t block_number)
cl::Program program(m_context[0], {m_device}, blobs);
try
{
program.build({m_device}, options);
program.build({m_device}, m_options);
cllog << "Build info success:"
<< program.getBuildInfo<CL_PROGRAM_BUILD_LOG>(m_device);
binaryProgram = program;
Expand Down Expand Up @@ -862,12 +812,8 @@ bool CLMiner::initEpoch_internal(uint64_t block_number)
m_dag.push_back(cl::Buffer(m_context[0], CL_MEM_READ_ONLY, m_epochContext.dagSize));
cllog << "Loading kernels";

// If we have a binary kernel to use, let's try it
// otherwise just do a normal opencl load
m_searchKernel = cl::Kernel(program, "ethash_search");

//m_dagKernel = cl::Kernel(program, "GenerateDAG");
m_dagKernel = cl::Kernel(program, "ethash_calculate_dag_item");
m_dagKernel = cl::Kernel(m_program, "ethash_calculate_dag_item");

cllog << "Writing light cache buffer";
m_queue[0].enqueueWriteBuffer(
Expand All @@ -879,15 +825,6 @@ bool CLMiner::initEpoch_internal(uint64_t block_number)
pause(MinerPauseEnum::PauseDueToInitEpochError);
return true;
}
// create buffer for header
ETHCL_LOG("Creating buffer for header.");
m_header.clear();
m_header.push_back(cl::Buffer(m_context[0], CL_MEM_READ_ONLY, 32));

m_searchKernel.setArg(1, m_header[0]);
m_searchKernel.setArg(2, m_dag[0]);
m_searchKernel.setArg(5, 0);

// create mining buffers
ETHCL_LOG("Creating mining buffer");
m_searchBuffer.clear();
Expand Down Expand Up @@ -931,3 +868,80 @@ bool CLMiner::initEpoch_internal(uint64_t block_number)
}
return true;
}

bool CLMiner::compileKernel(
uint64_t period_seed)
{
// patch source code
// note: The kernels here are simply compiled version of the respective .cl kernels
// into a byte array by bin2h.cmake. There is no need to load the file by hand in runtime
// See libethash-cl/CMakeLists.txt: add_custom_command()
// TODO: Just use C++ raw string literal.

std::string code = ProgPow::getKern(period_seed, ProgPow::KERNEL_CL);
code += string(CLMiner_kernel, sizeof(CLMiner_kernel));

addDefinition(code, "GROUP_SIZE", m_settings.localWorkSize);
addDefinition(code, "ACCESSES", 64);
addDefinition(code, "LIGHT_WORDS", m_epochContext.lightNumItems);
addDefinition(code, "PROGPOW_DAG_BYTES", m_epochContext.dagSize);
addDefinition(code, "PROGPOW_DAG_ELEMENTS", m_dagItems / 2);

addDefinition(code, "MAX_OUTPUTS", c_maxSearchResults);
int platform = 0;
switch (m_deviceDescriptor.clPlatformType) {
case ClPlatformTypeEnum::Nvidia:
platform = 1;
break;
case ClPlatformTypeEnum::Amd:
platform = 2;
break;
case ClPlatformTypeEnum::Clover:
platform = 3;
break;
default:
break;
}
addDefinition(code, "PLATFORM", platform);
addDefinition(code, "COMPUTE", m_computeCapability);

if (m_deviceDescriptor.clPlatformType == ClPlatformTypeEnum::Clover)
addDefinition(code, "LEGACY", 1);

ofstream out;
out.open("kernel.cl");
out << code;
out.close();

// create miner OpenCL program
cl::Program::Sources sources{{code.data(), code.size()}};
cl::Program program(m_context[0], sources);//, binaryProgram;
try
{
program.build({m_device}, m_options);
}
catch (cl::BuildError const& buildErr)
{
cwarn << "OpenCL kernel build log:\n"
<< program.getBuildInfo<CL_PROGRAM_BUILD_LOG>(m_device);
cwarn << "OpenCL kernel build error (" << buildErr.err() << "):\n" << buildErr.what();
pause(MinerPauseEnum::PauseDueToInitEpochError);
return true;
}
m_program = program;

// If we have a binary kernel to use, let's try it
// otherwise just do a normal opencl load
m_searchKernel = cl::Kernel(m_program, "ethash_search");

m_searchKernel.setArg(1, m_header[0]);
m_searchKernel.setArg(2, m_dag[0]);
m_searchKernel.setArg(5, 0);

// create buffer for header
ETHCL_LOG("Creating buffer for header.");
m_header.clear();
m_header.push_back(cl::Buffer(m_context[0], CL_MEM_READ_ONLY, 32));

return false;
}
4 changes: 4 additions & 0 deletions libethash-cl/CLMiner.h
View file
Expand Up @@ -78,6 +78,10 @@ class CLMiner : public Miner
unsigned m_dagItems = 0;
uint64_t m_lastNonce = 0;

cl::Program m_program;
char m_options[256] = {0};
int m_computeCapability = 0;
bool compileKernel(uint64_t prog_seed);
};

} // namespace eth
Expand Down

0 comments on commit 946d9c0

Please sign in to comment.