GPUOcelot
|
A device to control all of the cores in a single CPU. More...
#include <MulticoreCPUDevice.h>
Classes | |
class | Module |
Specialization of EmulatorDevice::Module for LLVM. | |
Public Member Functions | |
MulticoreCPUDevice (unsigned int flags=0) | |
Sets the device properties. | |
void | load (const ir::Module *module) |
Load a module, must have a unique name. | |
ExecutableKernel * | getKernel (const std::string &module, const std::string &kernel) |
Get a translated kernel from the device. | |
void | launch (const std::string &module, const std::string &kernel, const ir::Dim3 &grid, const ir::Dim3 &block, size_t sharedMemory, const void *argumentBlock, size_t argumentBlockSize, const trace::TraceGeneratorVector &traceGenerators=trace::TraceGeneratorVector(), const ir::ExternalFunctionSet *externals=0) |
helper function for launching a kernel | |
void | limitWorkerThreads (unsigned int threads) |
Limit the worker threads used by this device. | |
void | setOptimizationLevel (translator::Translator::OptimizationLevel level) |
Set the optimization level for kernels in this device. |
A device to control all of the cores in a single CPU.
executive::MulticoreCPUDevice::MulticoreCPUDevice | ( | unsigned int | flags = 0 | ) |
Sets the device properties.
ExecutableKernel * executive::MulticoreCPUDevice::getKernel | ( | const std::string & | module, |
const std::string & | kernel | ||
) | [virtual] |
Get a translated kernel from the device.
Reimplemented from executive::EmulatorDevice.
void executive::MulticoreCPUDevice::launch | ( | const std::string & | module, |
const std::string & | kernel, | ||
const ir::Dim3 & | grid, | ||
const ir::Dim3 & | block, | ||
size_t | sharedMemory, | ||
const void * | argumentBlock, | ||
size_t | argumentBlockSize, | ||
const trace::TraceGeneratorVector & | traceGenerators = trace::TraceGeneratorVector() , |
||
const ir::ExternalFunctionSet * | externals = 0 |
||
) | [virtual] |
helper function for launching a kernel
module | module name |
kernel | kernel name |
grid | grid dimensions |
block | block dimensions |
sharedMemory | shared memory size |
argumentBlock | array of bytes for parameter memory |
argumentBlockSize | number of bytes in parameter memory |
traceGenerators | vector of trace generators to add and remove from kernel |
Reimplemented from executive::EmulatorDevice.
void executive::MulticoreCPUDevice::limitWorkerThreads | ( | unsigned int | threads | ) | [virtual] |
Limit the worker threads used by this device.
Reimplemented from executive::EmulatorDevice.
void executive::MulticoreCPUDevice::load | ( | const ir::Module * | module | ) | [virtual] |
Load a module, must have a unique name.
Reimplemented from executive::EmulatorDevice.
void executive::MulticoreCPUDevice::setOptimizationLevel | ( | translator::Translator::OptimizationLevel | level | ) | [virtual] |
Set the optimization level for kernels in this device.
Reimplemented from executive::EmulatorDevice.