cl_kernel get_kernel(char *kernel_name, cl_context *context, cl_device_id *device)
{
cl_int status = CL_SUCCESS;
const char* program_source = load_program_source(PROGRAM_SRC);
if(program_source == NULL) {
fprintf(stderr, "Programm can not be created. File was not found.");
abort();
}
cl_program program = clCreateProgramWithSource(*context, 1,
&program_source, NULL,
&status);
CL_CHECK_ERROR(status);
status = clBuildProgram(program, 0, NULL, NULL, NULL, NULL);
/* Print build log */
char buf[0x10000];
clGetProgramBuildInfo(program,
*device,
CL_PROGRAM_BUILD_LOG,
0x10000,
buf,
NULL);
if(status != CL_SUCCESS) {
printf("\n-------BUILD LOG:\n %s \n-------\n", buf);
fprintf(stderr, "Programm can not be build. (%s)", opencl_map_error(status));
abort();
}
return clCreateKernel(program, kernel_name, &status);
}
void createKernel(const char* kernel, const char* path, int indice) {
// TheContext* tc = new TheContext();
// cl_context GPUContext_K = tc->getMyContext()->getContextCL();
// cl_device_id cdDevice_K = tc->getMyContext()->getDeviceCL();
// Creates the program
// Uses NVIDIA helper functions to get the code string and it's size (in bytes)
//size_t src_size = 0;
char full_path[256];
#ifdef _VIVID_STATIC_LIB
sprintf(full_path, "%s", path);
#else
sprintf(full_path, "%s", path);
#endif
char *program_source = load_program_source(full_path);
if (program_source == NULL) {
printf("Error: Failed to read the OpenCL kernel: %s\n",path);
exit(-1);
}
cl_int err;
program_list[indice] = clCreateProgramWithSource(GPUContext_K, 1, (const char **) &program_source, NULL, &err);
if (!program_list[indice]) {
printf("Error: Failed to create compute program for device %d Kernel: (%s)!\n", indice,kernel);
printf("************\n%s\n************\n", program_source);
}
// Build the program executable
const char * options = "-cl-fast-relaxed-math";
err = clBuildProgram(program_list[indice], 0, NULL, options, NULL, NULL);
if (err != CL_SUCCESS) {
size_t len;
char buffer[10000];
printf("Error: Failed to build program executable for device %d kernel: (%s)!\n",err,kernel);
cl_int get_err=clGetProgramBuildInfo(program_list[indice], cdDevice_K, CL_PROGRAM_BUILD_LOG, sizeof (buffer), buffer, &len);
printf("%d %s\n", get_err, buffer);
}
kernel_list[indice] = clCreateKernel(program_list[indice], kernel, &err);
if (!kernel_list[indice] || err != CL_SUCCESS) {
printf("Error: Failed to create compute kernel for device %d Kernel: (%s)!\n", indice, full_path);
exit(1);
}
}
void init_cl(int ker_id, char *kernel_path, t_cl *cl)
{
cl_device_id device;
char *source;
cl_program program;
clGetDeviceIDs(NULL, CL_DEVICE_TYPE_GPU, 1, &device, NULL);
if (!cl->context)
{
cl->context = clCreateContext(0, 1, &device, NULL, NULL, NULL);
cl->cmd_queue = clCreateCommandQueue(cl->context, device, 0, NULL);
}
source = load_program_source(kernel_path);
program = clCreateProgramWithSource(cl->context, 1, (const char **)&source, NULL, NULL);
free(source);
clBuildProgram(program, 0, NULL, NULL, NULL, NULL);
cl->kernel[ker_id] = clCreateKernel(program, "thread", NULL);
cl->input_cl_mem[ker_id] = clCreateBuffer(cl->context, CL_MEM_READ_ONLY, cl->gws[ker_id] * sizeof(float), NULL, NULL);
cl->output_cl_mem[ker_id] = clCreateBuffer(cl->context, CL_MEM_READ_WRITE, cl->gws[ker_id] * sizeof(float), NULL, NULL);
}
void opencl_setup(CLEnv& env)
{
/*****************************************/
/* Initialize OpenCL */
/*****************************************/
clGetPlatformIDs(1, &env.cpPlatform, NULL);
clGetDeviceIDs(env.cpPlatform, CL_DEVICE_TYPE_GPU, 1, &env.cdDevice, NULL);
env.context = clCreateContext(0, 1, &env.cdDevice, NULL, NULL, &env.errcode);
//env.context = clCreateContextFromType(0, CL_DEVICE_TYPE_GPU,NULL, NULL, &env.errcode);
opencl_check_error(env.errcode, CL_SUCCESS, __FILE__ , __LINE__ );
// get the list of GPU devices associated with context
env.errcode = clGetContextInfo(env.context, CL_CONTEXT_DEVICES, 0, NULL,&env.device_size);
env.devices = (cl_device_id *) malloc(env.device_size);
env.errcode |= clGetContextInfo(env.context, CL_CONTEXT_DEVICES, env.device_size, env.devices, NULL);
opencl_check_error(env.errcode, CL_SUCCESS, __FILE__ , __LINE__ );
//Create a command-queue
env.command_queue = clCreateCommandQueue(env.context, env.cdDevice, 0, &env.errcode);
opencl_check_error(env.errcode, CL_SUCCESS, __FILE__ , __LINE__ );
// Load and build OpenCL kernel
const char * filename = "kernel.cl";
char* kernel_source = load_program_source(filename);
env.program = clCreateProgramWithSource(env.context, 1, (const char**)&kernel_source, NULL, &env.errcode);
opencl_check_error(env.errcode, CL_SUCCESS, __FILE__ , __LINE__ );
env.errcode = clBuildProgram(env.program, 0, NULL, NULL, NULL, NULL);
opencl_check_error(env.errcode, CL_SUCCESS, __FILE__ , __LINE__ );
env.kernel = clCreateKernel(env.program, "matrix_mul", &env.errcode);
opencl_check_error(env.errcode, CL_SUCCESS, __FILE__ , __LINE__ );
free(kernel_source);
}
cl_int
load_kernel(cl_context context, cl_device_id *devices, unsigned int devc, cl_program *prog, cl_kernel *kern)
{
cl_int err;
char* source = load_program_source("ConwayKernel.cl");
*prog = clCreateProgramWithSource(context, 1, (const char**)&source,
NULL, &err);
if (*prog == NULL)
{
printf("clCreateProgramWithSource failed! (Error: %d)\n", err);
return err;
}
err = clBuildProgram(*prog, devc, devices, NULL, NULL, NULL);
if (err)
{
printf("clBuildProgram failed! (Error: %d)\n", err);
size_t length;
char buffer[2048];
clGetProgramBuildInfo(*prog, devices[0], CL_PROGRAM_BUILD_LOG, sizeof(buffer), buffer, &length);
printf("Build log: %s\n", buffer);
return err;
}
*kern = clCreateKernel(*prog, "evaluate_bit", &err);
if (*kern == NULL)
{
printf("clCreateKernel failed! (Error: %d)\n", err);
return err;
}
printf("Kernel build completed successfully\n");
return 0;
}
int main(int argc, char *argv[])
{
cl_platform_id platform;
cl_device_id device;
cl_context context;
cl_command_queue queue;
cl_program program;
cl_kernel kernel;
cl_mem buff_A, buff_B, buff_C;
int mult = 1;
uint32_t uiWA, uiHA, uiWB, uiHB, uiWC, uiHC;
uiWA = WA * mult;
uiHA = HA * mult;
uiWB = WB * mult;
uiHB = HB * mult;
uiWC = WC * mult;
uiHC = HC * mult;
printf("sizes WA %u HA %u WB %u HB %u WC %u HC %u\n",
uiWA, uiHA, uiWB, uiHB, uiWC, uiHC);
uint32_t size_A = uiWA * uiHA;
uint32_t size_B = uiWB * uiHB;
uint32_t size_C = uiWC * uiHC;
size_t mem_size_A = size_A * sizeof(float);
size_t mem_size_B = size_B * sizeof(float);
size_t mem_size_C = size_C * sizeof(float);
float *data_A = (float *)malloc(mem_size_A);
float *data_B = (float *)malloc(mem_size_B);
float *data_C = (float *)malloc(mem_size_C);
srand(2012);
shrFillArray(data_A, size_A);
shrFillArray(data_B, size_B);
size_t global_work_size[2];
size_t local_work_size[] = { BLOCK_SIZE, BLOCK_SIZE };
global_work_size[0] = shrRoundUp(BLOCK_SIZE, uiWC);
global_work_size[1] = shrRoundUp(BLOCK_SIZE, uiHA);
const char *source = load_program_source("MatrixMul.cl");
size_t source_len = strlen(source);;
cl_uint err = 0;
char *flags = "-cl-fast-relaxed-math";
clGetPlatformIDs(1, &platform, NULL);
printf("platform %p err %d\n", platform, err);
clGetDeviceIDs(platform, CL_DEVICE_TYPE_CPU, 1, &device, &err);
printf("device %p err %d\n", device, err);
context = clCreateContext(0, 1, &device, NULL, NULL, &err);
printf("context %p err %d\n", context, err);
queue = clCreateCommandQueue(context, device, 0, &err);
printf("queue %p err %d\n", queue, err);
program = clCreateProgramWithSource(context, 1, &source, &source_len, &err);
printf("program %p err %d\n", program, err);
err = clBuildProgram(program, 0, NULL, flags, NULL, NULL);
printf("err %d\n", err);
kernel = clCreateKernel(program, "matrixMul", &err);
printf("kernel %p err %d\n", kernel, err);
buff_A = clCreateBuffer(context, CL_MEM_READ_ONLY | CL_MEM_COPY_HOST_PTR,
mem_size_A, data_A, NULL);
printf("buff_A %p\n", buff_A);
buff_B = clCreateBuffer(context, CL_MEM_READ_ONLY | CL_MEM_COPY_HOST_PTR,
mem_size_B, data_B, NULL);
printf("buff_B %p\n", buff_B);
buff_C = clCreateBuffer(context, CL_MEM_WRITE_ONLY, mem_size_C, NULL, NULL);
printf("buff_C %p\n", buff_C);
err = clSetKernelArg(kernel, 0, sizeof(cl_mem), (void*)&buff_C);
printf("err %d\n", err);
err = clSetKernelArg(kernel, 1, sizeof(cl_mem), (void*)&buff_A);
printf("err %d\n", err);
err = clSetKernelArg(kernel, 2, sizeof(cl_mem), (void*)&buff_B);
printf("err %d\n", err);
err = clSetKernelArg(kernel, 3, sizeof(float) * BLOCK_SIZE * BLOCK_SIZE, NULL);
printf("err %d\n", err);
err = clSetKernelArg(kernel, 4, sizeof(float) * BLOCK_SIZE * BLOCK_SIZE, NULL);
printf("err %d\n", err);
err = clSetKernelArg(kernel, 5, sizeof(cl_int), (void*)&uiWA);
printf("err %d\n", err);
err = clSetKernelArg(kernel, 6, sizeof(cl_int), (void*)&uiWB);
printf("err %d\n", err);
err = clSetKernelArg(kernel, 7, sizeof(cl_int), (void*)&uiHA);
printf("err %d\n", err);
err = clEnqueueNDRangeKernel(queue, kernel, 2, NULL, global_work_size,
local_work_size, 0, NULL, NULL);
printf("err %d\n", err);
err = clFlush(queue);
printf("err %d\n", err);
err = clFinish(queue);
printf("err %d\n", err);
err = clEnqueueReadBuffer(queue, buff_C, CL_TRUE, 0, mem_size_C, data_C, 0,
NULL, NULL);
printf("err %d\n", err);
int i;
for (i = 0; i < size_C; i++) {
printf("%d %f\n", i, data_C[i]);
}
clReleaseMemObject(buff_A);
clReleaseMemObject(buff_B);
clReleaseMemObject(buff_C);
clReleaseProgram(program);
clReleaseKernel(kernel);
clReleaseCommandQueue(queue);
clReleaseProgram(program);
}
CLKernel::CLKernel(cl_context context, cl_command_queue commands, cl_device_id device,
const char * filename, const char * name, const char * options)
{
this->device = device;
this->commands = commands;
//this->profiling = true;
//Profiling doesn't work on neuro
this->profiling = false;
this->elapsed = 0;
#ifdef PV_USE_OPENCL
int status = CL_SUCCESS;
// Create the compute program from the source buffer
//
char * source = load_program_source(filename);
program = clCreateProgramWithSource(context, 1, (const char **) &source, NULL, &status);
if (!program || status != CL_SUCCESS)
{
printf("Error: Failed to create compute program!\n");
CLDevice::print_error_code(status);
exit(status);
}
// Build the program executable
//
// TODO - fix include path
status = clBuildProgram(program, 0, NULL, options, NULL, NULL);
if (status != CL_SUCCESS)
{
size_t len;
char buffer[150641]; //[12050]; //[8192];
printf("Error: Failed to build program executable!\n");
CLDevice::print_error_code(status);
status = clGetProgramBuildInfo(program, device, CL_PROGRAM_BUILD_LOG, sizeof(buffer), buffer, &len);
if (status != CL_SUCCESS) {
printf("CLKernel: error buffer length may be too small, is %ld, should be %ld\n", sizeof(buffer), len);
CLDevice::print_error_code(status);
}
printf("%s\n", buffer);
exit(status);
}
// Create the compute kernel in the program we wish to run
//
kernel = clCreateKernel(program, name, &status);
if (!kernel || status != CL_SUCCESS)
{
fprintf(stderr, "Error: Failed to create compute kernel!\n");
CLDevice::print_error_code(status);
exit(status);
}
#endif // PV_USE_OPENCL
}
int main( int argc, char* argv[] )
{
// Length of vectors
unsigned int n = 10;
struct timespec start, finish;
// Host input vectors
int *h_a;
int *h_b;
// Host output vector
int *h_c;
double elapsed;
// Device input buffers
cl_mem d_a;
cl_mem d_b;
// Device output buffer
cl_mem d_c;
cl_platform_id cpPlatform; // OpenCL platform
cl_device_id device_id; // device ID
cl_context context; // context
cl_command_queue queue; // command queue
cl_program program; // program
cl_kernel kernel; // kernel
// Size, in bytes, of each vector
size_t bytes = n*sizeof(int);
// Allocate memory for each vector on host
h_a = (int*)malloc(bytes);
h_b = (int*)malloc(bytes);
h_c = (int*)malloc(bytes);
// Initialize vectors on host
int i;
for( i = 0; i < n; i++ )
{
h_a[i] = i;
h_b[i] = i;
}
size_t globalSize, localSize;
cl_int err;
int workgrp;
int wrkitm;
//wrkitm=atoi(argv[1]);
// Number of work items in each local work group
// localSize = wrkitm ;
//workgrp=atoi(argv[2]);
// Number of total work items - localSize must be devisor
globalSize = n;//ceil(n/(float)localSize)*localSize;
//cl_uint platformCount;
//cl_platform_id* platforms;
//clGetPlatformIDs(0, NULL, &platformCount);
// platforms = (cl_platform_id*) malloc(sizeof(cl_platform_id) * platformCount);
// clGetPlatformIDs(platformCount, platforms, NULL);
//printf("%d",platformCount);
// Bind to platform
err = clGetPlatformIDs(1, &cpPlatform, NULL);
// Get ID for the device
err = clGetDeviceIDs(NULL, CL_DEVICE_TYPE_CPU, 1, &device_id, NULL);
if (err != CL_SUCCESS)
{
printf("Error: Failed to create a device group!\n");
}
// Create a context
context = clCreateContext(0, 1, &device_id, NULL, NULL, &err);
if (!context)
printf("Error: Failed to create a compute context!\n");
// Create a command queue
queue = clCreateCommandQueue(context, device_id, 0, &err);
//loading external cl file
const char *file="vectadd.cl";
const char *kernelSource = load_program_source(file);
// Create the compute program from the source buffer
program = clCreateProgramWithSource(context, 1,
(const char **) & kernelSource, NULL, &err);
// Build the program executable
clBuildProgram(program, 0, NULL, NULL, NULL, NULL);
// Create the compute kernel in the program we wish to run
kernel = clCreateKernel(program, "vecAdd", &err);
// Create the input and output arrays in device memory for our calculation
d_a = clCreateBuffer(context, CL_MEM_READ_ONLY, bytes, NULL, NULL);
d_b = clCreateBuffer(context, CL_MEM_READ_ONLY, bytes, NULL, NULL);
d_c = clCreateBuffer(context, CL_MEM_WRITE_ONLY, bytes, NULL, NULL);
// Write our data set into the input array in device memory
err = clEnqueueWriteBuffer(queue, d_a, CL_TRUE, 0,bytes, h_a, 0, NULL, NULL);
err = clEnqueueWriteBuffer(queue, d_b, CL_TRUE, 0,bytes, h_b, 0, NULL, NULL);
clFinish(queue);
// Set the arguments to our compute kernel
err = clSetKernelArg(kernel, 0, sizeof(cl_mem), &d_a);
err = clSetKernelArg(kernel, 1, sizeof(cl_mem), &d_b);
err = clSetKernelArg(kernel, 2, sizeof(cl_mem), &d_c);
err = clSetKernelArg(kernel, 3, sizeof(unsigned int), &n);
clock_gettime(CLOCK_MONOTONIC, &start);
// Execute the kernel over the entire range of the data set
err = clEnqueueNDRangeKernel(queue, kernel, 1, NULL, &globalSize, &localSize,
0, NULL, NULL);
clock_gettime(CLOCK_MONOTONIC, &finish);
elapsed = (finish.tv_sec - start.tv_sec);
elapsed += (finish.tv_nsec - start.tv_nsec)/ 1000000000.0;
// Wait for the command queue to get serviced before reading back results
clFinish(queue);
// Read the results from the device
clEnqueueReadBuffer(queue, d_c, CL_TRUE, 0,
bytes, h_c, 0, NULL, NULL );
clFinish(queue);
//Sum up vector c and print result divided by n, this should equal 1 within error
double sum = 0;
for(i=0; i<n; i++)
sum += h_c[i];
printf("Work Item/threads = %d \n",wrkitm);
printf("time taken by GPU = %le\n ",elapsed);
// release OpenCL resources
clReleaseMemObject(d_a);
clReleaseMemObject(d_b);
clReleaseMemObject(d_c);
clReleaseProgram(program);
clReleaseKernel(kernel);
clReleaseCommandQueue(queue);
clReleaseContext(context);
//release host memory
free(h_a);
free(h_b);
free(h_c);
return 0;
}
int main(int argc, char **argv)
{
int err;
cl_device_id device_id;
cl_command_queue commands;
cl_context context;
cl_mem output_buffer;
cl_mem input_buffer;
cl_mem partials_buffer;
size_t typesize;
int pass_count = 0;
size_t* group_counts = 0;
size_t* work_item_counts = 0;
int* operation_counts = 0;
int* entry_counts = 0;
int use_gpu = 1;
int i;
int c;
// Parse command line options
//
for( i = 0; i < argc && argv; i++)
{
if(!argv[i])
continue;
if(strstr(argv[i], "cpu"))
{
use_gpu = 0;
}
else if(strstr(argv[i], "gpu"))
{
use_gpu = 1;
}
}
channels=1;
// Create some random input data on the host
//
time_t tstart=0;
(void) time(&tstart);
srand48((long) tstart);
float *float_data = (float*)malloc(count * channels * sizeof(float));
for (i = 0; i < count * channels; i++)
{
float_data[i] = drand48();
}
//SETUP PLATFORM
cl_uint numPlatforms;
err = clGetPlatformIDs(0, NULL, &numPlatforms);
if (err != CL_SUCCESS) {
fprintf(stderr,"could not get platform\n");
exit(EXIT_FAILURE);
}
cl_platform_id platform_id;
if(numPlatforms > 0)
{
//we have at least one
//cl_platform_id* platforms = new cl_platform_id[numPlatforms];
cl_platform_id* platforms = calloc(numPlatforms, sizeof(cl_platform_id));
err = clGetPlatformIDs(numPlatforms, platforms, NULL);
if (err != CL_SUCCESS) {
fprintf(stderr,"could not get platform id\n");
exit(EXIT_FAILURE);
}
fprintf(stderr,"Found %d platforms\n", numPlatforms);
platform_id = platforms[0];
//delete[] platforms;
free(platforms);
}
else
exit(0);
// Connect to a compute device
//
err = clGetDeviceIDs(platform_id, use_gpu ? CL_DEVICE_TYPE_GPU : CL_DEVICE_TYPE_CPU, 1, &device_id, NULL);
if (err != CL_SUCCESS)
{
printf("Error: Failed to locate a compute device!\n");
return EXIT_FAILURE;
}
size_t returned_size = 0;
size_t max_workgroup_size = 0;
err = clGetDeviceInfo(device_id, CL_DEVICE_MAX_WORK_GROUP_SIZE, sizeof(size_t), &max_workgroup_size, &returned_size);
if (err != CL_SUCCESS)
{
printf("Error: Failed to retrieve device info!\n");
return EXIT_FAILURE;
}
cl_char vendor_name[1024] = {0};
cl_char device_name[1024] = {0};
err = clGetDeviceInfo(device_id, CL_DEVICE_VENDOR, sizeof(vendor_name), vendor_name, &returned_size);
err|= clGetDeviceInfo(device_id, CL_DEVICE_NAME, sizeof(device_name), device_name, &returned_size);
if (err != CL_SUCCESS)
{
printf("Error: Failed to retrieve device info!\n");
return EXIT_FAILURE;
}
printf(SEPARATOR);
printf("Connecting to %s %s...\n", vendor_name, device_name);
// Load the compute program from disk into a cstring buffer
//
typesize = (sizeof(float));
const char* filename = 0;
filename = "apple-reduce-kernel-float.cl";
printf(SEPARATOR);
printf("Loading program '%s'...\n", filename);
printf(SEPARATOR);
char *source = load_program_source(filename);
if(!source)
{
printf("Error: Failed to load compute program from file!\n");
return EXIT_FAILURE;
}
// Create a compute context
//
context = clCreateContext(0, 1, &device_id, NULL, NULL, &err);
if (!context)
{
printf("Error: Failed to create a compute context!\n");
return EXIT_FAILURE;
}
// Create a command queue
//
commands = clCreateCommandQueue(context, device_id, 0, &err);
if (!commands)
{
printf("Error: Failed to create a command commands!\n");
return EXIT_FAILURE;
}
// Create the input buffer on the device
//
size_t buffer_size = typesize * count * channels;
input_buffer = clCreateBuffer(context, CL_MEM_READ_WRITE, buffer_size, NULL, NULL);
if (!input_buffer)
{
printf("Error: Failed to allocate input buffer on device!\n");
return EXIT_FAILURE;
}
// Fill the input buffer with the host allocated random data
//
void *input_data = (void*)float_data;
err = clEnqueueWriteBuffer(commands, input_buffer, CL_TRUE, 0, buffer_size, input_data, 0, NULL, NULL);
if (err != CL_SUCCESS)
{
printf("Error: Failed to write to source array!\n");
return EXIT_FAILURE;
}
// Create an intermediate data buffer for intra-level results
//
partials_buffer = clCreateBuffer(context, CL_MEM_READ_WRITE, buffer_size, NULL, NULL);
if (!partials_buffer)
{
printf("Error: Failed to allocate partial sum buffer on device!\n");
return EXIT_FAILURE;
}
// Create the output buffer on the device
//
output_buffer = clCreateBuffer(context, CL_MEM_READ_WRITE, buffer_size, NULL, NULL);
if (!output_buffer)
{
printf("Error: Failed to allocate result buffer on device!\n");
return EXIT_FAILURE;
}
// Determine the reduction pass configuration for each level in the pyramid
//
create_reduction_pass_counts(
count, max_workgroup_size,
MAX_GROUPS, MAX_WORK_ITEMS,
&pass_count, &group_counts,
&work_item_counts, &operation_counts,
&entry_counts);
// Create specialized programs and kernels for each level of the reduction
//
cl_program *programs = (cl_program*)malloc(pass_count * sizeof(cl_program));
memset(programs, 0, pass_count * sizeof(cl_program));
cl_kernel *kernels = (cl_kernel*)malloc(pass_count * sizeof(cl_kernel));
memset(kernels, 0, pass_count * sizeof(cl_kernel));
for(i = 0; i < pass_count; i++)
{
char *block_source = malloc(strlen(source) + 1024);
size_t source_length = strlen(source) + 1024;
memset(block_source, 0, source_length);
// Insert macro definitions to specialize the kernel to a particular group size
//
const char group_size_macro[] = "#define GROUP_SIZE";
const char operations_macro[] = "#define OPERATIONS";
sprintf(block_source, "%s (%d) \n%s (%d)\n\n%s\n",
group_size_macro, (int)group_counts[i],
operations_macro, (int)operation_counts[i],
source);
// Create the compute program from the source buffer
//
programs[i] = clCreateProgramWithSource(context, 1, (const char **) & block_source, NULL, &err);
if (!programs[i] || err != CL_SUCCESS)
{
printf("%s\n", block_source);
printf("Error: Failed to create compute program!\n");
return EXIT_FAILURE;
}
// Build the program executable
//
err = clBuildProgram(programs[i], 0, NULL, NULL, NULL, NULL);
if (err != CL_SUCCESS)
{
size_t length;
char build_log[2048];
printf("%s\n", block_source);
printf("Error: Failed to build program executable!\n");
clGetProgramBuildInfo(programs[i], device_id, CL_PROGRAM_BUILD_LOG, sizeof(build_log), build_log, &length);
printf("%s\n", build_log);
return EXIT_FAILURE;
}
// Create the compute kernel from within the program
//
kernels[i] = clCreateKernel(programs[i], "reduce", &err);
if (!kernels[i] || err != CL_SUCCESS)
{
printf("Error: Failed to create compute kernel!\n");
return EXIT_FAILURE;
}
free(block_source);
}
// Do the reduction for each level
// this is one pass over it to establish the kernel args and such, so
// it is negligible time
//
cl_mem pass_swap;
cl_mem pass_input = output_buffer;
cl_mem pass_output = input_buffer;
for(i = 0; i < pass_count; i++)
{
size_t global = group_counts[i] * work_item_counts[i];
size_t local = work_item_counts[i];
unsigned int operations = operation_counts[i];
unsigned int entries = entry_counts[i];
size_t shared_size = typesize * channels * local * operations;
printf("Pass[%4d] Global[%4d] Local[%4d] Groups[%4d] WorkItems[%4d] Operations[%d] Entries[%d]\n", i,
(int)global, (int)local, (int)group_counts[i], (int)work_item_counts[i], operations, entries);
// Swap the inputs and outputs for each pass
//
pass_swap = pass_input;
pass_input = pass_output;
pass_output = pass_swap;
err = CL_SUCCESS;
err |= clSetKernelArg(kernels[i], 0, sizeof(cl_mem), &pass_output);
err |= clSetKernelArg(kernels[i], 1, sizeof(cl_mem), &pass_input);
err |= clSetKernelArg(kernels[i], 2, shared_size, NULL);
err |= clSetKernelArg(kernels[i], 3, sizeof(int), &entries);
if (err != CL_SUCCESS)
{
printf("Error: Failed to set kernel arguments!\n");
return EXIT_FAILURE;
}
// After the first pass, use the partial sums for the next input values
//
if(pass_input == input_buffer)
pass_input = partials_buffer;
err = CL_SUCCESS;
err |= clEnqueueNDRangeKernel(commands, kernels[i], 1, NULL, &global, &local, 0, NULL, NULL);
if (err != CL_SUCCESS)
{
printf("Error: Failed to execute kernel!\n");
return EXIT_FAILURE;
}
}
err = clFinish(commands);
if (err != CL_SUCCESS)
{
printf("Error: Failed to wait for command queue to finish! %d\n", err);
return EXIT_FAILURE;
}
// Start the timing loop and execute the kernel over several iterations
//
printf(SEPARATOR);
printf("Timing %d iterations of reduction with %d elements of type %s%s...\n",
iterations, count, "float",
(channels <= 1) ? (" ") : (channels == 2) ? "2" : "4");
printf(SEPARATOR);
int k;
err = CL_SUCCESS;
time_t t1 = clock();
for (k = 0 ; k < iterations; k++)
{
for(i = 0; i < pass_count; i++)
{
size_t global = group_counts[i] * work_item_counts[i];
size_t local = work_item_counts[i];
err = clEnqueueNDRangeKernel(commands, kernels[i], 1, NULL, &global, &local, 0, NULL, NULL);
if (err != CL_SUCCESS)
{
printf("Error: Failed to execute kernel!\n");
return EXIT_FAILURE;
}
}
}
err = clFinish(commands);
if (err != CL_SUCCESS)
{
printf("Error: Failed to wait for command queue to finish! %d\n", err);
return EXIT_FAILURE;
}
time_t t2 = clock();
// Calculate the statistics for execution time and throughput
//
double t = (t2-t1)/( (double)CLOCKS_PER_SEC );
printf("Exec Time: %.2f ms\n", t);
printf("Throughput: %.2f GB/sec\n", 1e-9 * buffer_size * iterations / t);
printf(SEPARATOR);
// Read back the results that were computed on the device
//
void *computed_result = malloc(typesize * channels);
memset(computed_result, 0, typesize * channels);
err = clEnqueueReadBuffer(commands, pass_output, CL_TRUE, 0, typesize * channels, computed_result, 0, NULL, NULL);
if (err)
{
printf("Error: Failed to read back results from the device!\n");
return EXIT_FAILURE;
}
// now do the speed test on standard
float reference=0;
t1 = clock();
for (k=0; k<iterations; k++) {
reference = reduce_validate_float(float_data, count);
}
t2 = clock();
double tcpu = (t2-t1)/( (double)CLOCKS_PER_SEC );
printf("CPU Exec Time: %.2f ms\n", tcpu);
printf("CPU Throughput: %.2f GB/sec\n", 1e-9 * buffer_size * iterations / tcpu);
printf("GPU is faster by %.16g\n", tcpu/t);
printf(SEPARATOR);
float result= ( (float *)computed_result )[0];
float ferror = fabs(reference - result)/reference;
if (ferror > MIN_ERROR)
{
printf("Result %.16g != %.16g\n", reference, result);
printf("Error: Incorrect results obtained! Rel error %.16g > Max allowed = %.16g\n", ferror, MIN_ERROR);
return EXIT_FAILURE;
}
else
{
printf("Results Validated!\n");
printf(SEPARATOR);
}
// Shutdown and cleanup
//
for(i = 0; i < pass_count; i++)
{
clReleaseKernel(kernels[i]);
clReleaseProgram(programs[i]);
}
clReleaseMemObject(input_buffer);
clReleaseMemObject(output_buffer);
clReleaseMemObject(partials_buffer);
clReleaseCommandQueue(commands);
clReleaseContext(context);
free(group_counts);
free(work_item_counts);
free(operation_counts);
free(entry_counts);
free(computed_result);
free(kernels);
free(float_data);
return 0;
}
int main(int argc, char **argv){
printf("Check OpenCL environtment\n");
cl_platform_id platid;
cl_device_id devid;
cl_int res;
size_t param;
/* Query OpenCL, get some information about the returned device */
clGetPlatformIDs(1u, &platid, NULL);
clGetDeviceIDs(platid, CL_DEVICE_TYPE_ALL, 1, &devid, NULL);
cl_char vendor_name[1024] = {0};
cl_char device_name[1024] = {0};
clGetDeviceInfo(devid, CL_DEVICE_VENDOR, sizeof(vendor_name), vendor_name, NULL);
clGetDeviceInfo(devid, CL_DEVICE_NAME, sizeof(device_name), device_name, NULL);
printf("Connecting to OpenCL device:\t%s %s\n", vendor_name, device_name);
clGetDeviceInfo(devid, CL_DEVICE_MAX_COMPUTE_UNITS, sizeof(cl_uint), ¶m, NULL);
printf("CL_DEVICE_MAX_COMPUTE_UNITS\t%d\n", param);
clGetDeviceInfo(devid, CL_DEVICE_MAX_WORK_GROUP_SIZE, sizeof(size_t), ¶m, NULL);
printf("CL_DEVICE_MAX_WORK_GROUP_SIZE\t%u\n", param);
clGetDeviceInfo(devid, CL_DEVICE_LOCAL_MEM_SIZE, sizeof(cl_ulong), ¶m, NULL);
printf("CL_DEVICE_LOCAL_MEM_SIZE\t%ub\n", param);
/* Check if kernel source exists, we compile argv[1] passed kernel */
if(argv[1] == NULL) { printf("\nUsage: %s kernel_source.cl kernel_function\n", argv[0]); exit(1); }
char *kernel_source;
if(load_program_source(argv[1], &kernel_source)) return 1;
printf("Building from OpenCL source: \t%s\n", argv[1]);
printf("Compile/query OpenCL_program:\t%s\n", argv[2]);
/* Create context and kernel program */
cl_context context = clCreateContext(0, 1, &devid, NULL, NULL, NULL);
cl_program pro = clCreateProgramWithSource(context, 1, (const char **)&kernel_source, NULL, NULL);
res = clBuildProgram(pro, 1, &devid, "-cl-fast-relaxed-math", NULL, NULL);
if(res != CL_SUCCESS){
printf("clBuildProgram failed: %d\n", res); char buf[0x10000];
clGetProgramBuildInfo(pro, devid, CL_PROGRAM_BUILD_LOG, 0x10000, buf, NULL);
printf("\n%s\n", buf); return(-1); }
cl_kernel kernelobj = clCreateKernel(pro, argv[2], &res); check_return(res);
/* Get the maximum work-group size for executing the kernel on the device */
size_t global, local;
res = clGetKernelWorkGroupInfo(kernelobj, devid, CL_KERNEL_WORK_GROUP_SIZE, sizeof(int), &local, NULL); check_return(res);
printf("CL_KERNEL_WORK_GROUP_SIZE\t%u\n", local);
res = clGetKernelWorkGroupInfo(kernelobj, devid, CL_KERNEL_LOCAL_MEM_SIZE, sizeof(cl_ulong), ¶m, NULL); check_return(res);
printf("CL_KERNEL_LOCAL_MEM_SIZE\t%ub\n", param);
cl_command_queue cmd_queue = clCreateCommandQueue(context, devid, CL_QUEUE_PROFILING_ENABLE, NULL);
if(cmd_queue == NULL) { printf("Compute device setup failed\n"); return(-1); }
local = 4;
int n = 2 * local; //num_group * local workgroup size
global = n;
int num_groups= global / local,
allocated_local= sizeof(data) * local +
sizeof(debug) * local;
data *DP __attribute__ ((aligned(16)));
DP = calloc(n, sizeof(data) *1);
debug *dbg __attribute__ ((aligned(16)));
dbg = calloc(n, sizeof(debug));
printf("global:%d, local:%d, (should be):%d groups\n", global, local, num_groups);
printf("structs size: %db, %db, %db\n", sizeof(data), sizeof(Elliptic_Curve), sizeof(inv256));
printf("sets:%d, total of %db needed, allocated _local: %db\n", n, n * sizeof(cl_uint4) *5 *4, allocated_local);
cl_mem cl_DP, cl_EC, cl_INV, DEBUG;
cl_DP = clCreateBuffer(context, CL_MEM_ALLOC_HOST_PTR, n * sizeof(data), NULL, &res); check_return(res);
cl_EC = clCreateBuffer(context, CL_MEM_ALLOC_HOST_PTR | CL_MEM_READ_ONLY, 1 * sizeof(Elliptic_Curve), NULL, &res); check_return(res); //_constant address space
cl_INV= clCreateBuffer(context, CL_MEM_ALLOC_HOST_PTR | CL_MEM_READ_ONLY, 1 * sizeof(u8) * 0x80, NULL, &res); check_return(res);
DEBUG = clCreateBuffer(context, CL_MEM_ALLOC_HOST_PTR | CL_MEM_WRITE_ONLY, n * sizeof(debug), NULL, &res); check_return(res);
Elliptic_Curve EC;
/*
Curve domain parameters, (test vectors)
-------------------------------------------------------------------------------------
p: c1c627e1638fdc8e24299bb041e4e23af4bb5427 is prime
a: c1c627e1638fdc8e24299bb041e4e23af4bb5424 divisor g = 62980
b: 877a6d84155a1de374b72d9f9d93b36bb563b2ab divisor g = 227169643
Gx: 010aff82b3ac72569ae645af3b527be133442131 divisor g = 32209245
Gy: 46b8ec1e6d71e5ecb549614887d57a287df573cc divisor g = 972
precomputed_per_curve_constants:
U: c1c627e1638fdc8e24299bb041e4e23af4bb5425
V: 3e39d81e9c702371dbd6644fbe1b1dc50b44abd9
already prepared mod p to test:
a: 07189f858e3f723890a66ec1079388ebd2ed509c
b: 6043379beb0dade6eed1e9d6de64f4a0c50639d4
gx: 5ef84aacf4f0ea6752f572d0741f40049f354dca
gy: 418c695435af6b3d4d7cbb72967395016ef67239
resulting point:
P.x: 01718f862ebe9423bd661a65355aa1c86ba330f8 program MUST got this point !!
P.y: 557e8ed53ffbfe2c990a121967b340f62e0e4fe2
taken mod p:
P.x: 41da1a8f74ff8d3f1ce20ef3e9d8865c96014fe3
P.y: 73ca143c9badedf2d9d3c7573307115ccfe04f13
*/
u8 *t;
t = _x_to_u8_buffer("c1c627e1638fdc8e24299bb041e4e23af4bb5427"); memcpy(EC.p, t, 20);
t = _x_to_u8_buffer("07189f858e3f723890a66ec1079388ebd2ed509c"); memcpy(EC.a, t, 20);
t = _x_to_u8_buffer("6043379beb0dade6eed1e9d6de64f4a0c50639d4"); memcpy(EC.b, t, 20);
t = _x_to_u8_buffer("5ef84aacf4f0ea6752f572d0741f40049f354dca"); memcpy(EC.Gx, t, 20);
t = _x_to_u8_buffer("418c695435af6b3d4d7cbb72967395016ef67239"); memcpy(EC.Gy, t, 20);
t = _x_to_u8_buffer("c1c627e1638fdc8e24299bb041e4e23af4bb5425"); memcpy(EC.U, t, 20);
t = _x_to_u8_buffer("3e39d81e9c702371dbd6644fbe1b1dc50b44abd9"); memcpy(EC.V, t, 20);
/* we need to map buffer now to load some k into data */
DP = clEnqueueMapBuffer(cmd_queue, cl_DP, CL_TRUE, CL_MAP_WRITE, 0, n * sizeof(data), 0, NULL, NULL, &res); check_return(res);
t = _x_to_u8_buffer("00542d46e7b3daac8aeb81e533873aabd6d74bb710");
for(u8 i = 0; i < n; i++) memcpy(DP[i].k, t, 21);
free(t);
//d for(u8 i = 0; i < n; i++) bn_print("", DP[i].k, 21, 1);
/* we can alter just a byte into a chosen k to verify that we'll get a different point! */
//DP[2].k[2] = 0x09;
//no res = clEnqueueWriteBuffer(cmd_queue, cl_DP, CL_TRUE, 0, n * sizeof(data), &DP, 0, NULL, NULL); check_return(res);
res = clEnqueueWriteBuffer(cmd_queue, cl_EC, CL_TRUE, 0, 1 * sizeof(Elliptic_Curve), &EC, 0, NULL, NULL); check_return(res);
res = clEnqueueWriteBuffer(cmd_queue, cl_INV, CL_TRUE, 0, 1 * sizeof(u8) * 0x80, &inv256, 0, NULL, NULL); check_return(res);
res = clSetKernelArg(kernelobj, 0, sizeof(cl_mem), &cl_DP); /* i/o buffer */
res|= clSetKernelArg(kernelobj, 1, sizeof(data) * local *1, NULL); //allocate space for __local in kernel (just this!) one * localsize
res|= clSetKernelArg(kernelobj, 2, sizeof(cl_mem), &cl_EC);
res|= clSetKernelArg(kernelobj, 3, sizeof(cl_mem), &cl_INV);
res|= clSetKernelArg(kernelobj, 4, sizeof(debug) * local *1, NULL); //allocate space for __local in kernel (just this!) one * localsize
res|= clSetKernelArg(kernelobj, 5, sizeof(cl_mem), &DEBUG); //this used to debug kernel output
check_return(res);
// printf("n:%d, total of %db needed, allocated _local: %db\n", n, n * sizeof(debug), allocated_local);
cl_event NDRangeEvent;
cl_ulong start, end;
/* Execute NDrange */
res = clEnqueueNDRangeKernel(cmd_queue, kernelobj, 1, NULL, &global, &local, 0, NULL, &NDRangeEvent); check_return(res);
// res = clEnqueueNDRangeKernel(cmd_queue, kernelobj, 1, NULL, &global, NULL, 0, NULL, &NDRangeEvent); check_return(res);
printf("Read back, Mapping buffer:\t%db\n", n * sizeof(data));
DP = clEnqueueMapBuffer(cmd_queue, cl_DP, CL_TRUE, CL_MAP_READ, 0, n * sizeof(data), 0, NULL, NULL, &res); check_return(res);
dbg =clEnqueueMapBuffer(cmd_queue, DEBUG, CL_TRUE, CL_MAP_READ, 0, n * sizeof(debug), 0, NULL, NULL, &res); check_return(res);
/* using clEnqueueReadBuffer template */
// res = clEnqueueReadBuffer(cmd_queue, ST, CL_TRUE, 0, sets * sizeof(cl_uint8), dbg, 0, NULL, NULL); check_return(res);
clFlush(cmd_queue);
clFinish(cmd_queue);
/* get NDRange execution time with internal ocl profiler */
res = clGetEventProfilingInfo(NDRangeEvent, CL_PROFILING_COMMAND_START, sizeof(cl_ulong), &start, NULL);
res|= clGetEventProfilingInfo(NDRangeEvent, CL_PROFILING_COMMAND_END, sizeof(cl_ulong), &end, NULL);
check_return(res);
printf("kernel execution time:\t\t%.2f ms\n", (float) ((end - start) /1000000)); //relative to NDRange call
printf("number of computes/sec:\t%.2f\n", (float) global *1000000 /((end - start)));
printf("i,\tgid\tlid0\tlsize0\tgid0/lsz0,\tgsz0,\tn_gr0,\tlid5,\toffset\n");
for(int i = 0; i < n; i++) {
// if(i %local == 0) {
printf("%d \t", i);
//printf("%u\t%u\t%u\t%u\t| %2u, %2u, %2u, %u\n", *p, *(p +1), *(p +2), *(p +3), *(p +4), *(p +5), *(p +6), *(p +7));
/* silence this doubled debug info
printf("%u\t%u\t%u\t%u\t| %2u, %2u, %2u, %u\n",
dbg[i].data[0], dbg[i].data[1], dbg[i].data[2], dbg[i].data[3],
dbg[i].data[4], dbg[i].data[5], dbg[i].data[6], dbg[i].data[7]);
*/
//printf("%d %d\n", P[i].dig, P[i].c);
bn_print("", DP[i].k, 21, 1);
bn_print("", DP[i].rx, 20, 0); bn_print(" ", DP[i].ry, 20, 1);
printf("%u(/%u) = %u*%u(/%u) +%u, offset:%u, stride:%u\n",
DP[i].pad[0], DP[i].pad[1], DP[i].pad[2], DP[i].pad[3],
DP[i].pad[4], DP[i].pad[5], DP[i].pad[6], DP[i].pad[7]);
// }
}
/* Release OpenCL stuff, free the rest */
clReleaseMemObject(cl_DP);
clReleaseMemObject(cl_EC);
clReleaseMemObject(cl_INV);
clReleaseMemObject(DEBUG);
clReleaseKernel(kernelobj);
clReleaseProgram(pro);
clReleaseCommandQueue(cmd_queue);
clReleaseContext(context);
free(kernel_source);
puts("Done!");
return 0;
}
int main(int argc, char *argv[]) {
/* Variables used to manage the OpenCL environment. */
cl_int rc;
size_t return_size[1];
unsigned int column_span = 0;
static cl_device_type device_type = CL_DEVICE_TYPE_DEFAULT;
static cl_uint kernel_type = KERNEL_DEFAULT;
static int gpu_wgsz = MAX_WGSZ;
/* The external file containing the matrix data in Matrix Market format */
static char *file_name;
/* These variables deal with the source file for the kernel, and the names of the kernels contained therein. */
char kernel_source_file[8] = "spmv.cl";
char kernel_name_LS[21] = "tiled_spmv_kernel_LS";
char kernel_name_AWGC[23] = "tiled_spmv_kernel_AWGC";
char kernel_name[32];
/* Basic "size of problem" variables. */
unsigned int nx; /* Number of elements in the X direction (length of the "input" vector. */
unsigned int ny; /* Number of elements in the Y direction (length of the "answer" vector. */
unsigned int non_zero; /* Number of non_zero elements in the matrix. */
unsigned int nx_pad, nyround; /* Rounded versions of nx and ny. */
/* Variables used to hold user-specified overrides and intermediate control values derived from them. */
unsigned int *slab_startrow = NULL;
unsigned int segcachesize;
unsigned int max_slabheight; /* Maximum matrix chunksize. */
unsigned int i, j, pdex = 0, ddex = 0;
size_t param_value_size_ret;
/* ================================================================================== */
/* Read in command line arguments. */
/* ================================================================================== */
int opt;
int option_index;
struct option long_options[] = {
{"help", no_argument, NULL, 'h'},
{"accel", no_argument, NULL, 'a'},
{"cpu", no_argument, NULL, 'c'},
{"gpu", no_argument, NULL, 'g'},
{"ls", no_argument, NULL, 'L'},
{"awgc", no_argument, NULL, 'A'},
{"verify", no_argument, NULL, 'v'},
{"lwgsize", required_argument, NULL, 'l'},
{"filename", required_argument, NULL, 'f'},
{NULL, 0, NULL, 0}
};
char *name;
/* ================================================================================== */
/* Change current working directory to that of the invocation path so that spmv can */
/* be run from any current working directory. */
/* ================================================================================== */
name = basename(argv[0]);
(void)chdir(dirname(argv[0]));
while (1) {
opt = getopt_long(argc, argv, "hacgLAl:f:", long_options, &option_index);
if (opt == -1) break;
switch (opt) {
/* -h, --help */
case 'h': usage(); exit(EXIT_SUCCESS);
/* -a, --accel */
case 'a': device_type = CL_DEVICE_TYPE_ACCELERATOR; break;
/* -c, --cpu */
case 'c': device_type = CL_DEVICE_TYPE_CPU; break;
/* -g, --gpu */
case 'g': device_type = CL_DEVICE_TYPE_GPU; break;
/* -L, --ls */
case 'L': kernel_type = KERNEL_LS; break;
/* -A, --awgc */
case 'A': kernel_type = KERNEL_AWGC; break;
/* -l, --lwgsize */
case 'l': gpu_wgsz = atoi(optarg); break;
/* -f, --filename */
case 'f':
posix_memalign((void **) &file_name, 128, 1+strlen(optarg));
strcpy(file_name, optarg);
break;
case '?':
printf("Try '%s --help' for more information.\n", name);
exit(EXIT_FAILURE);
}
}
if (optind != argc) {
printf("%s: unrecognized option '%s'.\n", name, argv[optind]);
printf("Try '%s --help' for more information.\n", name);
exit(EXIT_FAILURE);
}
/* ================================================================================== */
/* Start up OpenCL. */
/* ================================================================================== */
cl_uint preferred_alignment = 16; // used by "MEMORY_ALLOC_CHECK" macro
cl_uint num_platforms;
rc = clGetPlatformIDs(0, (cl_platform_id *) NULL, &num_platforms);
CHECK_RESULT("clGetPlatformIDs(num_platforms)")
platform_struct *platform;
MEMORY_ALLOC_CHECK(platform, num_platforms * sizeof(platform_struct), "platform");
cl_mem *buffer;
MEMORY_ALLOC_CHECK(buffer, num_platforms * sizeof(cl_mem), "buffer");
cl_platform_id *temp_platform_id_array;
MEMORY_ALLOC_CHECK(temp_platform_id_array, num_platforms * sizeof(cl_platform_id), "temp_platform_id_array");
rc = clGetPlatformIDs(num_platforms, temp_platform_id_array, (cl_uint *) NULL);
CHECK_RESULT("clGetPlatform IDs(Platform IDs)")
for (i=0; i<num_platforms; ++i) {
platform[i].id = temp_platform_id_array[i];
}
free(temp_platform_id_array);
printf("[START RUN]\n");
printf("command line: ");
for (i=0; i<(unsigned int) argc; ++i) {
printf("%s ", argv[i]);
}
printf("\n");
//printf("num_platforms = %d\n\n", num_platforms);
for (i=0; i<num_platforms; ++i) {
rc = clGetPlatformInfo(platform[i].id, CL_PLATFORM_NAME, (size_t) 0, NULL, (size_t *) ¶m_value_size_ret);
CHECK_RESULT("clGetPlatformInfo(size of platform name)")
MEMORY_ALLOC_CHECK(platform[i].name, param_value_size_ret, "platform name");
rc = clGetPlatformInfo(platform[i].id, CL_PLATFORM_NAME, param_value_size_ret, platform[i].name, (size_t *) NULL);
CHECK_RESULT("clGetPlatformInfo(platform name)")
rc = clGetDeviceIDs(platform[i].id, CL_DEVICE_TYPE_ALL, 0, NULL, (cl_uint *) &(platform[i].num_devices));
CHECK_RESULT("clGetDeviceIDs(number of devices)")
MEMORY_ALLOC_CHECK(platform[i].device, platform[i].num_devices * sizeof(device_struct), "device structure");
cl_device_id *tmpdevices;
MEMORY_ALLOC_CHECK(tmpdevices, platform[i].num_devices * sizeof(cl_device_id), "tmpdevices");
rc = clGetDeviceIDs(platform[i].id, CL_DEVICE_TYPE_ALL, platform[i].num_devices, tmpdevices, NULL);
CHECK_RESULT("clGetDeviceIDs(list of device IDs)")
for (j=0; j<platform[i].num_devices; ++j) {
platform[i].device[j].id = tmpdevices[j];
rc = clGetDeviceInfo(platform[i].device[j].id, CL_DEVICE_TYPE, sizeof(cl_device_type), &platform[i].device[j].type, NULL);
CHECK_RESULT("clGetDeviceInfo(device type)")
}
free(tmpdevices);
}
/* ================================================================================== */
/* Choose the best device to use, if one is not explicitly called for. */
/* If a device is specified, ensure that device is present on this hardware. */
/* ================================================================================== */
if (device_type == CL_DEVICE_TYPE_DEFAULT) {
int accel_found = 0;
for (i=0; i<num_platforms; ++i) {
for (j=0; j<platform[i].num_devices; ++j) {
if (platform[i].device[j].type == CL_DEVICE_TYPE_ACCELERATOR) {
accel_found = 1;
pdex = i;
ddex = j;
}
}
}
if (!accel_found) {
int gpu_found = 0;
for (i=0; i<num_platforms; ++i) {
for (j=0; j<platform[i].num_devices; ++j) {
if ((gpu_found == 0) && (platform[i].device[j].type == CL_DEVICE_TYPE_GPU)) {
gpu_found = 1;
pdex = i;
ddex = j;
}
}
}
if (!gpu_found) {
int cpu_found = 0;
for (i=0; i<num_platforms; ++i) {
for (j=0; j<platform[i].num_devices; ++j) {
if (platform[i].device[j].type == CL_DEVICE_TYPE_CPU) {
cpu_found = 1;
pdex = i;
ddex = j;
}
}
}
if (!cpu_found) {
fprintf(stderr, "no devices of any kind were found on this system. Leaving...\n");
fflush(stderr);
exit(EXIT_FAILURE);
}
}
}
}
else {
int device_found = 0;
for (i=0; i<num_platforms; ++i) for (j=0; j<platform[i].num_devices; ++j) {
if (platform[i].device[j].type == device_type) {
device_found = 1;
pdex = i;
ddex = j;
}
}
if (device_found == 0) {
fprintf(stderr, "no devices of the requested type were found on this system. Leaving...\n");
fflush(stderr);
exit(EXIT_FAILURE);
}
}
/* ================================================================================== */
/* Choose the best kernel to use, if one is not explicitly called for. */
/* ================================================================================== */
if (kernel_type == KERNEL_DEFAULT) {
kernel_type = (platform[pdex].device[ddex].type == CL_DEVICE_TYPE_ACCELERATOR) ? KERNEL_AWGC : KERNEL_LS;
}
/* ================================================================================== */
/* Create a context. */
/* ================================================================================== */
cl_context_properties properties[3];
properties[0] = CL_CONTEXT_PLATFORM;
properties[1] = (const cl_context_properties) platform[pdex].id;
properties[2] = 0;
platform[pdex].context = clCreateContext((const cl_context_properties *) properties, 1, &(platform[pdex].device[ddex].id), NULL, NULL, &rc);
CHECK_RESULT("clCreateContext")
/* ================================================================================== */
/* Build the kernel, create the Command Queue, and print kernel/device info. */
/* ================================================================================== */
switch (kernel_type) {
case KERNEL_LS:
strcpy(kernel_name, kernel_name_LS);
break;
case KERNEL_AWGC:
strcpy(kernel_name, kernel_name_AWGC);
break;
}
char *kernel_source;
kernel_source = load_program_source(kernel_source_file);
if (kernel_source == NULL) {
fprintf(stderr, "Error: Failed to load compute program from file!\n");
exit(EXIT_FAILURE);
}
platform[pdex].program = clCreateProgramWithSource(platform[pdex].context, 1, (const char **) &kernel_source, NULL, &rc);
CHECK_RESULT("clCreateProgramWithSource")
free(kernel_source);
rc = clBuildProgram(platform[pdex].program, 1, &(platform[pdex].device[ddex].id), "", NULL, NULL);
CHECK_RESULT("clBuildProgram")
platform[pdex].kernel = clCreateKernel(platform[pdex].program, kernel_name, &rc);
CHECK_RESULT("clCreateKernel")
platform[pdex].device[ddex].ComQ = clCreateCommandQueue(platform[pdex].context, platform[pdex].device[ddex].id, CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE, &rc);
CHECK_RESULT("clCreateCommandQueue")
rc = clGetDeviceInfo(platform[pdex].device[ddex].id, CL_DEVICE_NAME, (size_t) 0, NULL, (size_t *) ¶m_value_size_ret);
CHECK_RESULT("clGetDeviceInfo(size of CL_DEVICE_NAME)")
MEMORY_ALLOC_CHECK(platform[pdex].device[ddex].name, param_value_size_ret, "device name");
rc = clGetDeviceInfo(platform[pdex].device[ddex].id, CL_DEVICE_NAME, (size_t) param_value_size_ret, platform[pdex].device[ddex].name, (size_t *) NULL);
CHECK_RESULT("clGetDeviceInfo(CL_DEVICE_NAME)")
printf("We'll run kernel %s on device %s\n", ((kernel_type == KERNEL_LS) ? "kernel_ls" : "kernel_awgc"), platform[pdex].device[ddex].name);
/* ================================================================================== */
/* Determine device alignment, and whether "out-of-order" processing is supported. */
/* ================================================================================== */
rc = clGetDeviceInfo(platform[pdex].device[ddex].id, CL_DEVICE_MEM_BASE_ADDR_ALIGN, sizeof(cl_uint), &preferred_alignment, NULL);
CHECK_RESULT("clGetDeviceInfo(CL_DEVICE_MEM_BASE_ADDR_ALIGN)")
if (preferred_alignment > 1024) preferred_alignment = 1024;
preferred_alignment /= 8; /* Convert from units of bits to units of bytes. */
cl_command_queue_properties command_queue_properties;
clGetDeviceInfo (platform[pdex].device[ddex].id, CL_DEVICE_QUEUE_PROPERTIES, sizeof(cl_command_queue_properties), &command_queue_properties, NULL);
command_queue_properties &= CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE;
/* ================================================================================== */
/* Determine local memory size and maximum compute units. */
/* ================================================================================== */
size_t kernel_wg_size;
rc = clGetKernelWorkGroupInfo (platform[pdex].kernel, platform[pdex].device[ddex].id, CL_KERNEL_WORK_GROUP_SIZE, sizeof(size_t), (void *) &kernel_wg_size, return_size);
CHECK_RESULT("clGetKernelWorkGroupInfo(CL_KERNEL_WORK_GROUP_SIZE)")
cl_ulong total_local_mem;
rc = clGetDeviceInfo (platform[pdex].device[ddex].id, CL_DEVICE_LOCAL_MEM_SIZE, sizeof (cl_ulong), (void *) &total_local_mem, NULL);
CHECK_RESULT("clGetDeviceInfo(CL_DEVICE_LOCAL_MEM_SIZE)")
cl_ulong used_local_mem;
rc = clGetKernelWorkGroupInfo (platform[pdex].kernel, platform[pdex].device[ddex].id, CL_KERNEL_LOCAL_MEM_SIZE, sizeof (cl_ulong), &used_local_mem, NULL);
CHECK_RESULT("clGetKernelWorkGroupInfo(CL_KERNEL_LOCAL_MEM_SIZE)")
cl_ulong local_mem_size;
local_mem_size = total_local_mem - used_local_mem;
cl_uint max_compute_units;
clGetDeviceInfo (platform[pdex].device[ddex].id, CL_DEVICE_MAX_COMPUTE_UNITS, sizeof(cl_uint), &max_compute_units, NULL);
/* ================================================================================== */
/* Set up parameter structure and call the function that builds the tiled matrix. */
/* ================================================================================== */
matrix_gen_struct mgs;
unsigned int nslabs_round, memsize;
packet *seg_workspace;
slab_header *matrix_header;
unsigned int num_header_packets;
unsigned int *row_index_array = NULL;
unsigned int *x_index_array = NULL;
float *data_array = NULL;
mgs.matrix_header = &matrix_header;
mgs.seg_workspace = &seg_workspace;
mgs.num_header_packets = &num_header_packets;
mgs.row_index_array = &row_index_array;
mgs.x_index_array = &x_index_array;
mgs.data_array = &data_array;
mgs.nx_pad = &nx_pad;
mgs.nyround = &nyround;
mgs.slab_startrow = &slab_startrow;
mgs.nx = &nx;
mgs.ny = &ny;
mgs.non_zero = &non_zero;
mgs.file_name = (char *) file_name;
mgs.preferred_alignment = preferred_alignment;
mgs.max_compute_units = &max_compute_units;
mgs.kernel_type = kernel_type;
mgs.column_span = &column_span;
mgs.local_mem_size = (unsigned int) local_mem_size;
mgs.segcachesize = &segcachesize;
mgs.max_slabheight = &max_slabheight;
mgs.device_type = platform[pdex].device[ddex].type,
mgs.gpu_wgsz = &gpu_wgsz,
mgs.kernel_wg_size = kernel_wg_size;
mgs.nslabs_round = &nslabs_round;
mgs.memsize = &memsize;
rc = matrix_gen(&mgs);
/* =============================================================================================== */
/* Compute the local and global work group sizes. */
/* =============================================================================================== */
unsigned int ndims;
unsigned int team_size;
size_t global_work_size[3];
size_t local_work_size[3];
if (kernel_type == KERNEL_AWGC) {
ndims = 1;
global_work_size[0] = nslabs_round;
local_work_size[0] = 1;
}
else {
ndims = 2;
team_size = (platform[pdex].device[ddex].type == CL_DEVICE_TYPE_GPU) ? 16 : 1;
global_work_size[1] = nslabs_round;
local_work_size[1] = 1;
global_work_size[0] = local_work_size[0] = (platform[pdex].device[ddex].type == CL_DEVICE_TYPE_GPU) ? gpu_wgsz : CPU_WGSZ;
int max_aggregate_local_work_group_size = 0;
int aggregate_local_work_group_size = 1;
for (i=0; i<ndims; ++i) {
aggregate_local_work_group_size *= local_work_size[i];
}
max_aggregate_local_work_group_size = aggregate_local_work_group_size;
if (max_aggregate_local_work_group_size > (int) kernel_wg_size) {
while (max_aggregate_local_work_group_size > (int) kernel_wg_size) {
local_work_size[0] /= 2;
gpu_wgsz /= 2;
max_aggregate_local_work_group_size /= 2;
}
printf("coercing work group size to fit within hardware limits. New size is %d\n", gpu_wgsz);
}
}
/* =============================================================================================== */
/* Our Tiled format is now complete, but still in "working storage". We cannot allocate its */
/* buffer in OpenCL until we know how big it is, and now, we finally know how big it is. So, we */
/* create the Input and Output arrays, and the final array to hold the Tiled Format of the Matrix. */
/* =============================================================================================== */
/* Arrays to hold input and output data, and the finished tiled matrix data. */
float *input_array, *output_array, *output_array_verify;
unsigned int *tilebuffer;
MEMORY_ALLOC_CHECK(output_array_verify, (nyround * sizeof(float)), "output_array_verify")
if (output_array_verify == NULL) {
fprintf(stderr, "insufficient memory to perform this workload.\n"); fflush(stderr);
exit(EXIT_FAILURE);
}
cl_mem input_buffer;
cl_mem matrix_buffer;
cl_mem output_buffer;
unsigned int input_buffer_size;
unsigned int matrix_buffer_size;
/* Create the input and matrix buffer memory objects. */
input_buffer_size = (nx_pad * sizeof(float));
input_buffer = clCreateBuffer(platform[pdex].context, CL_MEM_ALLOC_HOST_PTR, input_buffer_size, NULL, &rc);
CHECK_RESULT("clCreateBuffer(input_buffer)")
matrix_buffer_size = memsize;
matrix_buffer = clCreateBuffer(platform[pdex].context, CL_MEM_ALLOC_HOST_PTR, matrix_buffer_size, NULL, &rc);
CHECK_RESULT("clCreateBuffer(matrix_buffer)")
cl_event events[2];
unsigned int output_buffer_size;
output_buffer_size = (slab_startrow[nslabs_round] - slab_startrow[0]) * sizeof(float);
output_buffer = clCreateBuffer(platform[pdex].context, CL_MEM_ALLOC_HOST_PTR, output_buffer_size, NULL, &rc);
CHECK_RESULT("clCreateBuffer(output_buffer)")
/* =============================================================================================== */
/* Map these buffers to allocate pointers into these buffers that we can use to load them. */
/* =============================================================================================== */
input_array = (float *) clEnqueueMapBuffer(platform[pdex].device[ddex].ComQ,
input_buffer,
CL_TRUE,
CL_MAP_WRITE,
0,
(size_t) input_buffer_size,
0,
NULL,
NULL,
&rc);
CHECK_RESULT("clEnqueueMapBuffer(input_array)")
tilebuffer = (unsigned int *) clEnqueueMapBuffer(platform[pdex].device[ddex].ComQ,
matrix_buffer,
CL_TRUE,
CL_MAP_WRITE,
0,
(size_t) matrix_buffer_size,
0,
NULL,
NULL,
&rc);
CHECK_RESULT("clEnqueueMapBuffer(tilebuffer)")
output_array = (float *) clEnqueueMapBuffer(platform[pdex].device[ddex].ComQ,
output_buffer,
CL_TRUE,
CL_MAP_WRITE,
0,
(size_t) output_buffer_size,
0,
NULL,
NULL,
&rc);
CHECK_RESULT("clEnqueueMapBuffer(output_array)")
/* =============================================================================================== */
/* Copy the tiled matrix into the memory buffer, and then unmap it. */
/* =============================================================================================== */
memcpy(tilebuffer, seg_workspace, sizeof(packet) * (matrix_header[nslabs_round].offset));
rc = clEnqueueUnmapMemObject(platform[pdex].device[ddex].ComQ, matrix_buffer, tilebuffer, 0, NULL, &events[0]);
CHECK_RESULT("clEnqueueUnmapMemObject(tilebuffer)")
clWaitForEvents(1, events);
/* Load random data into the input array. */
/* The user can substitute initialization of real data at this point in the code. */
for (i=0; i<nx; ++i) {
float rval;
rval = ((float) (rand() & 0x7fff)) * 0.001f - 15.0f;
input_array[i] = rval;
}
/* Zero out the output array. */
/* Note that this is only needed because some matrices are singular and have whole rows */
/* that are all zero, which is detected, and no work is done on those rows, so that they */
/* will never get written by the kernel, so to be safe, we zero it all out here, as well. */
memset((void *) output_array, 0, output_buffer_size);
/* =============================================================================================== */
/* Unmap the input and output memory buffers, to prepare for kernel execution. */
/* =============================================================================================== */
rc = clEnqueueUnmapMemObject(platform[pdex].device[ddex].ComQ, input_buffer, input_array, 0, NULL, &events[0]);
CHECK_RESULT("clEnqueueUnmapMemObject(input_array)")
rc = clEnqueueUnmapMemObject(platform[pdex].device[ddex].ComQ, output_buffer, output_array, 0, NULL, &events[1]);
CHECK_RESULT("clEnqueueUnmapMemObject(output_array)")
clWaitForEvents(2, events);
/* =============================================================================================== */
/* Execution: Multiplication of the input array times the Tiled Format of the Matrix. */
/* =============================================================================================== */
/* Run once to verifying correct answer, and computing a baseline number of repetitions for later performance measurements. */
rc = clSetKernelArg(platform[pdex].kernel, 0, sizeof(cl_mem), (const void *) &input_buffer);
CHECK_RESULT("clSetKernelArg(0)")
rc = clSetKernelArg(platform[pdex].kernel, 1, sizeof(cl_mem), (const void *) &output_buffer);
CHECK_RESULT("clSetKernelArg(1)")
rc = clSetKernelArg(platform[pdex].kernel, 2, sizeof(cl_mem), (const void *) &matrix_buffer);
CHECK_RESULT("clSetKernelArg(2)")
rc = clSetKernelArg(platform[pdex].kernel, 3, sizeof(cl_uint), &column_span);
CHECK_RESULT("clSetKernelArg(3)")
rc = clSetKernelArg(platform[pdex].kernel, 4, sizeof(cl_uint), &max_slabheight);
CHECK_RESULT("clSetKernelArg(4)")
if (kernel_type == KERNEL_LS) {
rc = clSetKernelArg(platform[pdex].kernel, 5, sizeof(cl_uint), &team_size);
CHECK_RESULT("clSetKernelArg(5)")
rc = clSetKernelArg(platform[pdex].kernel, 6, sizeof(cl_uint), &num_header_packets);
CHECK_RESULT("clSetKernelArg(6)")
rc = clSetKernelArg(platform[pdex].kernel, 7, (size_t) (max_slabheight * sizeof(float)), (void *) NULL);
CHECK_RESULT("clSetKernelArg(7)")
}
int main(int argc, char * argv[])
{
init_rpc(argv[1]);
cl_platform_id platform;
cl_device_id device;
cl_context context;
cl_command_queue queue;
cl_program program;
cl_kernel kernel;
cl_mem d_in_pos, d_in_vel, d_out_pos, d_out_vel;
int iterations = 100;
int num_bodies = 1024;
float espSqr = 500.0f;
float delT = 0.005f;
int exchange = 1;
size_t buf_size = 4 * num_bodies * sizeof(float);
float *ref_pos = (float *)malloc(buf_size);
float *ref_vel = (float *)malloc(buf_size);
int i, j;
for (i = 0; i < num_bodies; i++) {
int index = 4 * i;
for (j = 0; j < 3; ++j) {
ref_pos[index + j] = frandom(3, 50);
}
ref_pos[index + 3] = frandom(1, 1000);
for (j = 0; j < 3; ++j) {
ref_vel[index + j] = 0.0f;
}
ref_vel[3] = 0.0f;
}
size_t local_work_size[1];
size_t global_work_size[1];
local_work_size[0] = 256;
global_work_size[0] = num_bodies;
const char *source = load_program_source("NBody.cl");
size_t source_len = strlen(source);;
cl_uint err = 0;
char *flags = "";
clGetPlatformIDs(1, &platform, NULL);
printf("platform %p err %d\n", platform, err);
clGetDeviceIDs(platform, CL_DEVICE_TYPE_CPU, 1, &device, &err);
printf("device %p err %d\n", device, err);
context = clCreateContext(0, 1, &device, NULL, NULL, &err);
printf("context %p err %d\n", context, err);
queue = clCreateCommandQueue(context, device, 0, &err);
printf("queue %p err %d\n", queue, err);
program = clCreateProgramWithSource(context, 1, &source, &source_len, &err);
printf("program %p err %d\n", program, err);
err = clBuildProgram(program, 0, NULL, flags, NULL, NULL);
printf("err %d\n", err);
kernel = clCreateKernel(program, "nbody_sim", NULL);
printf("kernel %p\n", kernel);
d_in_pos = clCreateBuffer(context, CL_MEM_READ_WRITE | CL_MEM_COPY_HOST_PTR,
buf_size, ref_pos, &err);
printf("d_in_pos %p err %d\n", d_in_pos, err);
d_in_vel = clCreateBuffer(context, CL_MEM_READ_WRITE | CL_MEM_COPY_HOST_PTR,
buf_size, ref_vel, &err);
printf("d_in_vel %p err %d\n", d_in_vel, err);
d_out_pos = clCreateBuffer(context, CL_MEM_READ_WRITE,
buf_size, NULL, &err);
printf("d_out_pos %p err %d\n", d_out_pos, err);
d_out_vel = clCreateBuffer(context, CL_MEM_READ_WRITE,
buf_size, NULL, &err);
printf("d_out_vel %p err %d\n", d_out_vel, err);
err = clSetKernelArg(kernel, 0, sizeof(cl_mem), (void*)&d_in_pos);
printf("err %d\n", err);
err = clSetKernelArg(kernel, 1, sizeof(cl_mem), (void*)&d_in_vel);
printf("err %d\n", err);
err = clSetKernelArg(kernel, 2, sizeof(int), (void*)&num_bodies);
printf("err %d\n", err);
err = clSetKernelArg(kernel, 3, sizeof(float), (void*)&delT);
printf("err %d\n", err);
err = clSetKernelArg(kernel, 4, sizeof(float), (void*)&espSqr);
printf("err %d\n", err);
err = clSetKernelArg(kernel, 5, 256 * 4 * sizeof(float), NULL);
printf("err %d\n", err);
err = clSetKernelArg(kernel, 6, sizeof(cl_mem), (void*)&d_out_pos);
printf("err %d\n", err);
err = clSetKernelArg(kernel, 7, sizeof(cl_mem), (void*)&d_out_vel);
printf("err %d\n", err);
for (i = 0; i < iterations; i++) {
err = clEnqueueNDRangeKernel(queue, kernel, 1, NULL, global_work_size,
local_work_size, 0, NULL, NULL);
printf("err %d\n", err);
clFinish(queue);
err = clSetKernelArg(kernel, exchange ? 6 : 0, sizeof(cl_mem),
(void*)&d_in_pos);
printf("err %d\n", err);
err = clSetKernelArg(kernel, exchange ? 7 : 1, sizeof(cl_mem),
(void*)&d_in_vel);
printf("err %d\n", err);
err = clSetKernelArg(kernel, exchange ? 0 : 6, sizeof(cl_mem),
(void*)&d_out_pos);
printf("err %d\n", err);
err = clSetKernelArg(kernel, exchange ? 1 : 7, sizeof(cl_mem),
(void*)&d_out_vel);
printf("err %d\n", err);
exchange = exchange ? 0 : 1;
}
err = clEnqueueReadBuffer(queue, d_out_pos, CL_TRUE, 0, buf_size, ref_pos,
0, NULL, NULL);
printf("err %d\n", err);
err = clEnqueueReadBuffer(queue, d_out_vel, CL_TRUE, 0, buf_size, ref_vel,
0, NULL, NULL);
printf("err %d\n", err);
for (i = 0; i < num_bodies ; i++) {
printf("%i %f %f\n", i, ref_pos[i], ref_vel[i]);
}
clReleaseMemObject(d_in_pos);
clReleaseMemObject(d_in_vel);
clReleaseMemObject(d_out_pos);
clReleaseMemObject(d_out_vel);
clReleaseProgram(program);
clReleaseKernel(kernel);
clReleaseCommandQueue(queue);
}
int main(int argc, char *argv[])
{
cl_platform_id platform;
cl_device_id device;
cl_context context;
cl_command_queue queue;
cl_program program;
cl_kernel kernel_one, kernel_path;
cl_mem d_mt_state, d_mt_emit, d_max_prob_old;
cl_mem d_max_prob_new, d_path, v_prob, v_path;
int wg_size = 256;
int n_state = 256*16;
int n_emit = 128;
int n_obs = 100;
size_t init_prob_size = sizeof(float) * n_state;
size_t mt_state_size = sizeof(float) * n_state * n_state;
size_t mt_emit_size = sizeof(float) * n_emit * n_state;
float *init_prob = (float *) malloc(init_prob_size);
float *mt_state = (float *) malloc(mt_state_size);
float *mt_emit = (float *) malloc(mt_emit_size);
int *obs = (int *) malloc(sizeof(int) * n_obs);
int *viterbi_gpu = (int *) malloc(sizeof(int) * n_obs);
srand(2012);
initHMM(init_prob, mt_state, mt_emit, n_state, n_emit);
int i;
for (i = 0; i < n_obs; i++) {
obs[i] = i % 15;
}
const char *source = load_program_source("Viterbi.cl");
size_t source_len = strlen(source);;
cl_uint err = 0;
char *flags = "-cl-fast-relaxed-math";
clGetPlatformIDs(1, &platform, NULL);
printf("platform %p err %d\n", platform, err);
clGetDeviceIDs(platform, CL_DEVICE_TYPE_CPU, 1, &device, &err);
printf("device %p err %d\n", device, err);
context = clCreateContext(0, 1, &device, NULL, NULL, &err);
printf("context %p err %d\n", context, err);
queue = clCreateCommandQueue(context, device, 0, &err);
printf("queue %p err %d\n", queue, err);
program = clCreateProgramWithSource(context, 1, &source, &source_len, &err);
printf("program %p err %d\n", program, err);
err = clBuildProgram(program, 0, NULL, flags, NULL, NULL);
printf("err %d\n", err);
/*
char tmp[102400];
clGetProgramBuildInfo(program, device, CL_PROGRAM_BUILD_LOG, sizeof(tmp),
tmp, NULL);
printf("error %s\n", tmp);
*/
kernel_one = clCreateKernel(program, "ViterbiOneStep", &err);
printf("kernel %p err %d\n", kernel_one, err);
kernel_path = clCreateKernel(program, "ViterbiPath", &err);
printf("kernel %p err %d\n", kernel_path, err);
d_mt_state = clCreateBuffer(context, CL_MEM_READ_ONLY, mt_state_size,
NULL, &err);
printf("buffer %p\n", d_mt_state);
d_mt_emit = clCreateBuffer(context, CL_MEM_READ_ONLY, mt_emit_size,
NULL, &err);
printf("buffer %p\n", d_mt_emit);
d_max_prob_new = clCreateBuffer(context, CL_MEM_READ_WRITE,
init_prob_size, NULL, &err);
printf("buffer %p\n", d_max_prob_new);
d_max_prob_old = clCreateBuffer(context, CL_MEM_READ_WRITE,
init_prob_size, NULL, &err);
printf("buffer %p\n", d_max_prob_old);
d_path = clCreateBuffer(context, CL_MEM_READ_WRITE,
sizeof(int)*(n_obs-1)*n_state, NULL, &err);
printf("buffer %p\n", d_path);
v_prob = clCreateBuffer(context, CL_MEM_READ_WRITE, sizeof(float),
NULL, &err);
printf("buffer %p\n", v_prob);
v_path = clCreateBuffer(context, CL_MEM_READ_WRITE, sizeof(int)*n_obs,
NULL, &err);
printf("buffer %p\n", v_prob);
err = clEnqueueWriteBuffer(queue, d_mt_state, CL_TRUE, 0, mt_state_size,
mt_state, 0, NULL, NULL);
printf("err %d\n", err);
err = clEnqueueWriteBuffer(queue, d_mt_emit, CL_TRUE, 0, mt_emit_size,
mt_emit, 0, NULL, NULL);
printf("err %d\n", err);
err = clEnqueueWriteBuffer(queue, d_max_prob_old, CL_TRUE, 0, init_prob_size,
init_prob, 0, NULL, NULL);
printf("err %d\n", err);
// max_wg_size is 1024 for Intel Core 2 CPU
size_t max_wg_size;
err = clGetKernelWorkGroupInfo(kernel_one, device,
CL_KERNEL_WORK_GROUP_SIZE, sizeof(size_t), &max_wg_size, NULL);
printf("max_wg_size %d\n", max_wg_size);
size_t local_work_size[2], global_work_size[2];
local_work_size[0] = wg_size;
local_work_size[1] = 1;
global_work_size[0] = local_work_size[0] * 256;
global_work_size[1] = n_state/256;
for (i = 1; i < n_obs; i++) {
err = clSetKernelArg(kernel_one, 0, sizeof(cl_mem),
(void*)&d_max_prob_new);
printf("err %d\n", err);
err = clSetKernelArg(kernel_one, 1, sizeof(cl_mem),
(void*)&d_path);
printf("err %d\n", err);
err = clSetKernelArg(kernel_one, 2, sizeof(cl_mem),
(void*)&d_max_prob_old);
printf("err %d\n", err);
err = clSetKernelArg(kernel_one, 3, sizeof(cl_mem),
(void*)&d_mt_state);
printf("err %d\n", err);
err = clSetKernelArg(kernel_one, 4, sizeof(cl_mem),
(void*)&d_mt_emit);
printf("err %d\n", err);
err = clSetKernelArg(kernel_one, 5, sizeof(float)*local_work_size[0],
NULL);
printf("err %d\n", err);
err = clSetKernelArg(kernel_one, 6, sizeof(int)*local_work_size[0],
NULL);
printf("err %d\n", err);
err = clSetKernelArg(kernel_one, 7, sizeof(int),
(void*)&n_state);
printf("err %d\n", err);
err = clSetKernelArg(kernel_one, 8, sizeof(int),
(void*)&(obs[i]));
printf("err %d\n", err);
err = clSetKernelArg(kernel_one, 9, sizeof(int),
(void*)&i);
printf("err %d\n", err);
err = clEnqueueNDRangeKernel(queue, kernel_one, 2, NULL,
global_work_size, local_work_size, 0, NULL, NULL);
printf("err %d\n", err);
err = clEnqueueCopyBuffer(queue, d_max_prob_new, d_max_prob_old, 0, 0,
sizeof(float)*n_state, 0, NULL, NULL);
printf("err %d\n", err);
}
local_work_size[0] = 1;
global_work_size[0] = 1;
err = clSetKernelArg(kernel_path, 0, sizeof(cl_mem), (void*)&v_prob);
printf("err %d\n", err);
err = clSetKernelArg(kernel_path, 1, sizeof(cl_mem), (void*)&v_path);
printf("err %d\n", err);
err = clSetKernelArg(kernel_path, 2, sizeof(cl_mem),
(void*)&d_max_prob_new);
printf("err %d\n", err);
err = clSetKernelArg(kernel_path, 3, sizeof(cl_mem), (void*)&d_path);
printf("err %d\n", err);
err = clSetKernelArg(kernel_path, 4, sizeof(int), (void*)&n_state);
printf("err %d\n", err);
err = clSetKernelArg(kernel_path, 5, sizeof(int), (void*)&n_obs);
printf("err %d\n", err);
err = clEnqueueNDRangeKernel(queue, kernel_path, 1, NULL,
global_work_size, local_work_size, 0, NULL, NULL);
printf("err %d\n", err);
clFinish(queue);
printf("finish done\n");
err = clEnqueueReadBuffer(queue, v_path, CL_TRUE, 0, sizeof(int)*n_obs,
viterbi_gpu, 0, NULL, NULL);
printf("err %d\n", err);
for (i = 0; i < n_obs; i++) {
printf("%d %d\n", i, viterbi_gpu[i]);
}
clReleaseMemObject(d_mt_state);
clReleaseMemObject(d_mt_emit);
clReleaseMemObject(d_max_prob_old);
clReleaseMemObject(d_max_prob_new);
clReleaseMemObject(d_path);
clReleaseMemObject(v_prob);
clReleaseMemObject(v_path);
clReleaseProgram(program);
clReleaseKernel(kernel_one);
clReleaseKernel(kernel_path);
clReleaseCommandQueue(queue);
}
int main(int argc, char *argv[])
{
init_rpc(argv[1]);
cl_platform_id platform;
cl_device_id device;
cl_context context;
cl_command_queue queue;
cl_program program;
cl_kernel kernel;
cl_mem buffer;
size_t i;
int scale = 8; // scale should be higher than 8
size_t array_size = powl(2, scale) * 4;
cl_int *input = (cl_int *) malloc(sizeof(cl_int) * array_size);
cl_int *output = (cl_int *) malloc(sizeof(cl_int) * array_size);
cl_int dir = 1;
cl_int no_stages = 0;
cl_int temp;
generateInput(input, array_size);
//ExecuteSortReference(input, array_size, dir);
for (temp = array_size; temp > 2; temp >>= 1) {
no_stages++;
}
size_t local_work_size[1];
size_t global_work_size[1];
const char *source = load_program_source("BitonicSort.cl");
size_t source_len = strlen(source);;
cl_uint err = 0;
char *flags = "-cl-fast-relaxed-math";
clGetPlatformIDs(1, &platform, NULL);
printf("platform %p err %d\n", platform, err);
clGetDeviceIDs(platform, CL_DEVICE_TYPE_CPU, 1, &device, &err);
printf("device %p err %d\n", device, err);
context = clCreateContext(0, 1, &device, NULL, NULL, &err);
printf("context %p err %d\n", context, err);
queue = clCreateCommandQueue(context, device, 0, &err);
printf("queue %p err %d\n", queue, err);
program = clCreateProgramWithSource(context, 1, &source, &source_len, &err);
printf("program %p err %d\n", program, err);
err = clBuildProgram(program, 0, NULL, flags, NULL, NULL);
printf("err %d\n", err);
kernel = clCreateKernel(program, "BitonicSort", NULL);
printf("kernel %p\n", kernel);
buffer = clCreateBuffer(context, CL_MEM_COPY_HOST_PTR,
sizeof(cl_int) * array_size, input, &err);
printf("buffer %p err %d\n", buffer, err);
err = clSetKernelArg(kernel, 0, sizeof(cl_mem), (void*)&buffer);
printf("err %d\n", err);
err = clSetKernelArg(kernel, 3, sizeof(cl_int), (void*)&dir);
printf("err %d\n", err);
cl_int stage, pass_stage;
for (stage = 0; stage < no_stages; stage++) {
err = clSetKernelArg(kernel, 1, sizeof(cl_int), (void*)&stage);
printf("err %d\n", err);
for (pass_stage = stage; pass_stage >= 0; pass_stage--) {
err = clSetKernelArg(kernel, 2, sizeof(cl_int),
(void*)&pass_stage);
printf("err %d\n", err);
size_t gsz = array_size/(2*4);
global_work_size[0] = pass_stage ? gsz : gsz << 1;
local_work_size[0] = 128;
err = clEnqueueNDRangeKernel(queue, kernel, 1, NULL, global_work_size,
local_work_size, 0, NULL, NULL);
printf("err %d\n", err);
}
}
clFinish(queue);
err = clEnqueueReadBuffer(queue, buffer, CL_TRUE, 0,
sizeof(cl_int) * array_size, output, 0, NULL, NULL);
printf("err %d\n", err);
for (i = 0; i < array_size; i++) {
printf("%i %i\n", i, output[i]);
}
clReleaseMemObject(buffer);
clReleaseProgram(program);
clReleaseKernel(kernel);
clReleaseCommandQueue(queue);
}
int main( int argc, char* argv[] )
{
// Length of vectors
int m = atoi(argv[4]);
unsigned int n=(256*m);
//matrix variable
// OpenCL device memory for matrices
cl_mem d_A;
cl_mem d_B;
cl_mem d_C;
//########################Vector Add Variables
// Host input vectors
int *h_a;
int *h_b;
// Host output vector
int *h_c;
// Device input buffers
cl_mem d_a;
cl_mem d_b;
// Device output buffer
cl_mem d_c;
// cl_kernel *kernel;
cl_platform_id* cpPlatform; // OpenCL platform
cl_device_id device_id; // device ID
cl_context context; // context
//cl_command_queue* queue; // command queue
//cl_command_queue queue; // command queue
// cl_program *program; // program
cl_platform_id* platforms; // platform id,
// differnt for all the device we have in the system
cl_uint platformCount; //keeps the divice count
// Size, in bytes, of each vector
size_t bytes = n*sizeof(int);
// Allocate memory for each vector on host
h_a = (int*)malloc(bytes);
h_b = (int*)malloc(bytes);
h_c = (int*)malloc(bytes);
// Initialize vectors on host
int i;
for( i = 0; i < n; i++ )
{
h_a[i] = i;
h_b[i] = i;
// printf("%d ",h_a[i]);
}
size_t globalSize, localSize; //similar to cuda
cl_int err;//for errors
int workgrp;
int wrkitm;
int num_ker;
num_ker=atoi(argv[2]);
wrkitm=atoi(argv[3]);// i have tried automating lots of data,
// Number of work items in each local work group
localSize = wrkitm ;
// Number of total work items - localSize must be devisor
globalSize = n;
//################################# Done vector ###################
//#############Matrix Multiplication Variables ###############
int WA,HA,WB,HB,WC,HC;
WA = n;
HA = WA;
WB = WA;
HB = WB;
WC = WA;
HC = WA;
// set seed for rand()
srand(2006);
// 1. allocate host memory for matrices A and B
//automate the size of the matrix
unsigned int size_A = WA * HA;
unsigned int mem_size_A = sizeof(float) * size_A;
float* h_A = (float*) malloc(mem_size_A);
unsigned int size_B = WB * HB;
unsigned int mem_size_B = sizeof(float) * size_B;
float* h_B = (float*) malloc(mem_size_B);
// 4. allocate host memory for the result C
unsigned int size_C = WC * HC;
unsigned int mem_size_C = sizeof(float) * size_C;
float* h_C = (float*) malloc(mem_size_C);
// 2. initialize host memory
randomInit(h_A, size_A);
randomInit(h_B, size_B);
//######################## matrix done #######################
//mallocing for array of queues (break through)
cl_command_queue * queue = (cl_command_queue *)malloc(num_ker * sizeof(cl_command_queue));
cl_kernel *kernel=(cl_kernel *)malloc(num_ker * sizeof(cl_kernel));
cl_program *program=(cl_program *)malloc(num_ker * sizeof(cl_kernel));
//defining platform
clGetPlatformIDs(0, NULL, &platformCount);
cpPlatform = (cl_platform_id*) malloc(sizeof(cl_platform_id) * platformCount);
clGetPlatformIDs(platformCount, cpPlatform, NULL);//what ever is returned from last step will be used here
int choice = atoi(argv[1]);
if(choice ==1)
{
// we can have CL_DEVICE_GPU or ACCELERATOR or ALL as an option here
// we can these multiple times depending on requirements
err = clGetDeviceIDs(cpPlatform[0],CL_DEVICE_TYPE_CPU , 1, &device_id, NULL);
if (err != CL_SUCCESS)
printf("Error: Failed to create a device group!\n");
}
else
{
// Get ID for the device
err = clGetDeviceIDs(cpPlatform[1], CL_DEVICE_TYPE_GPU, 1, &device_id, NULL);
if (err != CL_SUCCESS)
{
printf("Error: Failed to create a device group!\n");
}
}
context = clCreateContext(0, 1, &device_id, NULL, NULL, &err);
//malloc file and kernel variable
char **file=(char **)malloc(num_ker * sizeof(char *));
char **KernelSource=(char **)malloc(num_ker * sizeof(char *));
for(i=0;i<num_ker;i++)
{
queue[i] = clCreateCommandQueue(context, device_id, 0, &err);
}
file[0]="vectadd.cl";
KernelSource[0] = load_program_source(file[0]);
file[1]="matxm.cl";
KernelSource[1] = load_program_source(file[1]);
for(i=0;i<num_ker;i++)
{
// Create the compute program from the source buffer
program[i] = clCreateProgramWithSource(context, 1,
(const char **) & KernelSource[i], NULL, &err);
// Build the program executable
clBuildProgram(program[i], 0, NULL, NULL, NULL, NULL);
// Create the compute kernel in the program we wish to run
kernel[i] = clCreateKernel(program[i], file[i], &err);
}
//Vector Start
// Create the input and output arrays in device memory for our calculation
d_a = clCreateBuffer(context, CL_MEM_READ_ONLY, bytes, NULL, NULL);
d_b = clCreateBuffer(context, CL_MEM_READ_ONLY, bytes, NULL, NULL);
d_c = clCreateBuffer(context, CL_MEM_WRITE_ONLY, bytes, NULL, NULL);
//vector finsih
//matrix start
d_C = clCreateBuffer(context, CL_MEM_READ_WRITE,
mem_size_A, NULL, &err);
d_A = clCreateBuffer(context,
CL_MEM_READ_WRITE,
mem_size_A, h_A, &err);
d_B = clCreateBuffer(context,
CL_MEM_READ_WRITE,
mem_size_B, h_B, &err);
//matrix finish
// Write our data set into the input array in device memory
for(i=0;i<num_ker;i++)
{
if(i=0)//for vectorADD
{
err = clEnqueueWriteBuffer(queue[i], d_a, CL_TRUE, 0,bytes, h_a, 0, NULL, NULL);
err = clEnqueueWriteBuffer(queue[i], d_b, CL_TRUE, 0,bytes, h_b, 0, NULL, NULL);
// Set the arguments to our compute kernel
err = clSetKernelArg(kernel[i], 0, sizeof(cl_mem), &d_a);
err = clSetKernelArg(kernel[i], 1, sizeof(cl_mem), &d_b);
err = clSetKernelArg(kernel[i], 2, sizeof(cl_mem), &d_c);
err = clSetKernelArg(kernel[i], 3, sizeof(unsigned int), &n);
// Get the maximum work group size for executing the kernel on the device
if (err != CL_SUCCESS)
{
printf("Error: Failed to retrieve kernel work group info! %d\n", err);
exit(1);
}
}
else if(i=1)
{ err = clEnqueueWriteBuffer(queue[i], d_A, CL_TRUE, 0,mem_size_A, h_A, 0, NULL, NULL);
err = clEnqueueWriteBuffer(queue[i], d_B, CL_TRUE, 0,mem_size_B, h_B, 0, NULL, NULL);
//size_t localWorkSize[2], globalWorkSize[2];
int wA = WA;
int wC = WC;
err = clSetKernelArg(kernel[i], 0,
sizeof(cl_mem), (void *)&d_C);
err = clSetKernelArg(kernel[i], 1,
sizeof(cl_mem), (void *)&d_A);
err = clSetKernelArg(kernel[i], 2,
sizeof(cl_mem), (void *)&d_B);
err = clSetKernelArg(kernel[i], 3,
sizeof(int), (void *)&wA);
err = clSetKernelArg(kernel[i], 4,
sizeof(int), (void *)&wC);
}
}
/* // Set the arguments to our compute kernel
err = clSetKernelArg(kernel, 0, sizeof(cl_mem), &d_a);
err = clSetKernelArg(kernel, 1, sizeof(cl_mem), &d_b);
err = clSetKernelArg(kernel, 2, sizeof(cl_mem), &d_c);
err = clSetKernelArg(kernel, 3, sizeof(unsigned int), &n);
// Get the maximum work group size for executing the kernel on the device
if (err != CL_SUCCESS)
{
printf("Error: Failed to retrieve kernel work group info! %d\n", err);
exit(1);
}
*/
//struct timeval tim;
//double t1,t2;
//gettimeofday(&tim, NULL);
//t1=tim.tv_sec+(tim.tv_usec/1000000.0);
//need to work on work size#############################
for(i=0;i<num_ker;i++)
{
err = clEnqueueNDRangeKernel(queue[i], kernel[i], 1, NULL, &globalSize, &localSize,
0, NULL, NULL);
}
//for(i=0;i<num_ker;i++)
//clFinish(queue[i]);
//gettimeofday(&tim, NULL);
// t2=tim.tv_sec+(tim.tv_usec/1000000.0);
//printf("GPU time %.4lf\t",(t2-t1));
for(i=0;i<num_ker;++i)
{
if(i=0)
{
clEnqueueReadBuffer(queue[i], d_c, CL_TRUE, 0,
bytes, h_c, 0, NULL, NULL );
}
else if(i=1)
{
err = clEnqueueReadBuffer(queue[i],
d_C, CL_TRUE, 0, mem_size_C,
h_C, 0, NULL, NULL);
}
}
for(i=0;i<num_ker;++i)
{
clFinish(queue[i]);
}
// release OpenCL resources
free(h_A);
free(h_B);
free(h_C);
clReleaseMemObject(d_A);
clReleaseMemObject(d_C);
clReleaseMemObject(d_B);
clReleaseMemObject(d_a);
clReleaseMemObject(d_b);
clReleaseMemObject(d_c);
// clReleaseProgram(program);
// clReleaseKernel(kernel);
for(i=0;i<num_ker;++i)
{
clReleaseCommandQueue(queue[i]);
clReleaseKernel(kernel[i]);
clReleaseProgram(program[i]);
}
clReleaseContext(context);
//release host memory
free(h_a);
free(h_b);
free(h_c);
return 0;
}
int main(int argc, char *argv[])
{
init_rpc(argv[1]);
cl_platform_id platform;
cl_device_id device;
cl_context context;
cl_command_queue queue;
cl_program program;
cl_kernel kernel;
cl_mem d_input_r;
cl_mem d_input_i;
int length = 1024;
size_t buf_size = length * sizeof(float);
float *input_r, *input_i, *output_r, *output_i;
posix_memalign((void **)&input_r, 16, buf_size);
posix_memalign((void **)&input_i, 16, buf_size);
posix_memalign((void **)&output_r, 16, buf_size);
posix_memalign((void **)&output_i, 16, buf_size);
fill_rand(input_r, length, 0, 255);
fill_rand(input_i, length, 0, 0);
memcpy(output_r, input_r, buf_size);
memcpy(output_i, input_i, buf_size);
size_t local_work_size[1];
size_t global_work_size[1];
local_work_size[0] = 64;
global_work_size[0] = 64;
const char *source = load_program_source("FFT.cl");
size_t source_len = strlen(source);;
cl_uint err = 0;
char *flags = "-x clc++";
clGetPlatformIDs(1, &platform, NULL);
printf("platform %p err %d\n", platform, err);
clGetDeviceIDs(platform, CL_DEVICE_TYPE_CPU, 1, &device, &err);
printf("device %p err %d\n", device, err);
context = clCreateContext(0, 1, &device, NULL, NULL, &err);
printf("context %p err %d\n", context, err);
queue = clCreateCommandQueue(context, device, 0, &err);
printf("queue %p err %d\n", queue, err);
program = clCreateProgramWithSource(context, 1, &source, &source_len, &err);
printf("program %p err %d\n", program, err);
err = clBuildProgram(program, 0, NULL, flags, NULL, NULL);
printf("err %d\n", err);
kernel = clCreateKernel(program, "kfft", NULL);
printf("kernel %p\n", kernel);
d_input_r = clCreateBuffer(context, CL_MEM_READ_WRITE | CL_MEM_COPY_HOST_PTR,
buf_size, input_r, &err);
printf("d_input_r %p err %d\n", d_input_r, err);
d_input_i = clCreateBuffer(context, CL_MEM_READ_WRITE | CL_MEM_COPY_HOST_PTR,
buf_size, input_i, &err);
printf("d_input_i %p err %d\n", d_input_i, err);
err = clSetKernelArg(kernel, 0, sizeof(cl_mem), (void*)&d_input_r);
printf("err %d\n", err);
err = clSetKernelArg(kernel, 1, sizeof(cl_mem), (void*)&d_input_i);
printf("err %d\n", err);
err = clEnqueueNDRangeKernel(queue, kernel, 1, NULL, global_work_size,
local_work_size, 0, NULL, NULL);
printf("err %d\n", err);
clFinish(queue);
err = clEnqueueReadBuffer(queue, d_input_r, CL_TRUE, 0, buf_size, output_r,
0, NULL, NULL);
printf("err %d\n", err);
err = clEnqueueReadBuffer(queue, d_input_i, CL_TRUE, 0, buf_size, output_i,
0, NULL, NULL);
printf("err %d\n", err);
int i;
for (i = 0; i < length; i++) {
printf("%i %f %f\n", i, output_r[i], output_i[i]);
}
clReleaseMemObject(d_input_r);
clReleaseMemObject(d_input_i);
clReleaseProgram(program);
clReleaseKernel(kernel);
clReleaseCommandQueue(queue);
}
int main(int argc, char** argv)
{
int err; // error code returned from api calls
float data[DATA_SIZE]; // original data set given to device
float results[DATA_SIZE]; // results returned from device
unsigned int correct; // number of correct results returned
size_t global; // global domain size for our calculation
size_t local; // local domain size for our calculation
cl_platform_id platform_id = NULL; // compute device platform id
cl_device_id device_id; // compute device id
cl_context context; // compute context
cl_command_queue commands; // compute command queue
cl_program program; // compute program
cl_kernel kernel; // compute kernel
cl_mem input; // device memory used for the input array
cl_mem output; // device memory used for the output array
cl_event event;
// Fill our data set with random float values
//
int i = 0;
unsigned int count = DATA_SIZE;
for(i = 0; i < count; i++)
data[i] = rand() / (float)RAND_MAX;
//Connect to a platform on device
err = clGetPlatformIDs(1, &platform_id, NULL);
if (err != CL_SUCCESS)
{
printf("Error: Failed to locate opencl platform!\n");
return EXIT_FAILURE;
}
// Connect to a compute device
//
int gpu = 0;
err = clGetDeviceIDs(platform_id, gpu ? CL_DEVICE_TYPE_GPU : CL_DEVICE_TYPE_CPU, 1, &device_id, NULL);
if (err != CL_SUCCESS)
{
printf("Error: Failed to create a device group!\n");
return EXIT_FAILURE;
}
// Create a compute context
//
context = clCreateContext(0, 1, &device_id, NULL, NULL, &err);
if (!context)
{
printf("Error: Failed to create a compute context!\n");
return EXIT_FAILURE;
}
// Create a command commands
//
commands = clCreateCommandQueue(context, device_id, CL_QUEUE_PROFILING_ENABLE, &err);
if (!commands)
{
printf("Error: Failed to create a command commands!\n");
return EXIT_FAILURE;
}
//Use function and load the kernel source from .cl files in the same folder
//
char *KernelSource = load_program_source("hello.cl");
// Create the compute program from the source buffer
//
program = clCreateProgramWithSource(context, 1, (const char **) & KernelSource, NULL, &err);
if (!program)
{
printf("Error: Failed to create compute program!\n");
return EXIT_FAILURE;
}
// Build the program executable
//
err = clBuildProgram(program, 0, NULL, NULL, NULL, NULL);
if (err != CL_SUCCESS)
{
size_t len;
char buffer[2048];
printf("Error: Failed to build program executable!\n");
clGetProgramBuildInfo(program, device_id, CL_PROGRAM_BUILD_LOG, sizeof(buffer), buffer, &len);
printf("%s\n", buffer);
exit(1);
}
// Create the compute kernel in the program we wish to run
//
kernel = clCreateKernel(program, "square", &err);
if (!kernel || err != CL_SUCCESS)
{
printf("Error: Failed to create compute kernel! - %d\n",err);
exit(1);
}
// Create the input and output arrays in device memory for our calculation
//
input = clCreateBuffer(context, CL_MEM_READ_ONLY, sizeof(float) * count, NULL, NULL);
output = clCreateBuffer(context, CL_MEM_WRITE_ONLY, sizeof(float) * count, NULL, NULL);
if (!input || !output)
{
printf("Error: Failed to allocate device memory!\n");
exit(1);
}
// Write our data set into the input array in device memory
//
err = clEnqueueWriteBuffer(commands, input, CL_TRUE, 0, sizeof(float) * count, data, 0, NULL, NULL);
if (err != CL_SUCCESS)
{
printf("Error: Failed to write to source array!\n");
exit(1);
}
// Set the arguments to our compute kernel
//
err = 0;
err = clSetKernelArg(kernel, 0, sizeof(cl_mem), &input);
err |= clSetKernelArg(kernel, 1, sizeof(cl_mem), &output);
if (err != CL_SUCCESS)
{
printf("Error: Failed to set kernel arguments! %d\n", err);
exit(1);
}
// Get the maximum work group size for executing the kernel on the device
//
err = clGetKernelWorkGroupInfo(kernel, device_id, CL_KERNEL_WORK_GROUP_SIZE, sizeof(local), &local, NULL);
if (err != CL_SUCCESS)
{
printf("Error: Failed to retrieve kernel work group info! %d\n", err);
exit(1);
}
// Execute the kernel over the entire range of our 1d input data set
// using the maximum number of work group items for this device
//
global = count;
err = clEnqueueNDRangeKernel(commands, kernel, 1, NULL, &global, NULL, 0, NULL, &event);
if (err)
{
printf("Error: Failed to execute kernel!-%d\n",err);
return EXIT_FAILURE;
}
// Wait for the command commands to get serviced before reading back results
//
clWaitForEvents(1, &event);
clFinish(commands);
cl_ulong time_start, time_end;
double total_time;
clGetEventProfilingInfo(event, CL_PROFILING_COMMAND_START, sizeof(time_start), &time_start, NULL);
clGetEventProfilingInfo(event, CL_PROFILING_COMMAND_END, sizeof(time_end), &time_end, NULL);
total_time = time_end - time_start;
printf("cl:main timing:opencl clEnqueueNDRangeKernel %0.3f us\n", total_time / 1000.0);
// Read back the results from the device to verify the output
//
err = clEnqueueReadBuffer( commands, output, CL_TRUE, 0, sizeof(float) * count, results, 0, NULL, NULL );
if (err != CL_SUCCESS)
{
printf("Error: Failed to read output array! %d\n", err);
exit(1);
}
// Validate our results
//
correct = 0;
for(i = 0; i < count; i++)
{
if(results[i] == data[i] * data[i])
correct++;
}
// Print a brief summary detailing the results
//
printf("Computed '%d/%d' correct values!\n", correct, count);
// Shutdown and cleanup
//
clReleaseMemObject(input);
clReleaseMemObject(output);
clReleaseProgram(program);
clReleaseKernel(kernel);
clReleaseCommandQueue(commands);
clReleaseContext(context);
return 0;
}
int main(int argc, char* argv[])
{
int num_ker=0,num_queue;
num_ker=atoi(argv[2]);
num_queue=atoi(argv[3]);
//variables
/*#define WA 1024
#define HA 1024
#define WB 1024
#define HB WA
#define WC WB
#define HC HA
*/
struct timeval tim,ftim;
double t1,t2,tim1,tim2;
// gettimeofday(&tim, NULL);
// t1=tim.tv_sec+(tim.tv_usec/1000000.0);
gettimeofday(&ftim, NULL);
tim1=ftim.tv_sec+(ftim.tv_usec/1000000.0);
int m,WA,HA,WB,HB,WC,HC;
m = atoi(argv[5]);
WA=(256*m);
HA = WA;
WB = WA;
HB = WB;
WC = WA;
HC = WA;
// set seed for rand()
srand(2006);
// 1. allocate host memory for matrices A and B
//automate the size of the matrix
unsigned int size_A = WA * HA;
unsigned int mem_size_A = sizeof(int) * size_A;
int* h_A = (int*) malloc(mem_size_A);
unsigned int size_B = WB * HB;
unsigned int mem_size_B = sizeof(int) * size_B;
int* h_B = (int*) malloc(mem_size_B);
// 2. initialize host memory
randomInit(h_A, size_A);
randomInit(h_B, size_B);
/* // 3. print out A and B
printf("\n\nMatrix A\n");
for(i = 0; i < size_A; i++)
{
printf("%f ", h_A[i]);
if(((i + 1) % WA) == 0)
printf("\n");
}
printf("\n\nMatrix B\n");
for(i = 0; i < size_B; i++)
{
printf("%f ", h_B[i]);
if(((i + 1) % WB) == 0)
printf("\n");
}
*/
// 4. allocate host memory for the result C
unsigned int size_C = WC * HC;
unsigned int mem_size_C = sizeof(int) * size_C;
int* h_C = (int*) malloc(mem_size_C);
// 5. Initialize OpenCL
// OpenCL specific variables
cl_context clGPUContext;
// cl_command_queue* clCommandQue;
//cl_program clProgram;
//cl_kernel clKernel;
cl_platform_id* cpPlatform; // OpenCL platform
cl_uint platformCount; //keeps the divice count
size_t dataBytes;
size_t kernelLength;
cl_int errcode;
// OpenCL device memory for matrices
cl_mem d_A;
cl_mem d_B;
cl_mem d_C;
/*****************************************/
/* Initialize OpenCL */
/*****************************************/
//cl_platform_id* cpPlatform; // OpenCL platform
//cl_device_id device_id;// = (cl_device_id)malloc(sizeof(cl_device_id));
// Bind to platform
// errcode = clGetPlatformIDs(1, &cpPlatform, NULL);
clGetPlatformIDs(0, NULL, &platformCount);
cpPlatform = (cl_platform_id*) malloc(sizeof(cl_platform_id) * platformCount);
clGetPlatformIDs(platformCount, cpPlatform, NULL);//what ever is returned from last step will be used here
cl_device_id device_id;
int choice =atoi(argv[1]);
if(choice ==1)
{
// Length of vectors
// n = 64;
// Connect to a compute device
// we can have CL_DEVICE_GPU or ACCELERATOR or ALL as an option here
//depending what device are we working on
// we can these multiple times depending on requirements
errcode = clGetDeviceIDs(cpPlatform[0],CL_DEVICE_TYPE_CPU , 1, &device_id, NULL);
if (errcode != CL_SUCCESS)
printf("Error: Failed to create a device group!\n");
}
else
{
// errcode = clGetPlatformIDs(1, &cpPlatform, NULL);
// Get ID for the device
errcode = clGetDeviceIDs(cpPlatform[1], CL_DEVICE_TYPE_GPU, 1, &device_id, NULL);
if (errcode != CL_SUCCESS)
{
printf("Error: Failed to create a device group!\n");
}
}
//printf("here");
// Create a context
clGPUContext = clCreateContext(0, 1, &device_id, NULL, NULL, &errcode);
// Create a command queue
//printf("here");
/*clGPUContext = clCreateContextFromType(NULL,
CL_DEVICE_TYPE_GPU,
NULL, NULL, &errcode);
//shrCheckError(errcode, CL_SUCCESS);
// get the list of GPU devices associated
// with context
errcode = clGetContextInfo(clGPUContext,
CL_CONTEXT_DEVICES, 0, NULL,
&dataBytes);
cl_device_id *clDevices = (cl_device_id *)
malloc(dataBytes);
errcode = clGetContextInfo(clGPUContext,
CL_CONTEXT_DEVICES, dataBytes,
clDevices, NULL);
//shrCheckError(errcode, CL_SUCCESS);
*/
//malloc for command queue, kernel and program
cl_kernel *clKernel=(cl_kernel *)malloc(num_ker * sizeof(cl_kernel));
cl_program *clProgram=(cl_program *)malloc(num_ker * sizeof(cl_kernel));
cl_command_queue * clCommandQue = (cl_command_queue *)malloc(num_ker * sizeof(cl_command_queue));
//Create a command-queue
for(i=0;i<num_queue;i++)
{
clCommandQue[i] = clCreateCommandQueue(clGPUContext,
device_id, 0, &errcode);
} //shrCheckError(errcode, CL_SUCCESS);
/* // Setup device memory
d_C = clCreateBuffer(clGPUContext,
CL_MEM_READ_WRITE,
mem_size_A, NULL, &errcode);
d_A = clCreateBuffer(clGPUContext,
printf("\nhere");
CL_MEM_READ_WRITE | CL_MEM_COPY_HOST_PTR,
mem_size_A, h_A, &errcode);
d_B = clCreateBuffer(clGPUContext,
CL_MEM_READ_WRITE | CL_MEM_COPY_HOST_PTR,
mem_size_B, h_B, &errcode);
*/
char *file="matxm.cl";
char *KernelSource = load_program_source(file);
for(i=0;i<num_ker;i++)
{
clProgram[i] = clCreateProgramWithSource(clGPUContext,
1, (const char **) & KernelSource,
NULL, &errcode);
//shrCheckError(errcode, CL_SUCCESS);
errcode = clBuildProgram(clProgram[i], 0,
NULL, NULL, NULL, NULL);
//shrCheckError(errcode, CL_SUCCESS);
clKernel[i] = clCreateKernel(clProgram[i],
"matrixMul", &errcode);
}
//shrCheckError(errcode, CL_SUCCESS);
// Setup device memory
d_C = clCreateBuffer(clGPUContext, CL_MEM_READ_WRITE,
mem_size_A, NULL, &errcode);
d_A = clCreateBuffer(clGPUContext,
CL_MEM_READ_WRITE,
mem_size_A, h_A, &errcode);
d_B = clCreateBuffer(clGPUContext,
CL_MEM_READ_WRITE,
mem_size_B, h_B, &errcode);
// Write our data set into the input array in device memory
for(i=0;i<num_queue;i++){
errcode = clEnqueueWriteBuffer(clCommandQue[i], d_A, CL_TRUE, 0,mem_size_A, h_A, 0, NULL, NULL);
errcode = clEnqueueWriteBuffer(clCommandQue[i], d_B, CL_TRUE, 0,mem_size_B, h_B, 0, NULL, NULL);
}
// 7. Launch OpenCL kernel
size_t localWorkSize[2], globalWorkSize[2];
int wA = WA;
int wC = WC;
for(i=0;i<num_ker;i++)
{
errcode = clSetKernelArg(clKernel[i], 0,
sizeof(cl_mem), (void *)&d_C);
errcode = clSetKernelArg(clKernel[i], 1,
sizeof(cl_mem), (void *)&d_A);
errcode = clSetKernelArg(clKernel[i], 2,
sizeof(cl_mem), (void *)&d_B);
errcode = clSetKernelArg(clKernel[i], 3,
sizeof(int), (void *)&wA);
errcode = clSetKernelArg(clKernel[i], 4,
sizeof(int), (void *)&wC);
}
// shrCheckError(errcode, CL_SUCCESS);
//struct timespec start, finish;
int value;
value =atoi(argv[4]);
localWorkSize[0] = value ;
localWorkSize[1] = value ;
globalWorkSize[0] = HA;
globalWorkSize[1] = HA;
//clFinish(clCommandQue);
//timer starting
// clock_gettime(CLOCK_MONOTONIC, &start);
//struct timeval tim;
//double t1,t2;
// gettimeofday(&tim, NULL);
// t1=tim.tv_sec+(tim.tv_usec/1000000.0);
gettimeofday(&tim, NULL);
t1=tim.tv_sec+(tim.tv_usec/1000000.0);
//multikernels inside queues
int j=0;
for(j=0;j<num_queue;j++)
{
for(i=0;i<num_ker;i++){
errcode = clEnqueueNDRangeKernel(clCommandQue[j],
clKernel[i], 2, NULL, globalWorkSize,
localWorkSize, 0, NULL, NULL);
}
}
for(i=0;i<num_queue;i++)
{
clFinish(clCommandQue[i]);
}
gettimeofday(&tim, NULL);
t2=tim.tv_sec+(tim.tv_usec/1000000.0);
printf("%.6lf\t",(t2-t1));
// shrCheckError(errcode, CL_SUCCESS);
/* clock_gettime(CLOCK_MONOTONIC, &finish);
elapsed = (finish.tv_sec - start.tv_sec);
elapsed += (finish.tv_nsec - start.tv_nsec)/ 1000000000.0;
printf("Work Item/threads = %d \n",value);
printf("time taken by GPU = %le\n ",elapsed);
*/
// 8. Retrieve result from device
for(i=0;i<num_queue;i++)
{
errcode = clEnqueueReadBuffer(clCommandQue[i],
d_C, CL_TRUE, 0, mem_size_C,
h_C, 0, NULL, NULL);
//shrCheckError(errcode, CL_SUCCESS);
}
for(i=0;i<num_queue;i++)
{
clFinish(clCommandQue[i]);
}
// shrCheckError(errcode, CL_SUCCESS);
//clock_gettime(CLOCK_MONOTONIC, &finish);
// elapsed = (finish.tv_sec - start.tv_sec);
// elapsed += (finish.tv_nsec - start.tv_nsec)/ 1000000000.0;
//printf("Work Item/threads = %d \n",value);
//printf("time taken by GPU = %le\n ",elapsed);
// 9. print out the results
/*printf("\n\nMatrix C (Results)\n");
for(i = 0; i < size_C; i++)
{
printf("%f ", h_C[i]);
if(((i + 1) % WC) == 0)
printf("\n");
}
printf("\n");*/
// 10. clean up memory
free(h_A);
free(h_B);
free(h_C);
clReleaseMemObject(d_A);
clReleaseMemObject(d_C);
clReleaseMemObject(d_B);
// free(device_id);
free(KernelSource);
clReleaseContext(clGPUContext);
for(i=0;i<num_ker;i++)
{
clReleaseKernel(clKernel[i]);
clReleaseProgram(clProgram[i]);
}
for(i=0;i<num_queue;i++){
clReleaseCommandQueue(clCommandQue[i]);
}
gettimeofday(&ftim, NULL);
tim2=ftim.tv_sec+(ftim.tv_usec/1000000.0);
printf("%.6lf\t",(tim2-tim1));
printf("\n");
exit(0);
}
int main(int argc, char* argv[])
{
int device_gpu = 1;
const char *source_files[1] = {
"mtgp32-opencl.cl"};
const char *buildOptions="-I. -Werror";
const char *program_source[1];
cl_int clerr;
cl_platform_id platform_ids[32];
unsigned int num_platforms;
clerr = clGetPlatformIDs(32, platform_ids, &num_platforms);
CLERR;
for (unsigned int i=0; i < num_platforms; ++i)
{
clerr = clGetDeviceIDs (platform_ids[i], device_gpu ? CL_DEVICE_TYPE_GPU : CL_DEVICE_TYPE_CPU, 1, &device_id, NULL);
if (CL_SUCCESS == clerr)
{
platform_id = platform_ids[i];
break;
}
else if (CL_DEVICE_NOT_FOUND == clerr)
continue;
CLERR;
}
{
char platform_name[1024];
char platform_vendor[1024];
char device_name[1024];
clerr = clGetPlatformInfo(platform_id, CL_PLATFORM_NAME, sizeof(platform_name), platform_name, NULL);
CLERR;
clerr = clGetPlatformInfo(platform_id, CL_PLATFORM_VENDOR, sizeof(platform_vendor), platform_vendor, NULL);
CLERR;
clerr = clGetDeviceInfo(device_id, CL_DEVICE_NAME, sizeof(device_name), device_name, NULL);
CLERR;
printf("Platform name: %s\nPlatform vendor: %s\nDevice name: %s\n", platform_name, platform_vendor, device_name);
}
context = clCreateContext(NULL, 1, &device_id, NULL, NULL, &clerr);
CLERR;
commands = clCreateCommandQueue(context, device_id, 0, &clerr);
CLERR;
program_source[0] = load_program_source(source_files[0]);
program = clCreateProgramWithSource(context, 1, program_source, NULL, &clerr);
CLERR;
clerr = clBuildProgram(program, 0, NULL, buildOptions, NULL, NULL);
//CLERR;
size_t log_size;
clerr = clGetProgramBuildInfo(program, device_id, CL_PROGRAM_BUILD_LOG, 0, NULL, &log_size);
CLERR;
char* build_log = (char*) malloc(log_size);
clerr = clGetProgramBuildInfo(program, device_id, CL_PROGRAM_BUILD_LOG, log_size, build_log, NULL);
CLERR;
build_log[log_size-1] = '\0';
printf("BUILD LOG %s\n", build_log);
free(build_log);
free((void*)program_source[0]);
mtgp32_uint32_kernel = clCreateKernel(program, "mtgp32_uint32_kernel", &clerr); CLERR;
mtgp32_single_kernel = clCreateKernel(program, "mtgp32_single_kernel", &clerr); CLERR;
// LARGE_SIZE is a multiple of 16
int num_data = 10000000;
int block_num;
int num_unit;
int r;
cl_mem d_status;
cl_mem d_params;
int mb, mp;
block_num = 96;
/* if (argc >= 2) {
errno = 0;
block_num = strtol(argv[1], NULL, 10);
if (errno) {
printf("%s number_of_block number_of_output\n", argv[0]);
return 1;
}
if (block_num < 1 || block_num > BLOCK_NUM_MAX) {
printf("%s block_num should be between 1 and %d\n",
argv[0], BLOCK_NUM_MAX);
return 1;
}
errno = 0;
num_data = strtol(argv[2], NULL, 10);
if (errno) {
printf("%s number_of_block number_of_output\n", argv[0]);
return 1;
}
argc -= 2;
argv += 2;
} else {
printf("%s number_of_block number_of_output\n", argv[0]);
block_num = get_suitable_block_num(device,
&mb,
&mp,
sizeof(uint32_t),
THREAD_NUM,
LARGE_SIZE);
if (block_num <= 0) {
printf("can't calculate sutable number of blocks.\n");
return 1;
}
printf("the suitable number of blocks for device 0 "
"will be multiple of %d, or multiple of %d\n", block_num,
(mb - 1) * mp);
return 1;
}
*/
num_unit = LARGE_SIZE * block_num;
d_status = clCreateBuffer(context, CL_MEM_READ_WRITE, sizeof(struct mtgp32_kernel_status_t) * block_num, NULL, &clerr); CLERR;
d_params = clCreateBuffer(context, CL_MEM_READ_ONLY, sizeof(struct mtgp32_param_t), NULL, &clerr); CLERR;
// ccudaMalloc((void**)&d_status, sizeof(mtgp32_kernel_status_t) * block_num);
r = num_data % num_unit;
if (r != 0) {
num_data = num_data + num_unit - r;
}
make_constant(commands, d_params, MTGPDC_PARAM_TABLE, block_num);
make_kernel_data32(commands, d_status, MTGPDC_PARAM_TABLE, block_num);
make_uint32_random(d_status, d_params, num_data, block_num);
make_single_random(d_status, d_params, num_data, block_num);
clReleaseMemObject(d_status);
clReleaseMemObject(d_params);
/*Close connection with devices*/
clReleaseKernel(mtgp32_uint32_kernel);
clReleaseKernel(mtgp32_single_kernel);
clReleaseProgram(program);
clReleaseCommandQueue(commands);
clReleaseContext(context);
}
int main(int argc, char **argv)
{
uint64_t t1 = 0;
uint64_t t2 = 0;
int err;
cl_device_id device_id;
cl_command_queue commands;
cl_context context;
cl_mem output_buffer;
cl_mem input_buffer;
cl_mem partials_buffer;
size_t typesize;
int pass_count = 0;
size_t* group_counts = 0;
size_t* work_item_counts = 0;
int* operation_counts = 0;
int* entry_counts = 0;
int use_gpu = 1;
int i;
int c;
// Parse command line options
//
for( i = 0; i < argc && argv; i++)
{
if(!argv[i])
continue;
if(strstr(argv[i], "cpu"))
{
use_gpu = 0;
}
else if(strstr(argv[i], "gpu"))
{
use_gpu = 1;
}
else if(strstr(argv[i], "float2"))
{
integer = false;
channels = 2;
}
else if(strstr(argv[i], "float4"))
{
integer = false;
channels = 4;
}
else if(strstr(argv[i], "float"))
{
integer = false;
channels = 1;
}
else if(strstr(argv[i], "int2"))
{
integer = true;
channels = 2;
}
else if(strstr(argv[i], "int4"))
{
integer = true;
channels = 4;
}
else if(strstr(argv[i], "int"))
{
integer = true;
channels = 1;
}
}
// Create some random input data on the host
//
float *float_data = (float*)malloc(count * channels * sizeof(float));
int *integer_data = (int*)malloc(count * channels * sizeof(int));
for (i = 0; i < count * channels; i++)
{
float_data[i] = ((float) rand() / (float) RAND_MAX);
integer_data[i] = (int) (255.0f * float_data[i]);
}
// Connect to a compute device
//
err = clGetDeviceIDs(NULL, use_gpu ? CL_DEVICE_TYPE_GPU : CL_DEVICE_TYPE_CPU, 1, &device_id, NULL);
if (err != CL_SUCCESS)
{
printf("Error: Failed to locate a compute device!\n");
return EXIT_FAILURE;
}
size_t returned_size = 0;
size_t max_workgroup_size = 0;
err = clGetDeviceInfo(device_id, CL_DEVICE_MAX_WORK_GROUP_SIZE, sizeof(size_t), &max_workgroup_size, &returned_size);
if (err != CL_SUCCESS)
{
printf("Error: Failed to retrieve device info!\n");
return EXIT_FAILURE;
}
cl_char vendor_name[1024] = {0};
cl_char device_name[1024] = {0};
err = clGetDeviceInfo(device_id, CL_DEVICE_VENDOR, sizeof(vendor_name), vendor_name, &returned_size);
err|= clGetDeviceInfo(device_id, CL_DEVICE_NAME, sizeof(device_name), device_name, &returned_size);
if (err != CL_SUCCESS)
{
printf("Error: Failed to retrieve device info!\n");
return EXIT_FAILURE;
}
printf(SEPARATOR);
printf("Connecting to %s %s...\n", vendor_name, device_name);
// Load the compute program from disk into a cstring buffer
//
typesize = integer ? (sizeof(int)) : (sizeof(float));
const char* filename = 0;
switch(channels)
{
case 4:
filename = integer ? "reduce_int4_kernel.cl" : "reduce_float4_kernel.cl";
break;
case 2:
filename = integer ? "reduce_int2_kernel.cl" : "reduce_float2_kernel.cl";
break;
case 1:
filename = integer ? "reduce_int_kernel.cl" : "reduce_float_kernel.cl";
break;
default:
printf("Invalid channel count specified!\n");
return EXIT_FAILURE;
};
printf(SEPARATOR);
printf("Loading program '%s'...\n", filename);
printf(SEPARATOR);
char *source = load_program_source(filename);
if(!source)
{
printf("Error: Failed to load compute program from file!\n");
return EXIT_FAILURE;
}
// Create a compute context
//
context = clCreateContext(0, 1, &device_id, NULL, NULL, &err);
if (!context)
{
printf("Error: Failed to create a compute context!\n");
return EXIT_FAILURE;
}
// Create a command queue
//
commands = clCreateCommandQueue(context, device_id, 0, &err);
if (!commands)
{
printf("Error: Failed to create a command commands!\n");
return EXIT_FAILURE;
}
// Create the input buffer on the device
//
size_t buffer_size = typesize * count * channels;
input_buffer = clCreateBuffer(context, CL_MEM_READ_WRITE, buffer_size, NULL, NULL);
if (!input_buffer)
{
printf("Error: Failed to allocate input buffer on device!\n");
return EXIT_FAILURE;
}
// Fill the input buffer with the host allocated random data
//
void *input_data = (integer) ? (void*)integer_data : (void*)float_data;
err = clEnqueueWriteBuffer(commands, input_buffer, CL_TRUE, 0, buffer_size, input_data, 0, NULL, NULL);
if (err != CL_SUCCESS)
{
printf("Error: Failed to write to source array!\n");
return EXIT_FAILURE;
}
// Create an intermediate data buffer for intra-level results
//
partials_buffer = clCreateBuffer(context, CL_MEM_READ_WRITE, buffer_size, NULL, NULL);
if (!partials_buffer)
{
printf("Error: Failed to allocate partial sum buffer on device!\n");
return EXIT_FAILURE;
}
// Create the output buffer on the device
//
output_buffer = clCreateBuffer(context, CL_MEM_READ_WRITE, buffer_size, NULL, NULL);
if (!output_buffer)
{
printf("Error: Failed to allocate result buffer on device!\n");
return EXIT_FAILURE;
}
// Determine the reduction pass configuration for each level in the pyramid
//
create_reduction_pass_counts(
count, max_workgroup_size,
MAX_GROUPS, MAX_WORK_ITEMS,
&pass_count, &group_counts,
&work_item_counts, &operation_counts,
&entry_counts);
// Create specialized programs and kernels for each level of the reduction
//
cl_program *programs = (cl_program*)malloc(pass_count * sizeof(cl_program));
memset(programs, 0, pass_count * sizeof(cl_program));
cl_kernel *kernels = (cl_kernel*)malloc(pass_count * sizeof(cl_kernel));
memset(kernels, 0, pass_count * sizeof(cl_kernel));
for(i = 0; i < pass_count; i++)
{
char *block_source = malloc(strlen(source) + 1024);
size_t source_length = strlen(source) + 1024;
memset(block_source, 0, source_length);
// Insert macro definitions to specialize the kernel to a particular group size
//
const char group_size_macro[] = "#define GROUP_SIZE";
const char operations_macro[] = "#define OPERATIONS";
sprintf(block_source, "%s (%d) \n%s (%d)\n\n%s\n",
group_size_macro, (int)group_counts[i],
operations_macro, (int)operation_counts[i],
source);
// Create the compute program from the source buffer
//
programs[i] = clCreateProgramWithSource(context, 1, (const char **) & block_source, NULL, &err);
if (!programs[i] || err != CL_SUCCESS)
{
printf("%s\n", block_source);
printf("Error: Failed to create compute program!\n");
return EXIT_FAILURE;
}
// Build the program executable
//
err = clBuildProgram(programs[i], 0, NULL, NULL, NULL, NULL);
if (err != CL_SUCCESS)
{
size_t length;
char build_log[2048];
printf("%s\n", block_source);
printf("Error: Failed to build program executable!\n");
clGetProgramBuildInfo(programs[i], device_id, CL_PROGRAM_BUILD_LOG, sizeof(build_log), build_log, &length);
printf("%s\n", build_log);
return EXIT_FAILURE;
}
// Create the compute kernel from within the program
//
kernels[i] = clCreateKernel(programs[i], "reduce", &err);
if (!kernels[i] || err != CL_SUCCESS)
{
printf("Error: Failed to create compute kernel!\n");
return EXIT_FAILURE;
}
free(block_source);
}
// Do the reduction for each level
//
cl_mem pass_swap;
cl_mem pass_input = output_buffer;
cl_mem pass_output = input_buffer;
for(i = 0; i < pass_count; i++)
{
size_t global = group_counts[i] * work_item_counts[i];
size_t local = work_item_counts[i];
unsigned int operations = operation_counts[i];
unsigned int entries = entry_counts[i];
size_t shared_size = typesize * channels * local * operations;
printf("Pass[%4d] Global[%4d] Local[%4d] Groups[%4d] WorkItems[%4d] Operations[%d] Entries[%d]\n", i,
(int)global, (int)local, (int)group_counts[i], (int)work_item_counts[i], operations, entries);
// Swap the inputs and outputs for each pass
//
pass_swap = pass_input;
pass_input = pass_output;
pass_output = pass_swap;
err = CL_SUCCESS;
err |= clSetKernelArg(kernels[i], 0, sizeof(cl_mem), &pass_output);
err |= clSetKernelArg(kernels[i], 1, sizeof(cl_mem), &pass_input);
err |= clSetKernelArg(kernels[i], 2, shared_size, NULL);
err |= clSetKernelArg(kernels[i], 3, sizeof(int), &entries);
if (err != CL_SUCCESS)
{
printf("Error: Failed to set kernel arguments!\n");
return EXIT_FAILURE;
}
// After the first pass, use the partial sums for the next input values
//
if(pass_input == input_buffer)
pass_input = partials_buffer;
err = CL_SUCCESS;
err |= clEnqueueNDRangeKernel(commands, kernels[i], 1, NULL, &global, &local, 0, NULL, NULL);
if (err != CL_SUCCESS)
{
printf("Error: Failed to execute kernel!\n");
return EXIT_FAILURE;
}
}
err = clFinish(commands);
if (err != CL_SUCCESS)
{
printf("Error: Failed to wait for command queue to finish! %d\n", err);
return EXIT_FAILURE;
}
// Start the timing loop and execute the kernel over several iterations
//
printf(SEPARATOR);
printf("Timing %d iterations of reduction with %d elements of type %s%s...\n",
iterations, count, integer ? "int" : "float",
(channels <= 1) ? (" ") : (channels == 2) ? "2" : "4");
printf(SEPARATOR);
int k;
err = CL_SUCCESS;
t1 = current_time();
for (k = 0 ; k < iterations; k++)
{
for(i = 0; i < pass_count; i++)
{
size_t global = group_counts[i] * work_item_counts[i];
size_t local = work_item_counts[i];
err = clEnqueueNDRangeKernel(commands, kernels[i], 1, NULL, &global, &local, 0, NULL, NULL);
if (err != CL_SUCCESS)
{
printf("Error: Failed to execute kernel!\n");
return EXIT_FAILURE;
}
}
}
err = clFinish(commands);
if (err != CL_SUCCESS)
{
printf("Error: Failed to wait for command queue to finish! %d\n", err);
return EXIT_FAILURE;
}
t2 = current_time();
// Calculate the statistics for execution time and throughput
//
double t = subtract_time_in_seconds(t2, t1);
printf("Exec Time: %.2f ms\n", 1000.0 * t / (double)(iterations));
printf("Throughput: %.2f GB/sec\n", 1e-9 * buffer_size * iterations / t);
printf(SEPARATOR);
// Read back the results that were computed on the device
//
void *computed_result = malloc(typesize * channels);
memset(computed_result, 0, typesize * channels);
err = clEnqueueReadBuffer(commands, pass_output, CL_TRUE, 0, typesize * channels, computed_result, 0, NULL, NULL);
if (err)
{
printf("Error: Failed to read back results from the device!\n");
return EXIT_FAILURE;
}
// Verify the results are correct
//
if(integer)
{
int reference[4] = { 0, 0, 0, 0};
switch(channels)
{
case 4:
reduce_validate_int4(integer_data, count, reference);
break;
case 2:
reduce_validate_int2(integer_data, count, reference);
break;
case 1:
reduce_validate_int(integer_data, count, reference);
break;
default:
printf("Invalid channel count specified!\n");
return EXIT_FAILURE;
}
int result[4] = { 0.0f, 0.0f, 0.0f, 0.0f};
for(c = 0; c < channels; c++)
{
int v = ((int*) computed_result)[c];
result[c] += v;
}
float error = 0.0f;
float diff = 0.0f;
for(c = 0; c < channels; c++)
{
diff = fabs(reference[c] - result[c]);
error = diff > error ? diff : error;
}
if (error > MIN_ERROR)
{
for(c = 0; c < channels; c++)
printf("Result[%d] %d != %d\n", c, reference[c], result[c]);
printf("Error: Incorrect results obtained! Max error = %f\n", error);
return EXIT_FAILURE;
}
else
{
printf("Results Validated!\n");
printf(SEPARATOR);
}
}
else
{
float reference[4] = { 0.0f, 0.0f, 0.0f, 0.0f};
switch(channels)
{
case 4:
reduce_validate_float4(float_data, count, reference);
break;
case 2:
reduce_validate_float2(float_data, count, reference);
break;
case 1:
reduce_validate_float(float_data, count, reference);
break;
default:
printf("Invalid channel count specified!\n");
return EXIT_FAILURE;
}
float result[4] = { 0.0f, 0.0f, 0.0f, 0.0f};
for(c = 0; c < channels; c++)
{
float v = ((float*) computed_result)[c];
result[c] += v;
}
float error = 0.0f;
float diff = 0.0f;
for(c = 0; c < channels; c++)
{
diff = fabs(reference[c] - result[c]);
error = diff > error ? diff : error;
}
if (error > MIN_ERROR)
{
for(c = 0; c < channels; c++)
printf("Result[%d] %f != %f\n", c, reference[c], result[c]);
printf("Error: Incorrect results obtained! Max error = %f\n", error);
return EXIT_FAILURE;
}
else
{
printf("Results Validated!\n");
printf(SEPARATOR);
}
}
// Shutdown and cleanup
//
for(i = 0; i < pass_count; i++)
{
clReleaseKernel(kernels[i]);
clReleaseProgram(programs[i]);
}
clReleaseMemObject(input_buffer);
clReleaseMemObject(output_buffer);
clReleaseMemObject(partials_buffer);
clReleaseCommandQueue(commands);
clReleaseContext(context);
free(group_counts);
free(work_item_counts);
free(operation_counts);
free(entry_counts);
free(computed_result);
free(kernels);
free(float_data);
free(integer_data);
return 0;
}
int initGPU(int n)
{
#pragma mark Device Information
// Find the CPU CL device, as a fallback
err = clGetDeviceIDs(NULL, CL_DEVICE_TYPE_CPU, 1, &cpu, NULL);
assert(err == CL_SUCCESS);
// Find the GPU CL device, this is what we really want
// If there is no GPU device is CL capable, fall back to CPU
err |= clGetDeviceIDs(NULL, CL_DEVICE_TYPE_GPU, 1, &device, NULL);
if (err != CL_SUCCESS) device = cpu;
assert(device);
// Get some information about the returned device
cl_char vendor_name[1024] = {0};
cl_char device_name[1024] = {0};
err |= clGetDeviceInfo(device, CL_DEVICE_VENDOR, sizeof(vendor_name), vendor_name, &returned_size);
err |= clGetDeviceInfo(device, CL_DEVICE_NAME, sizeof(device_name), device_name, &returned_size);
assert(err == CL_SUCCESS);
printf("Connecting to %s %s...", vendor_name, device_name);
#pragma mark Context and Command Queue
// Now create a context to perform our calculation with the
// specified device
context = clCreateContext(0, 1, &device, NULL, NULL, &err);
assert(err == CL_SUCCESS);
// And also a command queue for the context
cmd_queue = clCreateCommandQueue(context, device, 0, NULL);
#pragma mark Program and Kernel Creation
// Load the program source from disk
// The kernel/program is the project directory and in Xcode the executable
// is set to launch from that directory hence we use a relative path
const char * filename = "kernel.cl";
char *program_source = load_program_source(filename);
program[0] = clCreateProgramWithSource(context, 1, (const char**)&program_source, NULL, &err);
assert(err == CL_SUCCESS);
err |= clBuildProgram(program[0], 0, NULL, NULL, NULL, NULL);
assert(err == CL_SUCCESS);
// Now create the kernel "objects" that we want to use in the example file
kernel[0] = clCreateKernel(program[0], "add", &err);
assert(err == CL_SUCCESS);
#pragma mark Memory Allocation
// Allocate memory on the device to hold our data and store the results into
buffer_size = sizeof(int) * n;
mem_c_position = clCreateBuffer(context, CL_MEM_READ_ONLY, buffer_size, NULL, &err);
mem_c_velocity = clCreateBuffer(context, CL_MEM_READ_ONLY, buffer_size, NULL, &err);
mem_p_angle = clCreateBuffer(context, CL_MEM_READ_ONLY, buffer_size, NULL, &err);
mem_p_velocity = clCreateBuffer(context, CL_MEM_READ_ONLY, buffer_size, NULL, &err);
assert(err == CL_SUCCESS);
mem_fitness = clCreateBuffer(context, CL_MEM_WRITE_ONLY, buffer_size, NULL, &err);
assert(err == CL_SUCCESS);
// Get all of the stuff written and allocated
clFinish(cmd_queue);
printf(" done\n");
return err; // CL_SUCCESS
}