github-mirrors/vortex
1
0
Fork 0
mirror of https://github.com/vortexgpgpu/vortex.git synced 2025-04-23 13:27:29 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

Add files via upload

Browse source
This commit is contained in:
Lyons, Ethan Tyler 2019-11-25 01:49:17 -05:00 • committed by GitHub Enterprise
parent 1ff385daaf
commit a38987e7af
9 changed files with 1694 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

101
benchmarks/opencl/BlackScholes/BlackScholes.cl Normal file
View file

@ -0,0 +1,101 @@
/*
* Copyright 1993-2010 NVIDIA Corporation. All rights reserved.
*
* Please refer to the NVIDIA end user license agreement (EULA) associated
* with this source code for terms and conditions that govern your use of
* this software. Any use, reproduction, disclosure, or distribution of
* this software and related documentation outside the terms of the EULA
* is strictly prohibited.
*
*/
#if(0)
#define EXP(a) native_exp(a)
#define LOG(a) native_log(a)
#define SQRT(a) native_sqrt(a)
#else
#define EXP(a) exp(a)
#define LOG(a) log(a)
#define SQRT(a) sqrt(a)
#endif
///////////////////////////////////////////////////////////////////////////////
// Predefine functions to avoid bug in OpenCL compiler on Mac OSX 10.7 systems
///////////////////////////////////////////////////////////////////////////////
float CND(float d);
void BlackScholesBody(__global float *call, __global float *put, float S,
float X, float T, float R, float V);
///////////////////////////////////////////////////////////////////////////////
// Rational approximation of cumulative normal distribution function
///////////////////////////////////////////////////////////////////////////////
float CND(float d){
const float A1 = 0.31938153f;
const float A2 = -0.356563782f;
const float A3 = 1.781477937f;
const float A4 = -1.821255978f;
const float A5 = 1.330274429f;
const float RSQRT2PI = 0.39894228040143267793994605993438f;
float
K = 1.0f / (1.0f + 0.2316419f * fabs(d));
float
cnd = RSQRT2PI * EXP(- 0.5f * d * d) *
(K * (A1 + K * (A2 + K * (A3 + K * (A4 + K * A5)))));
if(d > 0)
cnd = 1.0f - cnd;
return cnd;
}
///////////////////////////////////////////////////////////////////////////////
// Black-Scholes formula for both call and put
///////////////////////////////////////////////////////////////////////////////
void BlackScholesBody(
__global float *call, //Call option price
__global float *put, //Put option price
float S, //Current stock price
float X, //Option strike price
float T, //Option years
float R, //Riskless rate of return
float V //Stock volatility
){
float sqrtT = SQRT(T);
float d1 = (LOG(S / X) + (R + 0.5f * V * V) * T) / (V * sqrtT);
float d2 = d1 - V * sqrtT;
float CNDD1 = CND(d1);
float CNDD2 = CND(d2);
//Calculate Call and Put simultaneously
float expRT = EXP(- R * T);
*call = (S * CNDD1 - X * expRT * CNDD2);
*put = (X * expRT * (1.0f - CNDD2) - S * (1.0f - CNDD1));
}
__kernel void BlackScholes(
__global float *d_Call, //Call option price
__global float *d_Put, //Put option price
__global float *d_S, //Current stock price
__global float *d_X, //Option strike price
__global float *d_T, //Option years
float R, //Riskless rate of return
float V, //Stock volatility
unsigned int optN
){
for(unsigned int opt = get_global_id(0); opt < optN; opt += get_global_size(0))
BlackScholesBody(
&d_Call[opt],
&d_Put[opt],
d_S[opt],
d_X[opt],
d_T[opt],
R,
V
);
}

248
benchmarks/opencl/BlackScholes/main.cpp Normal file
View file

@ -0,0 +1,248 @@
/*
* Copyright 1993-2010 NVIDIA Corporation. All rights reserved.
*
* Please refer to the NVIDIA end user license agreement (EULA) associated
* with this source code for terms and conditions that govern your use of
* this software. Any use, reproduction, disclosure, or distribution of
* this software and related documentation outside the terms of the EULA
* is strictly prohibited.
*
*/
// standard utilities and systems includes
#include <oclUtils.h>
#include <shrQATest.h>
#include "oclBlackScholes_common.h"
////////////////////////////////////////////////////////////////////////////////
// Helper functions
////////////////////////////////////////////////////////////////////////////////
double executionTime(cl_event &event){
cl_ulong start, end;
clGetEventProfilingInfo(event, CL_PROFILING_COMMAND_END, sizeof(cl_ulong), &end, NULL);
clGetEventProfilingInfo(event, CL_PROFILING_COMMAND_START, sizeof(cl_ulong), &start, NULL);
return (double)1.0e-9 * (end - start); // convert nanoseconds to seconds on return
}
////////////////////////////////////////////////////////////////////////////////
// Random float helper
////////////////////////////////////////////////////////////////////////////////
float randFloat(float low, float high){
float t = (float)rand() / (float)RAND_MAX;
return (1.0f - t) * low + t * high;
}
////////////////////////////////////////////////////////////////////////////////
// Main program
////////////////////////////////////////////////////////////////////////////////
int main(int argc, char **argv)
{
cl_platform_id cpPlatform; //OpenCL platform
cl_device_id* cdDevices = NULL; //OpenCL devices list (array)
cl_context cxGPUContext; //OpenCL context
cl_command_queue cqCommandQueue; //OpenCL command que
cl_mem //OpenCL memory buffer objects
d_Call,
d_Put,
d_S,
d_X,
d_T;
cl_int ciErrNum;
float
*h_CallCPU,
*h_PutCPU,
*h_CallGPU,
*h_PutGPU,
*h_S,
*h_X,
*h_T;
const unsigned int optionCount = 4000000;
const float R = 0.02f;
const float V = 0.30f;
shrQAStart(argc, argv);
// Get the NVIDIA platform
ciErrNum = oclGetPlatformID(&cpPlatform);
//oclCheckErrorEX(ciErrNum, CL_SUCCESS, NULL);
shrLog("clGetPlatformID...\n");
//Get all the devices
cl_uint uiNumDevices = 0; // Number of devices available
cl_uint uiTargetDevice = 0; // Default Device to compute on
cl_uint uiNumComputeUnits; // Number of compute units (SM's on NV GPU)
shrLog("Get the Device info and select Device...\n");
ciErrNum = clGetDeviceIDs(cpPlatform, CL_DEVICE_TYPE_DEFAULT, 0, NULL, &uiNumDevices);
//oclCheckErrorEX(ciErrNum, CL_SUCCESS, NULL);
cdDevices = (cl_device_id *)malloc(uiNumDevices * sizeof(cl_device_id) );
ciErrNum = clGetDeviceIDs(cpPlatform, CL_DEVICE_TYPE_DEFAULT, uiNumDevices, cdDevices, NULL);
//oclCheckErrorEX(ciErrNum, CL_SUCCESS, NULL);
// Get command line device options and config accordingly
shrLog(" # of Devices Available = %u\n", uiNumDevices);
if(shrGetCmdLineArgumentu(argc, (const char**)argv, "device", &uiTargetDevice)== shrTRUE)
{
uiTargetDevice = CLAMP(uiTargetDevice, 0, (uiNumDevices - 1));
}
shrLog(" Using Device %u: ", uiTargetDevice);
oclPrintDevName(LOGBOTH, cdDevices[uiTargetDevice]);
ciErrNum = clGetDeviceInfo(cdDevices[uiTargetDevice], CL_DEVICE_MAX_COMPUTE_UNITS, sizeof(uiNumComputeUnits), &uiNumComputeUnits, NULL);
//oclCheckErrorEX(ciErrNum, CL_SUCCESS, NULL);
shrLog("\n # of Compute Units = %u\n", uiNumComputeUnits);
// set logfile name and start logs
shrSetLogFileName ("oclBlackScholes.txt");
shrLog("%s Starting...\n\n", argv[0]);
shrLog("Allocating and initializing host memory...\n");
h_CallCPU = (float *)malloc(optionCount * sizeof(float));
h_PutCPU = (float *)malloc(optionCount * sizeof(float));
h_CallGPU = (float *)malloc(optionCount * sizeof(float));
h_PutGPU = (float *)malloc(optionCount * sizeof(float));
h_S = (float *)malloc(optionCount * sizeof(float));
h_X = (float *)malloc(optionCount * sizeof(float));
h_T = (float *)malloc(optionCount * sizeof(float));
srand(2009);
for(unsigned int i = 0; i < optionCount; i++){
h_CallCPU[i] = -1.0f;
h_PutCPU[i] = -1.0f;
h_S[i] = randFloat(5.0f, 30.0f);
h_X[i] = randFloat(1.0f, 100.0f);
h_T[i] = randFloat(0.25f, 10.0f);
}
shrLog("Initializing OpenCL...\n");
// Get the NVIDIA platform
ciErrNum = oclGetPlatformID(&cpPlatform);
//oclCheckError(ciErrNum, CL_SUCCESS);
// Get a GPU device
ciErrNum = clGetDeviceIDs(cpPlatform, CL_DEVICE_TYPE_DEFAULT, 1, &cdDevices[uiTargetDevice], NULL);
//oclCheckError(ciErrNum, CL_SUCCESS);
// Create the context
cxGPUContext = clCreateContext(0, 1, &cdDevices[uiTargetDevice], NULL, NULL, &ciErrNum);
//oclCheckError(ciErrNum, CL_SUCCESS);
//Create a command-queue
cqCommandQueue = clCreateCommandQueue(cxGPUContext, cdDevices[uiTargetDevice], CL_QUEUE_PROFILING_ENABLE, &ciErrNum);
//oclCheckError(ciErrNum, CL_SUCCESS);
shrLog("Creating OpenCL memory objects...\n");
d_Call = clCreateBuffer(cxGPUContext, CL_MEM_READ_WRITE, optionCount * sizeof(float), NULL, &ciErrNum);
//oclCheckError(ciErrNum, CL_SUCCESS);
d_Put = clCreateBuffer(cxGPUContext, CL_MEM_READ_WRITE, optionCount * sizeof(float), NULL, &ciErrNum);
//oclCheckError(ciErrNum, CL_SUCCESS);
d_S = clCreateBuffer(cxGPUContext, CL_MEM_READ_WRITE | CL_MEM_COPY_HOST_PTR, optionCount * sizeof(float), h_S, &ciErrNum);
//oclCheckError(ciErrNum, CL_SUCCESS);
d_X = clCreateBuffer(cxGPUContext, CL_MEM_READ_WRITE | CL_MEM_COPY_HOST_PTR, optionCount * sizeof(float), h_X, &ciErrNum);
//oclCheckError(ciErrNum, CL_SUCCESS);
d_T = clCreateBuffer(cxGPUContext, CL_MEM_READ_WRITE | CL_MEM_COPY_HOST_PTR, optionCount * sizeof(float), h_T, &ciErrNum);
//oclCheckError(ciErrNum, CL_SUCCESS);
shrLog("Starting up BlackScholes...\n");
initBlackScholes(cxGPUContext, cqCommandQueue, (const char **)argv);
shrLog("Running OpenCL BlackScholes...\n\n");
//Just a single run or a warmup iteration
BlackScholes(
NULL,
d_Call,
d_Put,
d_S,
d_X,
d_T,
R,
V,
optionCount
);
#ifdef GPU_PROFILING
const int numIterations = 16;
cl_event startMark, endMark;
ciErrNum = clEnqueueMarker(cqCommandQueue, &startMark);
ciErrNum |= clFinish(cqCommandQueue);
shrCheckError(ciErrNum, CL_SUCCESS);
shrDeltaT(0);
for(int i = 0; i < numIterations; i++){
BlackScholes(
cqCommandQueue,
d_Call,
d_Put,
d_S,
d_X,
d_T,
R,
V,
optionCount
);
}
ciErrNum = clEnqueueMarker(cqCommandQueue, &endMark);
ciErrNum |= clFinish(cqCommandQueue);
shrCheckError(ciErrNum, CL_SUCCESS);
//Calculate performance metrics by wallclock time
double gpuTime = shrDeltaT(0) / numIterations;
shrLogEx(LOGBOTH | MASTER, 0, "oclBlackScholes, Throughput = %.4f GOptions/s, Time = %.5f s, Size = %u options, NumDevsUsed = %i, Workgroup = %u\n",
(double)(2.0 * optionCount * 1.0e-9)/gpuTime, gpuTime, (2 * optionCount), 1, 0);
//Get profiling info
cl_ulong startTime = 0, endTime = 0;
ciErrNum = clGetEventProfilingInfo(startMark, CL_PROFILING_COMMAND_END, sizeof(cl_ulong), &startTime, NULL);
ciErrNum |= clGetEventProfilingInfo(endMark, CL_PROFILING_COMMAND_END, sizeof(cl_ulong), &endTime, NULL);
shrCheckError(ciErrNum, CL_SUCCESS);
shrLog("\nOpenCL time: %.5f s\n\n", 1.0e-9 * ((double)endTime - (double)startTime) / (double)numIterations);
#endif
shrLog("\nReading back OpenCL BlackScholes results...\n");
ciErrNum = clEnqueueReadBuffer(cqCommandQueue, d_Call, CL_TRUE, 0, optionCount * sizeof(float), h_CallGPU, 0, NULL, NULL);
//oclCheckError(ciErrNum, CL_SUCCESS);
ciErrNum = clEnqueueReadBuffer(cqCommandQueue, d_Put, CL_TRUE, 0, optionCount * sizeof(float), h_PutGPU, 0, NULL, NULL);
//oclCheckError(ciErrNum, CL_SUCCESS);
shrLog("Comparing against Host/C++ computation...\n");
BlackScholesCPU(h_CallCPU, h_PutCPU, h_S, h_X, h_T, R, V, optionCount);
double deltaCall = 0, deltaPut = 0, sumCall = 0, sumPut = 0;
double L1call, L1put;
for(unsigned int i = 0; i < optionCount; i++)
{
sumCall += fabs(h_CallCPU[i]);
sumPut += fabs(h_PutCPU[i]);
deltaCall += fabs(h_CallCPU[i] - h_CallGPU[i]);
deltaPut += fabs(h_PutCPU[i] - h_PutGPU[i]);
}
L1call = deltaCall / sumCall;
L1put = deltaPut / sumPut;
shrLog("Relative L1 (call, put) = (%.3e, %.3e)\n\n", L1call, L1put);
shrLog("Shutting down...\n");
closeBlackScholes();
ciErrNum = clReleaseMemObject(d_T);
ciErrNum |= clReleaseMemObject(d_X);
ciErrNum |= clReleaseMemObject(d_S);
ciErrNum |= clReleaseMemObject(d_Put);
ciErrNum |= clReleaseMemObject(d_Call);
ciErrNum |= clReleaseCommandQueue(cqCommandQueue);
ciErrNum |= clReleaseContext(cxGPUContext);
//oclCheckError(ciErrNum, CL_SUCCESS);
free(h_T);
free(h_X);
free(h_S);
free(h_PutGPU);
free(h_CallGPU);
free(h_PutCPU);
free(h_CallCPU);
if(cdDevices)free(cdDevices);
shrQAFinishExit(argc, (const char **)argv, ((L1call < 1E-6) && (L1put < 1E-6)) ? QA_PASSED : QA_FAILED );
}

BIN
benchmarks/opencl/BlackScholes/oclBlackScholes.pdf Normal file
View file

Binary file not shown.

50
benchmarks/opencl/BlackScholes/oclBlackScholes_common.h Normal file
View file

@ -0,0 +1,50 @@
/*
* Copyright 1993-2010 NVIDIA Corporation. All rights reserved.
*
* Please refer to the NVIDIA end user license agreement (EULA) associated
* with this source code for terms and conditions that govern your use of
* this software. Any use, reproduction, disclosure, or distribution of
* this software and related documentation outside the terms of the EULA
* is strictly prohibited.
*
*/
#include <oclUtils.h>
////////////////////////////////////////////////////////////////////////////////
// Process an array of optN options on CPU
////////////////////////////////////////////////////////////////////////////////
extern "C" void BlackScholesCPU(
float *h_Call, //Call option price
float *h_Put, //Put option price
float *h_S, //Current stock price
float *h_X, //Option strike price
float *h_T, //Option years
float R, //Riskless rate of return
float V, //Stock volatility
unsigned int optionCount
);
////////////////////////////////////////////////////////////////////////////////
// OpenCL Black-Scholes kernel launcher
////////////////////////////////////////////////////////////////////////////////
extern "C" void initBlackScholes(cl_context cxGPUContext, cl_command_queue cqParamCommandQue, const char **argv);
extern "C" void closeBlackScholes(void);
extern "C" void BlackScholes(
cl_command_queue cqCommandQueue,
cl_mem d_Call, //Call option price
cl_mem d_Put, //Put option price
cl_mem d_S, //Current stock price
cl_mem d_X, //Option strike price
cl_mem d_T, //Option years
cl_float R, //Riskless rate of return
cl_float V, //Stock volatility
cl_uint optionCount
);

92
benchmarks/opencl/BlackScholes/oclBlackScholes_gold.cpp Normal file
View file

@ -0,0 +1,92 @@
/*
* Copyright 1993-2010 NVIDIA Corporation. All rights reserved.
*
* Please refer to the NVIDIA end user license agreement (EULA) associated
* with this source code for terms and conditions that govern your use of
* this software. Any use, reproduction, disclosure, or distribution of
* this software and related documentation outside the terms of the EULA
* is strictly prohibited.
*
*/
#include <math.h>
#include "oclBlackScholes_common.h"
///////////////////////////////////////////////////////////////////////////////
// Rational approximation of cumulative normal distribution function
///////////////////////////////////////////////////////////////////////////////
static double CND(double d){
const double A1 = 0.31938153;
const double A2 = -0.356563782;
const double A3 = 1.781477937;
const double A4 = -1.821255978;
const double A5 = 1.330274429;
const double RSQRT2PI = 0.39894228040143267793994605993438;
double
K = 1.0 / (1.0 + 0.2316419 * fabs(d));
double
cnd = RSQRT2PI * exp(- 0.5 * d * d) *
(K * (A1 + K * (A2 + K * (A3 + K * (A4 + K * A5)))));
if(d > 0)
cnd = 1.0 - cnd;
return cnd;
}
///////////////////////////////////////////////////////////////////////////////
// Black-Scholes formula for both call and put
///////////////////////////////////////////////////////////////////////////////
static void BlackScholesBodyCPU(
float& call, //Call option price
float& put, //Put option price
float Sf, //Current stock price
float Xf, //Option strike price
float Tf, //Option years
float Rf, //Riskless rate of return
float Vf //Stock volatility
){
double S = Sf, X = Xf, T = Tf, R = Rf, V = Vf;
double sqrtT = sqrt(T);
double d1 = (log(S / X) + (R + 0.5 * V * V) * T) / (V * sqrtT);
double d2 = d1 - V * sqrtT;
double CNDD1 = CND(d1);
double CNDD2 = CND(d2);
//Calculate Call and Put simultaneously
double expRT = exp(- R * T);
call = (float)(S * CNDD1 - X * expRT * CNDD2);
put = (float)(X * expRT * (1.0 - CNDD2) - S * (1.0 - CNDD1));
}
////////////////////////////////////////////////////////////////////////////////
// Process an array of optN options
////////////////////////////////////////////////////////////////////////////////
extern "C" void BlackScholesCPU(
float *h_Call, //Call option price
float *h_Put, //Put option price
float *h_S, //Current stock price
float *h_X, //Option strike price
float *h_T, //Option years
float R, //Riskless rate of return
float V, //Stock volatility
unsigned int optionCount
){
for(unsigned int i = 0; i < optionCount; i++)
BlackScholesBodyCPU(
h_Call[i],
h_Put[i],
h_S[i],
h_X[i],
h_T[i],
R,
V
);
}

125
benchmarks/opencl/BlackScholes/oclBlackScholes_launcher.cpp Normal file
View file

@ -0,0 +1,125 @@
/*
* Copyright 1993-2010 NVIDIA Corporation. All rights reserved.
*
* Please refer to the NVIDIA end user license agreement (EULA) associated
* with this source code for terms and conditions that govern your use of
* this software. Any use, reproduction, disclosure, or distribution of
* this software and related documentation outside the terms of the EULA
* is strictly prohibited.
*
*/
#include <oclUtils.h>
#include "oclBlackScholes_common.h"
static cl_program cpBlackScholes; //OpenCL program
static cl_kernel ckBlackScholes; //OpenCL kernel
static cl_command_queue cqDefaultCommandQueue;
extern "C" void initBlackScholes(cl_context cxGPUContext, cl_command_queue cqParamCommandQueue, const char **argv){
cl_int ciErrNum;
size_t kernelLength;
shrLog("...loading BlackScholes.cl\n");
char *cPathAndName = shrFindFilePath("BlackScholes.cl", argv[0]);
shrCheckError(cPathAndName != NULL, shrTRUE);
char *cBlackScholes = oclLoadProgSource(cPathAndName, "// My comment\n", &kernelLength);
shrCheckError(cBlackScholes != NULL, shrTRUE);
shrLog("...creating BlackScholes program\n");
//cpBlackScholes = clCreateProgramWithSource(cxGPUContext, 1, (const char **)&cBlackScholes, &kernelLength, &ciErrNum);
cpBlackScholes = clCreateProgramWithBuiltInKernels(context, 1, &device_id, "BlackScholes", NULL);
shrCheckError(ciErrNum, CL_SUCCESS);
shrLog("...building BlackScholes program\n");
ciErrNum = clBuildProgram(cpBlackScholes, 0, NULL, "-cl-fast-relaxed-math -Werror", NULL, NULL);
if(ciErrNum != CL_BUILD_SUCCESS){
shrLog("*** Compilation failure ***\n");
size_t deviceNum;
cl_device_id *cdDevices;
ciErrNum = clGetContextInfo(cxGPUContext, CL_CONTEXT_DEVICES, 0, NULL, &deviceNum);
shrCheckError(ciErrNum, CL_SUCCESS);
cdDevices = (cl_device_id *)malloc(deviceNum * sizeof(cl_device_id));
shrCheckError(cdDevices != NULL, shrTRUE);
ciErrNum = clGetContextInfo(cxGPUContext, CL_CONTEXT_DEVICES, deviceNum * sizeof(cl_device_id), cdDevices, NULL);
shrCheckError(ciErrNum, CL_SUCCESS);
size_t logSize;
char *logTxt;
ciErrNum = clGetProgramBuildInfo(cpBlackScholes, cdDevices[0], CL_PROGRAM_BUILD_LOG, 0, NULL, &logSize);
shrCheckError(ciErrNum, CL_SUCCESS);
logTxt = (char *)malloc(logSize);
shrCheckError(logTxt != NULL, shrTRUE);
ciErrNum = clGetProgramBuildInfo(cpBlackScholes, cdDevices[0], CL_PROGRAM_BUILD_LOG, logSize, logTxt, NULL);
shrCheckError(ciErrNum, CL_SUCCESS);
shrLog("%s\n", logTxt);
shrLog("*** Exiting ***\n");
free(logTxt);
free(cdDevices);
exit(666);
}
//Save ptx code to separate file
oclLogPtx(cpBlackScholes, oclGetFirstDev(cxGPUContext), "BlackScholes.ptx");
shrLog("...creating BlackScholes kernels\n");
ckBlackScholes = clCreateKernel(cpBlackScholes, "BlackScholes", &ciErrNum);
shrCheckError(ciErrNum, CL_SUCCESS);
cqDefaultCommandQueue = cqParamCommandQueue;
free(cBlackScholes);
free(cPathAndName);
}
extern "C" void closeBlackScholes(void){
cl_int ciErrNum;
ciErrNum = clReleaseKernel(ckBlackScholes);
ciErrNum |= clReleaseProgram(cpBlackScholes);
shrCheckError(ciErrNum, CL_SUCCESS);
}
////////////////////////////////////////////////////////////////////////////////
// OpenCL Black-Scholes kernel launcher
////////////////////////////////////////////////////////////////////////////////
extern "C" void BlackScholes(
cl_command_queue cqCommandQueue,
cl_mem d_Call, //Call option price
cl_mem d_Put, //Put option price
cl_mem d_S, //Current stock price
cl_mem d_X, //Option strike price
cl_mem d_T, //Option years
cl_float R, //Riskless rate of return
cl_float V, //Stock volatility
cl_uint optionCount
){
cl_int ciErrNum;
if(!cqCommandQueue)
cqCommandQueue = cqDefaultCommandQueue;
ciErrNum = clSetKernelArg(ckBlackScholes, 0, sizeof(cl_mem), (void *)&d_Call);
ciErrNum |= clSetKernelArg(ckBlackScholes, 1, sizeof(cl_mem), (void *)&d_Put);
ciErrNum |= clSetKernelArg(ckBlackScholes, 2, sizeof(cl_mem), (void *)&d_S);
ciErrNum |= clSetKernelArg(ckBlackScholes, 3, sizeof(cl_mem), (void *)&d_X);
ciErrNum |= clSetKernelArg(ckBlackScholes, 4, sizeof(cl_mem), (void *)&d_T);
ciErrNum |= clSetKernelArg(ckBlackScholes, 5, sizeof(cl_float), (void *)&R);
ciErrNum |= clSetKernelArg(ckBlackScholes, 6, sizeof(cl_float), (void *)&V);
ciErrNum |= clSetKernelArg(ckBlackScholes, 7, sizeof(cl_uint), (void *)&optionCount);
shrCheckError(ciErrNum, CL_SUCCESS);
//Run the kernel
size_t globalWorkSize = 60 * 1024;
size_t localWorkSize = 128;
ciErrNum = clEnqueueNDRangeKernel(cqCommandQueue, ckBlackScholes, 1, NULL, &globalWorkSize, &localWorkSize, 0, NULL, NULL);
shrCheckError(ciErrNum, CL_SUCCESS);
}

198
benchmarks/opencl/BlackScholes/oclUtils.h Normal file
View file

@ -0,0 +1,198 @@
/*
* Copyright 1993-2010 NVIDIA Corporation. All rights reserved.
*
* Please refer to the NVIDIA end user license agreement (EULA) associated
* with this source code for terms and conditions that govern your use of
* this software. Any use, reproduction, disclosure, or distribution of
* this software and related documentation outside the terms of the EULA
* is strictly prohibited.
*
*/
#ifndef OCL_UTILS_H
#define OCL_UTILS_H
// *********************************************************************
// Utilities specific to OpenCL samples in NVIDIA GPU Computing SDK
// *********************************************************************
// Common headers: Cross-API utililties and OpenCL header
#include <shrUtils.h>
// All OpenCL headers
#if defined (__APPLE__) || defined(MACOSX)
#include <OpenCL/opencl.h>
#else
#include <CL/opencl.h>
#endif
// Includes
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
// For systems with CL_EXT that are not updated with these extensions, we copied these
// extensions from <CL/cl_ext.h>
#ifndef CL_DEVICE_COMPUTE_CAPABILITY_MAJOR_NV
/* cl_nv_device_attribute_query extension - no extension #define since it has no functions */
#define CL_DEVICE_COMPUTE_CAPABILITY_MAJOR_NV 0x4000
#define CL_DEVICE_COMPUTE_CAPABILITY_MINOR_NV 0x4001
#define CL_DEVICE_REGISTERS_PER_BLOCK_NV 0x4002
#define CL_DEVICE_WARP_SIZE_NV 0x4003
#define CL_DEVICE_GPU_OVERLAP_NV 0x4004
#define CL_DEVICE_KERNEL_EXEC_TIMEOUT_NV 0x4005
#define CL_DEVICE_INTEGRATED_MEMORY_NV 0x4006
#endif
// reminders for build output window and log
#ifdef _WIN32
#pragma message ("Note: including shrUtils.h")
#pragma message ("Note: including opencl.h")
#endif
// SDK Revision #
#define OCL_SDKREVISION "7027912"
// Error and Exit Handling Macros...
// *********************************************************************
// Full error handling macro with Cleanup() callback (if supplied)...
// (Companion Inline Function lower on page)
#define oclCheckErrorEX(a, b, c) __oclCheckErrorEX(a, b, c, __FILE__ , __LINE__)
// Short version without Cleanup() callback pointer
// Both Input (a) and Reference (b) are specified as args
#define oclCheckError(a, b) oclCheckErrorEX(a, b, 0)
//////////////////////////////////////////////////////////////////////////////
//! Gets the platform ID for NVIDIA if available, otherwise default to platform 0
//!
//! @return the id
//! @param clSelectedPlatformID OpenCL platform ID
//////////////////////////////////////////////////////////////////////////////
extern "C" cl_int oclGetPlatformID(cl_platform_id* clSelectedPlatformID);
//////////////////////////////////////////////////////////////////////////////
//! Print info about the device
//!
//! @param iLogMode enum LOGBOTH, LOGCONSOLE, LOGFILE
//! @param device OpenCL id of the device
//////////////////////////////////////////////////////////////////////////////
extern "C" void oclPrintDevInfo(int iLogMode, cl_device_id device);
//////////////////////////////////////////////////////////////////////////////
//! Get and return device capability
//!
//! @return the 2 digit integer representation of device Cap (major minor). return -1 if NA
//! @param device OpenCL id of the device
//////////////////////////////////////////////////////////////////////////////
extern "C" int oclGetDevCap(cl_device_id device);
//////////////////////////////////////////////////////////////////////////////
//! Print the device name
//!
//! @param iLogMode enum LOGBOTH, LOGCONSOLE, LOGFILE
//! @param device OpenCL id of the device
//////////////////////////////////////////////////////////////////////////////
extern "C" void oclPrintDevName(int iLogMode, cl_device_id device);
//////////////////////////////////////////////////////////////////////////////
//! Gets the id of the first device from the context
//!
//! @return the id
//! @param cxGPUContext OpenCL context
//////////////////////////////////////////////////////////////////////////////
extern "C" cl_device_id oclGetFirstDev(cl_context cxGPUContext);
//////////////////////////////////////////////////////////////////////////////
//! Gets the id of the nth device from the context
//!
//! @return the id or -1 when out of range
//! @param cxGPUContext OpenCL context
//! @param device_idx index of the device of interest
//////////////////////////////////////////////////////////////////////////////
extern "C" cl_device_id oclGetDev(cl_context cxGPUContext, unsigned int device_idx);
//////////////////////////////////////////////////////////////////////////////
//! Gets the id of device with maximal FLOPS from the context
//!
//! @return the id
//! @param cxGPUContext OpenCL context
//////////////////////////////////////////////////////////////////////////////
extern "C" cl_device_id oclGetMaxFlopsDev(cl_context cxGPUContext);
//////////////////////////////////////////////////////////////////////////////
//! Loads a Program file and prepends the cPreamble to the code.
//!
//! @return the source string if succeeded, 0 otherwise
//! @param cFilename program filename
//! @param cPreamble code that is prepended to the loaded file, typically a set of #defines or a header
//! @param szFinalLength returned length of the code string
//////////////////////////////////////////////////////////////////////////////
extern "C" char* oclLoadProgSource(const char* cFilename, const char* cPreamble, size_t* szFinalLength);
//////////////////////////////////////////////////////////////////////////////
//! Get the binary (PTX) of the program associated with the device
//!
//! @param cpProgram OpenCL program
//! @param cdDevice device of interest
//! @param binary returned code
//! @param length length of returned code
//////////////////////////////////////////////////////////////////////////////
extern "C" void oclGetProgBinary( cl_program cpProgram, cl_device_id cdDevice, char** binary, size_t* length);
//////////////////////////////////////////////////////////////////////////////
//! Get and log the binary (PTX) from the OpenCL compiler for the requested program & device
//!
//! @param cpProgram OpenCL program
//! @param cdDevice device of interest
//! @param const char* cPtxFileName optional PTX file name
//////////////////////////////////////////////////////////////////////////////
extern "C" void oclLogPtx(cl_program cpProgram, cl_device_id cdDevice, const char* cPtxFileName);
//////////////////////////////////////////////////////////////////////////////
//! Get and log the Build Log from the OpenCL compiler for the requested program & device
//!
//! @param cpProgram OpenCL program
//! @param cdDevice device of interest
//////////////////////////////////////////////////////////////////////////////
extern "C" void oclLogBuildInfo(cl_program cpProgram, cl_device_id cdDevice);
// Helper function for De-allocating cl objects
// *********************************************************************
extern "C" void oclDeleteMemObjs(cl_mem* cmMemObjs, int iNumObjs);
// Helper function to get OpenCL error string from constant
// *********************************************************************
extern "C" const char* oclErrorString(cl_int error);
// Helper function to get OpenCL image format string (channel order and type) from constant
// *********************************************************************
extern "C" const char* oclImageFormatString(cl_uint uiImageFormat);
// companion inline function for error checking and exit on error WITH Cleanup Callback (if supplied)
// *********************************************************************
inline void __oclCheckErrorEX(cl_int iSample, cl_int iReference, void (*pCleanup)(int), const char* cFile, const int iLine)
{
// An error condition is defined by the sample/test value not equal to the reference
if (iReference != iSample)
{
// If the sample/test value isn't equal to the ref, it's an error by defnition, so override 0 sample/test value
iSample = (iSample == 0) ? -9999 : iSample;
// Log the error info
shrLog("\n !!! Error # %i (%s) at line %i , in file %s !!!\n\n", iSample, oclErrorString(iSample), iLine, cFile);
// Cleanup and exit, or just exit if no cleanup function pointer provided. Use iSample (error code in this case) as process exit code.
if (pCleanup != NULL)
{
pCleanup(iSample);
}
else
{
shrLogEx(LOGBOTH | CLOSELOG, 0, "Exiting...\n");
exit(iSample);
}
}
}
#endif

238
benchmarks/opencl/BlackScholes/shrQATest.h Normal file
View file

@ -0,0 +1,238 @@
/*
* Copyright 1993-2010 NVIDIA Corporation. All rights reserved.
*
* Please refer to the NVIDIA end user license agreement (EULA) associated
* with this source code for terms and conditions that govern your use of
* this software. Any use, reproduction, disclosure, or distribution of
* this software and related documentation outside the terms of the EULA
* is strictly prohibited.
*
*/
#ifndef SHR_QATEST_H
#define SHR_QATEST_H
// *********************************************************************
// Generic utilities for NVIDIA GPU Computing SDK
// *********************************************************************
// OS dependent includes
#ifdef _WIN32
#pragma message ("Note: including windows.h")
#pragma message ("Note: including math.h")
#pragma message ("Note: including assert.h")
#pragma message ("Note: including time.h")
// Headers needed for Windows
#include <windows.h>
#include <time.h>
#else
// Headers needed for Linux
#include <sys/stat.h>
#include <sys/types.h>
#include <sys/time.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <stdarg.h>
#include <unistd.h>
#include <time.h>
#endif
#ifndef STRCASECMP
#ifdef _WIN32
#define STRCASECMP _stricmp
#else
#define STRCASECMP strcasecmp
#endif
#endif
#ifndef STRNCASECMP
#ifdef _WIN32
#define STRNCASECMP _strnicmp
#else
#define STRNCASECMP strncasecmp
#endif
#endif
// Standardized QA Start/Finish for CUDA SDK tests
#define shrQAStart(a, b) __shrQAStart(a, b)
#define shrQAFinish(a, b, c) __shrQAFinish(a, b, c)
#define shrQAFinish2(a, b, c, d) __shrQAFinish2(a, b, c, d)
inline int findExeNameStart(const char *exec_name)
{
int exename_start = (int)strlen(exec_name);
while( (exename_start > 0) &&
(exec_name[exename_start] != '\\') &&
(exec_name[exename_start] != '/') )
{
exename_start--;
}
if (exec_name[exename_start] == '\\' ||
exec_name[exename_start] == '/')
{
return exename_start+1;
} else {
return exename_start;
}
}
inline int __shrQAStart(int argc, char **argv)
{
bool bQATest = false;
// First clear the output buffer
fflush(stdout);
fflush(stdout);
for (int i=1; i < argc; i++) {
int string_start = 0;
while (argv[i][string_start] == '-')
string_start++;
char *string_argv = &argv[i][string_start];
if (!STRCASECMP(string_argv, "qatest")) {
bQATest = true;
}
}
// We don't want to print the entire path, so we search for the first
int exename_start = findExeNameStart(argv[0]);
if (bQATest) {
fprintf(stdout, "&&&& RUNNING %s", &(argv[0][exename_start]));
for (int i=1; i < argc; i++) fprintf(stdout, " %s", argv[i]);
fprintf(stdout, "\n");
} else {
fprintf(stdout, "[%s] starting...\n", &(argv[0][exename_start]));
}
fflush(stdout);
printf("\n"); fflush(stdout);
return exename_start;
}
enum eQAstatus {
QA_FAILED = 0,
QA_PASSED = 1,
QA_WAIVED = 2
};
inline void __ExitInTime(int seconds)
{
fprintf(stdout, "> exiting in %d seconds: ", seconds);
fflush(stdout);
time_t t;
int count;
for (t=time(0)+seconds, count=seconds; time(0) < t; count--) {
fprintf(stdout, "%d...", count);
#ifdef WIN32
Sleep(1000);
#else
sleep(1);
#endif
}
fprintf(stdout,"done!\n\n");
fflush(stdout);
}
inline void __shrQAFinish(int argc, const char **argv, int iStatus)
{
// By default QATest is disabled and NoPrompt is Enabled (times out at seconds passed into __ExitInTime() )
bool bQATest = false, bNoPrompt = true, bQuitInTime = true;
const char *sStatus[] = { "FAILED", "PASSED", "WAIVED", NULL };
for (int i=1; i < argc; i++) {
int string_start = 0;
while (argv[i][string_start] == '-')
string_start++;
const char *string_argv = &argv[i][string_start];
if (!STRCASECMP(string_argv, "qatest")) {
bQATest = true;
}
// For SDK individual samples that don't specify -noprompt or -prompt,
// a 3 second delay will happen before exiting, giving a user time to view results
if (!STRCASECMP(string_argv, "noprompt") || !STRCASECMP(string_argv, "help")) {
bNoPrompt = true;
bQuitInTime = false;
}
if (!STRCASECMP(string_argv, "prompt")) {
bNoPrompt = false;
bQuitInTime = false;
}
}
int exename_start = findExeNameStart(argv[0]);
if (bQATest) {
fprintf(stdout, "&&&& %s %s", sStatus[iStatus], &(argv[0][exename_start]));
for (int i=1; i < argc; i++) fprintf(stdout, " %s", argv[i]);
fprintf(stdout, "\n");
} else {
fprintf(stdout, "[%s] test results...\n%s\n", &(argv[0][exename_start]), sStatus[iStatus]);
}
fflush(stdout);
printf("\n"); fflush(stdout);
if (bQuitInTime) {
__ExitInTime(3);
} else {
if (!bNoPrompt) {
fprintf(stdout, "\nPress <Enter> to exit...\n");
fflush(stdout);
getchar();
}
}
}
inline void __shrQAFinish2(bool bQATest, int argc, const char **argv, int iStatus)
{
bool bQuitInTime = true;
const char *sStatus[] = { "FAILED", "PASSED", "WAIVED", NULL };
for (int i=1; i < argc; i++) {
int string_start = 0;
while (argv[i][string_start] == '-')
string_start++;
const char *string_argv = &argv[i][string_start];
// For SDK individual samples that don't specify -noprompt or -prompt,
// a 3 second delay will happen before exiting, giving a user time to view results
if (!STRCASECMP(string_argv, "noprompt") || !STRCASECMP(string_argv, "help")) {
bQuitInTime = false;
}
if (!STRCASECMP(string_argv, "prompt")) {
bQuitInTime = false;
}
}
int exename_start = findExeNameStart(argv[0]);
if (bQATest) {
fprintf(stdout, "&&&& %s %s", sStatus[iStatus], &(argv[0][exename_start]));
for (int i=1; i < argc; i++) fprintf(stdout, " %s", argv[i]);
fprintf(stdout, "\n");
} else {
fprintf(stdout, "[%s] test results...\n%s\n", &(argv[0][exename_start]), sStatus[iStatus]);
}
fflush(stdout);
if (bQuitInTime) {
__ExitInTime(3);
}
}
inline void shrQAFinishExit(int argc, const char **argv, int iStatus)
{
__shrQAFinish(argc, argv, iStatus);
exit(iStatus ? EXIT_SUCCESS : EXIT_FAILURE);
}
inline void shrQAFinishExit2(bool bQAtest, int argc, const char **argv, int iStatus)
{
__shrQAFinish2(bQAtest, argc, argv, iStatus);
exit(iStatus ? EXIT_SUCCESS : EXIT_FAILURE);
}
#endif

642
benchmarks/opencl/BlackScholes/shrUtils.h Normal file
View file

@ -0,0 +1,642 @@
/*
* Copyright 1993-2010 NVIDIA Corporation. All rights reserved.
*
* Please refer to the NVIDIA end user license agreement (EULA) associated
* with this source code for terms and conditions that govern your use of
* this software. Any use, reproduction, disclosure, or distribution of
* this software and related documentation outside the terms of the EULA
* is strictly prohibited.
*
*/
#ifndef SHR_UTILS_H
#define SHR_UTILS_H
// *********************************************************************
// Generic utilities for NVIDIA GPU Computing SDK
// *********************************************************************
// reminders for output window and build log
#ifdef _WIN32
#pragma message ("Note: including windows.h")
#pragma message ("Note: including math.h")
#pragma message ("Note: including assert.h")
#endif
// OS dependent includes
#ifdef _WIN32
// Headers needed for Windows
#include <windows.h>
#else
// Headers needed for Linux
#include <sys/stat.h>
#include <sys/types.h>
#include <sys/time.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <stdarg.h>
#endif
// Other headers needed for both Windows and Linux
#include <math.h>
#include <assert.h>
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
// Un-comment the following #define to enable profiling code in SDK apps
//#define GPU_PROFILING
// Beginning of GPU Architecture definitions
inline int ConvertSMVer2Cores(int major, int minor)
{
// Defines for GPU Architecture types (using the SM version to determine the # of cores per SM
typedef struct {
int SM; // 0xMm (hexidecimal notation), M = SM Major version, and m = SM minor version
int Cores;
} sSMtoCores;
sSMtoCores nGpuArchCoresPerSM[] =
{ { 0x10, 8 }, // Tesla Generation (SM 1.0) G80 class
{ 0x11, 8 }, // Tesla Generation (SM 1.1) G8x class
{ 0x12, 8 }, // Tesla Generation (SM 1.2) G9x class
{ 0x13, 8 }, // Tesla Generation (SM 1.3) GT200 class
{ 0x20, 32 }, // Fermi Generation (SM 2.0) GF100 class
{ 0x21, 48 }, // Fermi Generation (SM 2.1) GF10x class
{ 0x30, 192}, // Fermi Generation (SM 3.0) GK10x class
{ -1, -1 }
};
int index = 0;
while (nGpuArchCoresPerSM[index].SM != -1) {
if (nGpuArchCoresPerSM[index].SM == ((major << 4) + minor) ) {
return nGpuArchCoresPerSM[index].Cores;
}
index++;
}
printf("MapSMtoCores SM %d.%d is undefined (please update to the latest SDK)!\n", major, minor);
return -1;
}
// end of GPU Architecture definitions
// Defines and enum for use with logging functions
// *********************************************************************
#define DEFAULTLOGFILE "SdkConsoleLog.txt"
#define MASTERLOGFILE "SdkMasterLog.csv"
enum LOGMODES
{
LOGCONSOLE = 1, // bit to signal "log to console"
LOGFILE = 2, // bit to signal "log to file"
LOGBOTH = 3, // convenience union of first 2 bits to signal "log to both"
APPENDMODE = 4, // bit to set "file append" mode instead of "replace mode" on open
MASTER = 8, // bit to signal master .csv log output
ERRORMSG = 16, // bit to signal "pre-pend Error"
CLOSELOG = 32 // bit to close log file, if open, after any requested file write
};
#define HDASHLINE "-----------------------------------------------------------\n"
// Standardized boolean
enum shrBOOL
{
shrFALSE = 0,
shrTRUE = 1
};
// Standardized MAX, MIN and CLAMP
#define MAX(a, b) ((a > b) ? a : b)
#define MIN(a, b) ((a < b) ? a : b)
#define CLAMP(a, b, c) MIN(MAX(a, b), c) // double sided clip of input a
#define TOPCLAMP(a, b) (a < b ? a:b) // single top side clip of input a
// Error and Exit Handling Macros...
// *********************************************************************
// Full error handling macro with Cleanup() callback (if supplied)...
// (Companion Inline Function lower on page)
#define shrCheckErrorEX(a, b, c) __shrCheckErrorEX(a, b, c, __FILE__ , __LINE__)
// Short version without Cleanup() callback pointer
// Both Input (a) and Reference (b) are specified as args
#define shrCheckError(a, b) shrCheckErrorEX(a, b, 0)
// Standardized Exit Macro for leaving main()... extended version
// (Companion Inline Function lower on page)
#define shrExitEX(a, b, c) __shrExitEX(a, b, c)
// Standardized Exit Macro for leaving main()... short version
// (Companion Inline Function lower on page)
#define shrEXIT(a, b) __shrExitEX(a, b, EXIT_SUCCESS)
// Simple argument checker macro
#define ARGCHECK(a) if((a) != shrTRUE)return shrFALSE
// Define for user-customized error handling
#define STDERROR "file %s, line %i\n\n" , __FILE__ , __LINE__
// Function to deallocate memory allocated within shrUtils
// *********************************************************************
extern "C" void shrFree(void* ptr);
// *********************************************************************
// Helper function to log standardized information to Console, to File or to both
//! Examples: shrLogEx(LOGBOTH, 0, "Function A\n");
//! : shrLogEx(LOGBOTH | ERRORMSG, ciErrNum, STDERROR);
//!
//! Automatically opens file and stores handle if needed and not done yet
//! Closes file and nulls handle on request
//!
//! @param 0 iLogMode: LOGCONSOLE, LOGFILE, LOGBOTH, APPENDMODE, MASTER, ERRORMSG, CLOSELOG.
//! LOGFILE and LOGBOTH may be | 'd with APPENDMODE to select file append mode instead of overwrite mode
//! LOGFILE and LOGBOTH may be | 'd with CLOSELOG to "write and close"
//! First 3 options may be | 'd with MASTER to enable independent write to master data log file
//! First 3 options may be | 'd with ERRORMSG to start line with standard error message
//! @param 2 dValue:
//! Positive val = double value for time in secs to be formatted to 6 decimals.
//! Negative val is an error code and this give error preformatting.
//! @param 3 cFormatString: String with formatting specifiers like printf or fprintf.
//! ALL printf flags, width, precision and type specifiers are supported with this exception:
//! Wide char type specifiers intended for wprintf (%S and %C) are NOT supported
//! Single byte char type specifiers (%s and %c) ARE supported
//! @param 4... variable args: like printf or fprintf. Must match format specifer type above.
//! @return 0 if OK, negative value on error or if error occurs or was passed in.
// *********************************************************************
extern "C" int shrLogEx(int iLogMode, int iErrNum, const char* cFormatString, ...);
// Short version of shrLogEx defaulting to shrLogEx(LOGBOTH, 0,
// *********************************************************************
extern "C" int shrLog(const char* cFormatString, ...);
// *********************************************************************
// Delta timer function for up to 3 independent timers using host high performance counters
// Maintains state for 3 independent counters
//! Example: double dElapsedTime = shrDeltaTime(0);
//!
//! @param 0 iCounterID: Which timer to check/reset. (0, 1, 2)
//! @return delta time of specified counter since last call in seconds. Otherwise -9999.0 if error
// *********************************************************************
extern "C" double shrDeltaT(int iCounterID);
// Optional LogFileNameOverride function
// *********************************************************************
extern "C" void shrSetLogFileName (const char* cOverRideName);
// Helper function to init data arrays
// *********************************************************************
extern "C" void shrFillArray(float* pfData, int iSize);
// Helper function to print data arrays
// *********************************************************************
extern "C" void shrPrintArray(float* pfData, int iSize);
////////////////////////////////////////////////////////////////////////////
//! Find the path for a filename
//! @return the path if succeeded, otherwise 0
//! @param filename name of the file
//! @param executablePath optional absolute path of the executable
////////////////////////////////////////////////////////////////////////////
extern "C" char* shrFindFilePath(const char* filename, const char* executablePath);
////////////////////////////////////////////////////////////////////////////
//! Read file \filename containing single precision floating point data
//! @return shrTRUE if reading the file succeeded, otherwise shrFALSE
//! @param filename name of the source file
//! @param data uninitialized pointer, returned initialized and pointing to
//! the data read
//! @param len number of data elements in data, -1 on error
//! @note If a NULL pointer is passed to this function and it is initialized
//! within shrUtils, then free() has to be used to deallocate the memory
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrReadFilef( const char* filename, float** data, unsigned int* len,
bool verbose = false);
////////////////////////////////////////////////////////////////////////////
//! Read file \filename containing double precision floating point data
//! @return shrTRUE if reading the file succeeded, otherwise shrFALSE
//! @param filename name of the source file
//! @param data uninitialized pointer, returned initialized and pointing to
//! the data read
//! @param len number of data elements in data, -1 on error
//! @note If a NULL pointer is passed to this function and it is
//! @note If a NULL pointer is passed to this function and it is initialized
//! within shrUtils, then free() has to be used to deallocate the memory
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrReadFiled( const char* filename, double** data, unsigned int* len,
bool verbose = false);
////////////////////////////////////////////////////////////////////////////
//! Read file \filename containing integer data
//! @return shrTRUE if reading the file succeeded, otherwise shrFALSE
//! @param filename name of the source file
//! @param data uninitialized pointer, returned initialized and pointing to
//! the data read
//! @param len number of data elements in data, -1 on error
//! @note If a NULL pointer is passed to this function and it is
//! @note If a NULL pointer is passed to this function and it is initialized
//! within shrUtils, then free() has to be used to deallocate the memory
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrReadFilei( const char* filename, int** data, unsigned int* len, bool verbose = false);
////////////////////////////////////////////////////////////////////////////
//! Read file \filename containing unsigned integer data
//! @return shrTRUE if reading the file succeeded, otherwise shrFALSE
//! @param filename name of the source file
//! @param data uninitialized pointer, returned initialized and pointing to
//! the data read
//! @param len number of data elements in data, -1 on error
//! @note If a NULL pointer is passed to this function and it is
//! @note If a NULL pointer is passed to this function and it is initialized
//! within shrUtils, then free() has to be used to deallocate the memory
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrReadFileui( const char* filename, unsigned int** data,
unsigned int* len, bool verbose = false);
////////////////////////////////////////////////////////////////////////////
//! Read file \filename containing char / byte data
//! @return shrTRUE if reading the file succeeded, otherwise shrFALSE
//! @param filename name of the source file
//! @param data uninitialized pointer, returned initialized and pointing to
//! the data read
//! @param len number of data elements in data, -1 on error
//! @note If a NULL pointer is passed to this function and it is
//! @note If a NULL pointer is passed to this function and it is initialized
//! within shrUtils, then free() has to be used to deallocate the memory
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrReadFileb( const char* filename, char** data, unsigned int* len,
bool verbose = false);
////////////////////////////////////////////////////////////////////////////
//! Read file \filename containing unsigned char / byte data
//! @return shrTRUE if reading the file succeeded, otherwise shrFALSE
//! @param filename name of the source file
//! @param data uninitialized pointer, returned initialized and pointing to
//! the data read
//! @param len number of data elements in data, -1 on error
//! @note If a NULL pointer is passed to this function and it is
//! @note If a NULL pointer is passed to this function and it is initialized
//! within shrUtils, then free() has to be used to deallocate the memory
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrReadFileub( const char* filename, unsigned char** data,
unsigned int* len, bool verbose = false);
////////////////////////////////////////////////////////////////////////////
//! Write a data file \filename containing single precision floating point
//! data
//! @return shrTRUE if writing the file succeeded, otherwise shrFALSE
//! @param filename name of the file to write
//! @param data pointer to data to write
//! @param len number of data elements in data, -1 on error
//! @param epsilon epsilon for comparison
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrWriteFilef( const char* filename, const float* data, unsigned int len,
const float epsilon, bool verbose = false);
////////////////////////////////////////////////////////////////////////////
//! Write a data file \filename containing double precision floating point
//! data
//! @return shrTRUE if writing the file succeeded, otherwise shrFALSE
//! @param filename name of the file to write
//! @param data pointer to data to write
//! @param len number of data elements in data, -1 on error
//! @param epsilon epsilon for comparison
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrWriteFiled( const char* filename, const float* data, unsigned int len,
const double epsilon, bool verbose = false);
////////////////////////////////////////////////////////////////////////////
//! Write a data file \filename containing integer data
//! @return shrTRUE if writing the file succeeded, otherwise shrFALSE
//! @param filename name of the file to write
//! @param data pointer to data to write
//! @param len number of data elements in data, -1 on error
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrWriteFilei( const char* filename, const int* data, unsigned int len,
bool verbose = false);
////////////////////////////////////////////////////////////////////////////
//! Write a data file \filename containing unsigned integer data
//! @return shrTRUE if writing the file succeeded, otherwise shrFALSE
//! @param filename name of the file to write
//! @param data pointer to data to write
//! @param len number of data elements in data, -1 on error
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrWriteFileui( const char* filename, const unsigned int* data,
unsigned int len, bool verbose = false);
////////////////////////////////////////////////////////////////////////////
//! Write a data file \filename containing char / byte data
//! @return shrTRUE if writing the file succeeded, otherwise shrFALSE
//! @param filename name of the file to write
//! @param data pointer to data to write
//! @param len number of data elements in data, -1 on error
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrWriteFileb( const char* filename, const char* data, unsigned int len,
bool verbose = false);
////////////////////////////////////////////////////////////////////////////
//! Write a data file \filename containing unsigned char / byte data
//! @return shrTRUE if writing the file succeeded, otherwise shrFALSE
//! @param filename name of the file to write
//! @param data pointer to data to write
//! @param len number of data elements in data, -1 on error
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrWriteFileub( const char* filename, const unsigned char* data,
unsigned int len, bool verbose = false);
////////////////////////////////////////////////////////////////////////////
//! Load PPM image file (with unsigned char as data element type), padding
//! 4th component
//! @return shrTRUE if reading the file succeeded, otherwise shrFALSE
//! @param file name of the image file
//! @param OutData handle to the data read
//! @param w width of the image
//! @param h height of the image
//!
//! Note: If *OutData is NULL this function allocates buffer that must be freed by caller
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrLoadPPM4ub(const char* file, unsigned char** OutData,
unsigned int *w, unsigned int *h);
////////////////////////////////////////////////////////////////////////////
//! Save PPM image file (with unsigned char as data element type, padded to
//! 4 bytes)
//! @return shrTRUE if saving the file succeeded, otherwise shrFALSE
//! @param file name of the image file
//! @param data handle to the data read
//! @param w width of the image
//! @param h height of the image
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrSavePPM4ub( const char* file, unsigned char *data,
unsigned int w, unsigned int h);
////////////////////////////////////////////////////////////////////////////////
//! Save PGM image file (with unsigned char as data element type)
//! @return shrTRUE if saving the file succeeded, otherwise shrFALSE
//! @param file name of the image file
//! @param data handle to the data read
//! @param w width of the image
//! @param h height of the image
////////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrSavePGMub( const char* file, unsigned char *data,
unsigned int w, unsigned int h);
////////////////////////////////////////////////////////////////////////////
//! Load PGM image file (with unsigned char as data element type)
//! @return shrTRUE if saving the file succeeded, otherwise shrFALSE
//! @param file name of the image file
//! @param data handle to the data read
//! @param w width of the image
//! @param h height of the image
//! @note If a NULL pointer is passed to this function and it is initialized
//! within shrUtils, then free() has to be used to deallocate the memory
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrLoadPGMub( const char* file, unsigned char** data,
unsigned int *w,unsigned int *h);
////////////////////////////////////////////////////////////////////////////
// Command line arguments: General notes
// * All command line arguments begin with '--' followed by the token;
// token and value are seperated by '='; example --samples=50
// * Arrays have the form --model=[one.obj,two.obj,three.obj]
// (without whitespaces)
////////////////////////////////////////////////////////////////////////////
////////////////////////////////////////////////////////////////////////////
//! Check if command line argument \a flag-name is given
//! @return shrTRUE if command line argument \a flag_name has been given,
//! otherwise shrFALSE
//! @param argc argc as passed to main()
//! @param argv argv as passed to main()
//! @param flag_name name of command line flag
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrCheckCmdLineFlag( const int argc, const char** argv,
const char* flag_name);
////////////////////////////////////////////////////////////////////////////
//! Get the value of a command line argument of type int
//! @return shrTRUE if command line argument \a arg_name has been given and
//! is of the requested type, otherwise shrFALSE
//! @param argc argc as passed to main()
//! @param argv argv as passed to main()
//! @param arg_name name of the command line argument
//! @param val value of the command line argument
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrGetCmdLineArgumenti( const int argc, const char** argv,
const char* arg_name, int* val);
////////////////////////////////////////////////////////////////////////////
//! Get the value of a command line argument of type unsigned int
//! @return shrTRUE if command line argument \a arg_name has been given and
//! is of the requested type, otherwise shrFALSE
//! @param argc argc as passed to main()
//! @param argv argv as passed to main()
//! @param arg_name name of the command line argument
//! @param val value of the command line argument
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrGetCmdLineArgumentu( const int argc, const char** argv,
const char* arg_name, unsigned int* val);
////////////////////////////////////////////////////////////////////////////
//! Get the value of a command line argument of type float
//! @return shrTRUE if command line argument \a arg_name has been given and
//! is of the requested type, otherwise shrFALSE
//! @param argc argc as passed to main()
//! @param argv argv as passed to main()
//! @param arg_name name of the command line argument
//! @param val value of the command line argument
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrGetCmdLineArgumentf( const int argc, const char** argv,
const char* arg_name, float* val);
////////////////////////////////////////////////////////////////////////////
//! Get the value of a command line argument of type string
//! @return shrTRUE if command line argument \a arg_name has been given and
//! is of the requested type, otherwise shrFALSE
//! @param argc argc as passed to main()
//! @param argv argv as passed to main()
//! @param arg_name name of the command line argument
//! @param val value of the command line argument
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrGetCmdLineArgumentstr( const int argc, const char** argv,
const char* arg_name, char** val);
////////////////////////////////////////////////////////////////////////////
//! Get the value of a command line argument list those element are strings
//! @return shrTRUE if command line argument \a arg_name has been given and
//! is of the requested type, otherwise shrFALSE
//! @param argc argc as passed to main()
//! @param argv argv as passed to main()
//! @param arg_name name of the command line argument
//! @param val command line argument list
//! @param len length of the list / number of elements
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrGetCmdLineArgumentListstr( const int argc, const char** argv,
const char* arg_name, char** val,
unsigned int* len);
////////////////////////////////////////////////////////////////////////////
//! Compare two float arrays
//! @return shrTRUEif \a reference and \a data are identical, otherwise shrFALSE
//! @param reference handle to the reference data / gold image
//! @param data handle to the computed data
//! @param len number of elements in reference and data
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrComparef( const float* reference, const float* data,
const unsigned int len);
////////////////////////////////////////////////////////////////////////////
//! Compare two integer arrays
//! @return shrTRUEif \a reference and \a data are identical, otherwise shrFALSE
//! @param reference handle to the reference data / gold image
//! @param data handle to the computed data
//! @param len number of elements in reference and data
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrComparei( const int* reference, const int* data,
const unsigned int len );
////////////////////////////////////////////////////////////////////////////////
//! Compare two unsigned integer arrays, with epsilon and threshold
//! @return shrTRUEif \a reference and \a data are identical, otherwise shrFALSE
//! @param reference handle to the reference data / gold image
//! @param data handle to the computed data
//! @param len number of elements in reference and data
//! @param threshold tolerance % # of comparison errors (0.15f = 15%)
////////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrCompareuit( const unsigned int* reference, const unsigned int* data,
const unsigned int len, const float epsilon, const float threshold );
////////////////////////////////////////////////////////////////////////////
//! Compare two unsigned char arrays
//! @return shrTRUEif \a reference and \a data are identical, otherwise shrFALSE
//! @param reference handle to the reference data / gold image
//! @param data handle to the computed data
//! @param len number of elements in reference and data
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrCompareub( const unsigned char* reference, const unsigned char* data,
const unsigned int len );
////////////////////////////////////////////////////////////////////////////////
//! Compare two integers with a tolernance for # of byte errors
//! @return shrTRUEif \a reference and \a data are identical, otherwise shrFALSE
//! @param reference handle to the reference data / gold image
//! @param data handle to the computed data
//! @param len number of elements in reference and data
//! @param epsilon epsilon to use for the comparison
//! @param threshold tolerance % # of comparison errors (0.15f = 15%)
////////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrCompareubt( const unsigned char* reference, const unsigned char* data,
const unsigned int len, const float epsilon, const float threshold );
////////////////////////////////////////////////////////////////////////////////
//! Compare two integer arrays witha n epsilon tolerance for equality
//! @return shrTRUEif \a reference and \a data are identical, otherwise shrFALSE
//! @param reference handle to the reference data / gold image
//! @param data handle to the computed data
//! @param len number of elements in reference and data
//! @param epsilon epsilon to use for the comparison
////////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrCompareube( const unsigned char* reference, const unsigned char* data,
const unsigned int len, const float epsilon );
////////////////////////////////////////////////////////////////////////////
//! Compare two float arrays with an epsilon tolerance for equality
//! @return shrTRUEif \a reference and \a data are identical, otherwise shrFALSE
//! @param reference handle to the reference data / gold image
//! @param data handle to the computed data
//! @param len number of elements in reference and data
//! @param epsilon epsilon to use for the comparison
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrComparefe( const float* reference, const float* data,
const unsigned int len, const float epsilon );
////////////////////////////////////////////////////////////////////////////////
//! Compare two float arrays with an epsilon tolerance for equality and a
//! threshold for # pixel errors
//! @return shrTRUEif \a reference and \a data are identical, otherwise shrFALSE
//! @param reference handle to the reference data / gold image
//! @param data handle to the computed data
//! @param len number of elements in reference and data
//! @param epsilon epsilon to use for the comparison
////////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrComparefet( const float* reference, const float* data,
const unsigned int len, const float epsilon, const float threshold );
////////////////////////////////////////////////////////////////////////////
//! Compare two float arrays using L2-norm with an epsilon tolerance for
//! equality
//! @return shrTRUEif \a reference and \a data are identical, otherwise shrFALSE
//! @param reference handle to the reference data / gold image
//! @param data handle to the computed data
//! @param len number of elements in reference and data
//! @param epsilon epsilon to use for the comparison
////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrCompareL2fe( const float* reference, const float* data,
const unsigned int len, const float epsilon );
////////////////////////////////////////////////////////////////////////////////
//! Compare two PPM image files with an epsilon tolerance for equality
//! @return shrTRUEif \a reference and \a data are identical, otherwise shrFALSE
//! @param src_file filename for the image to be compared
//! @param data filename for the reference data / gold image
//! @param epsilon epsilon to use for the comparison
//! @param threshold threshold of pixels that can still mismatch to pass (i.e. 0.15f = 15% must pass)
//! $param verboseErrors output details of image mismatch to std::err
////////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrComparePPM( const char *src_file, const char *ref_file, const float epsilon, const float threshold);
////////////////////////////////////////////////////////////////////////////////
//! Compare two PGM image files with an epsilon tolerance for equality
//! @return shrTRUEif \a reference and \a data are identical, otherwise shrFALSE
//! @param src_file filename for the image to be compared
//! @param data filename for the reference data / gold image
//! @param epsilon epsilon to use for the comparison
//! @param threshold threshold of pixels that can still mismatch to pass (i.e. 0.15f = 15% must pass)
//! $param verboseErrors output details of image mismatch to std::err
////////////////////////////////////////////////////////////////////////////////
extern "C" shrBOOL shrComparePGM( const char *src_file, const char *ref_file, const float epsilon, const float threshold);
extern "C" unsigned char* shrLoadRawFile(const char* filename, size_t size);
extern "C" size_t shrRoundUp(int group_size, int global_size);
// companion inline function for error checking and exit on error WITH Cleanup Callback (if supplied)
// *********************************************************************
inline void __shrCheckErrorEX(int iSample, int iReference, void (*pCleanup)(int), const char* cFile, const int iLine)
{
if (iReference != iSample)
{
shrLogEx(LOGBOTH | ERRORMSG, iSample, "line %i , in file %s !!!\n\n" , iLine, cFile);
if (pCleanup != NULL)
{
pCleanup(EXIT_FAILURE);
}
else
{
shrLogEx(LOGBOTH | CLOSELOG, 0, "Exiting...\n");
exit(EXIT_FAILURE);
}
}
}
// Standardized Exit
// *********************************************************************
inline void __shrExitEX(int argc, const char** argv, int iExitCode)
{
#ifdef WIN32
if (!shrCheckCmdLineFlag(argc, argv, "noprompt") && !shrCheckCmdLineFlag(argc, argv, "qatest"))
#else
if (shrCheckCmdLineFlag(argc, argv, "prompt") && !shrCheckCmdLineFlag(argc, argv, "qatest"))
#endif
{
shrLogEx(LOGBOTH | CLOSELOG, 0, "\nPress <Enter> to Quit...\n");
getchar();
}
else
{
shrLogEx(LOGBOTH | CLOSELOG, 0, "%s Exiting...\n", argv[0]);
}
fflush(stderr);
exit(iExitCode);
}
#endif