OpenCL Dot Product

小灰灰 2023-02-25 02:10 4阅读 0赞

#  OpenCL Dot Product #

## 1. OpenCL Dot Product ##

Dot Product (scalar product) of set of input vector pairs. Implemented in OpenCL for CUDA GPU’s, with functional comparison against a simple C++ host CPU implementation.  
输入向量对的点积 (标量积)。在针对 CUDA GPU 的 OpenCL 中实现，并与简单的 C++ 主机 CPU 实现进行了函数的比较。

函数调用
    DotProductHost ((const float*)srcA, (const float*)srcB, (float*)Golden, iNumElements);

// *********************************************************************
    void *srcA, *srcB, *dst;        // Host buffers for OpenCL test
    void* Golden;                   // Host buffer for host golden processing cross check
    
    // set and log Global and Local work size dimensions
    szLocalWorkSize = 256;
    szGlobalWorkSize = shrRoundUp((int)szLocalWorkSize, iNumElements);  // rounded up to the nearest multiple of the LocalWorkSize
        
    srcA = (void *)malloc(sizeof(cl_float4) * szGlobalWorkSize);
    srcB = (void *)malloc(sizeof(cl_float4) * szGlobalWorkSize);
    dst = (void *)malloc(sizeof(cl_float) * szGlobalWorkSize);
    Golden = (void *)malloc(sizeof(cl_float) * iNumElements);
    
    // Forward Declarations
    // *********************************************************************
    void DotProductHost(const float* pfData1, const float* pfData2, float* pfResult, int iNumElements);
    
    // "Golden" Host processing dot product function for comparison purposes
    // *********************************************************************
    void DotProductHost(const float* pfData1, const float* pfData2, float* pfResult, int iNumElements)
    {
    	int i, j, k;
    	for (i = 0, j = 0; i < iNumElements; i++)
    	{
    		pfResult[i] = 0.0f;
    		for (k = 0; k < 4; k++, j++)
    		{
    			pfResult[i] += pfData1[j] * pfData2[j];
    		}
    	}
    }

__kernel void DotProduct(__global float* a, __global float* b, __global float* c, int iNumElements)
    {
    	// find position in global arrays
    	int iGID = get_global_id(0);
    
    	// bound check (equivalent to the limit on a 'for' loop for standard/serial C code
    	if (iGID >= iNumElements)
    	{
    		return;
    	}
    
    	// process 
    	int iInOffset = iGID << 2;
    	c[iGID] = a[iInOffset] * b[iInOffset]
    		+ a[iInOffset + 1] * b[iInOffset + 1]
    		+ a[iInOffset + 2] * b[iInOffset + 2]
    		+ a[iInOffset + 3] * b[iInOffset + 3];
    }