SSE指令集的介绍网上一大堆, 这里贴一个用VS2008环境下的SSE测试程序, 分别用C++代码, C++内联汇编, C++的SSE Intrinsics三种方式计算卷积的程序...这是一个win32控制台程序.....

程序下载地址 : http://download.csdn.net/detail/hemmingway/4598506

主文件的代码一览:

// Test_SSE.cpp : 定义控制台应用程序的入口点。
// calc conversion
//#include "stdafx.h"
#include <xmmintrin.h>      // __m128 data type and SSE functions
#include <float.h>
#include <math.h>
#include <Windows.h>		// Support odprintf
#include <stdarg.h>
#include <ctype.h>
#include "MMX_SSESupport.h"
#include "TimeCounter.h"#define ARRAY_SIZE 100000#pragma warning(disable : 4324)// Arrays processed by SSE should have 16 bytes alignment:
__declspec(align(16)) float m_fInitialArray[ARRAY_SIZE];
__declspec(align(16)) float m_fResultArray[ARRAY_SIZE];// minimum and maximum values in the result array
float m_fMin;
float m_fMax;#define TIME_START CTimeCounter* pT = new CTimeCounter()
#define TIME_END   ShowTime(pT->GetExecutionTime())//
//odprintf -- debug function
void __cdecl odprintf(const char* fmt, ...)
{char buf[4096], *p = buf;va_list args;va_start(args, fmt);p += vsnprintf_s(p, sizeof(buf), _TRUNCATE, fmt, args);va_end(args);while ( p > buf  &&  isspace(p[-1]) )*--p = '\0';*p++ = '\r';*p++ = '\n';*p   = '\0';OutputDebugStringA(buf);	//output as ANSI string	//OutputDebugString
}//
// Show execution time (ms)
void ShowTime(__int64 nTime)
{printf("usage time: %I64d\n\n",nTime);		//在g++中对应的是<stdint.h> int64_t, 应该用%lld输出
}//
// ShowArray, display array's data
void ShowArray(float* pArray)
{if ( !(*pArray))return;float* p = pArray;for ( int i = 0; i < ARRAY_SIZE; i += 500 )		//没有显示所有的数据出来{printf("%f  ", p[i]);if (i == 5)printf("\n");}printf("\n\n");
}//
// InitArray, Fill initial array
void InitArray()
{m_fMin = FLT_MAX;m_fMax = FLT_MIN;float f;int i;for ( i = 0; i < ARRAY_SIZE; i++ ){// Fill array with one sin cycle and ensure that all values are positive// (to use sqrt in conversion)f = (float) sin(((double)i * 6.29 / ARRAY_SIZE)) + 2.0f;if ( f < m_fMin )m_fMin = f;if ( f > m_fMax )m_fMax = f;m_fInitialArray[i] = f;}ShowArray(m_fInitialArray);
}//
// Make conversion using C++ code
//
// Each initial array member is converted to result array member
// using some formula (just to demonstrate SSE features).
// Minimum and maximum result values are calculated and shown.
//
// Function also calculates and shows conversion time (ms).
//
void OnCplusplus()
{TIME_START;m_fMin = FLT_MAX;m_fMax = FLT_MIN;int i;for ( i = 0; i < ARRAY_SIZE; i++ ){m_fResultArray[i] = sqrt(m_fInitialArray[i]  * 2.8f);if ( m_fResultArray[i] < m_fMin )m_fMin = m_fResultArray[i];if ( m_fResultArray[i] > m_fMax )m_fMax = m_fResultArray[i];}TIME_END;ShowArray(m_fResultArray);
}//
//OnSseAssembly, Make conversion using C++ code with inline Assembly
void OnSseAssembly()
{TIME_START;float* pIn = m_fInitialArray;float* pOut = m_fResultArray;float f = 2.8f;float flt_min = FLT_MIN;float flt_max = FLT_MAX;__m128 min128;__m128 max128;// using additional registers:// xmm2 - multiplication coefficient// xmm3 - minimum// xmm4 - maximum_asm{movss   xmm2, f                         // xmm2[0] = 2.8shufps  xmm2, xmm2, 0                   // xmm2[1, 2, 3] = xmm2[0]movss   xmm3, flt_max                   // xmm3 = FLT_MAXshufps  xmm3, xmm3, 0                   // xmm3[1, 2, 3] = xmm3[0]movss   xmm4, flt_min                   // xmm4 = FLT_MINshufps  xmm4, xmm4, 0                   // xmm3[1, 2, 3] = xmm3[0]mov         esi, pIn                    // input pointermov         edi, pOut                   // output pointermov         ecx, ARRAY_SIZE/4           // loop counterstart_loop:movaps      xmm1, [esi]                 // xmm1 = [esi]mulps       xmm1, xmm2                  // xmm1 = xmm1 * xmm2sqrtps      xmm1, xmm1                  // xmm1 = sqrt(xmm1)movaps      [edi], xmm1                 // [edi] = xmm1minps       xmm3, xmm1maxps       xmm4, xmm1add         esi, 16add         edi, 16dec         ecxjnz         start_loopmovaps      min128, xmm3movaps      max128, xmm4}// extract minimum and maximum values from min128 and max128union u{__m128 m;float f[4];} x;x.m = min128;m_fMin = min(x.f[0], min(x.f[1], min(x.f[2], x.f[3])));x.m = max128;m_fMax = max(x.f[0], max(x.f[1], max(x.f[2], x.f[3])));TIME_END;ShowArray(m_fResultArray);
}//
// OnSseCpp, Make conversion using C++ code with SSE Intrinsics
void OnSseCpp()
{TIME_START;__m128 coeff = _mm_set_ps1(2.8f);      // coeff[0, 1, 2, 3] = 2.8__m128 tmp;__m128 min128 = _mm_set_ps1(FLT_MAX);  // min128[0, 1, 2, 3] = FLT_MAX__m128 max128 = _mm_set_ps1(FLT_MIN);  // max128[0, 1, 2, 3] = FLT_MIN__m128* pSource = (__m128*) m_fInitialArray;__m128* pDest = (__m128*) m_fResultArray;for ( int i = 0; i < ARRAY_SIZE/4; i++ ){tmp = _mm_mul_ps(*pSource, coeff);      // tmp = *pSource * coeff*pDest = _mm_sqrt_ps(tmp);              // *pDest = sqrt(tmp)min128 =  _mm_min_ps(*pDest, min128);max128 =  _mm_max_ps(*pDest, max128);pSource++;pDest++;}// extract minimum and maximum values from min128 and max128union u{__m128 m;float f[4];} x;x.m = min128;m_fMin = min(x.f[0], min(x.f[1], min(x.f[2], x.f[3])));x.m = max128;m_fMax = max(x.f[0], max(x.f[1], max(x.f[2], x.f[3])));TIME_END;ShowArray(m_fResultArray);
}int _tmain(int argc, _TCHAR* argv[])
{// Test SSE support ?bool bMMX, bSSE;TestFeatures(&bMMX, &bSSE);if ( !bSSE ){// Do not support SSEodprintf("Do not support SSE.\n");return 0;}odprintf("everything is ok...");//first, prepare dataprintf("program generate %d floating point(Not all data are displayed)...\n\n", ARRAY_SIZE);InitArray();//second, Make conversion using C++ codegetchar();printf("Make conversion using C++ code\n\n");OnCplusplus();//third,Make conversion using C++ code with inline Assemblygetchar();printf("Make conversion using C++ code with inline Assembly\n\n");OnSseAssembly();//finally, Make conversion using C++ code with SSE Intrinsics getchar();printf("Make conversion using C++ code with SSE Intrinsics\n\n");OnSseCpp();getchar();return 0;
}