MPI 应用程序中的分段和中止陷阱错误
Segmentation and Abort Trap error in MPI app
我正在尝试并行化一些使用 OpenMPI 计算 Mandelbrot 集的串行代码。作为第一步，我试图像这样在不同的进程之间划分工作：
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <math.h>
#include <time.h>
#include "mpi.h"
// Main program
int main(int argc, char* argv[])
{
/* screen ( integer) coordinate */
int iX,iY,i,j;
const int iXmax = 5; // default
const int iYmax = 5; // default
/* world ( double) coordinate = parameter plane*/
double Cx, Cy;
const double CxMin = -2.5;
const double CxMax = 1.5;
const double CyMin = -2.0;
const double CyMax = 2.0;
/* */
double PixelWidth = (CxMax - CxMin)/iXmax;
double PixelHeight = (CyMax - CyMin)/iYmax;
int linePerProcess, remainingLines, processMinY, processMaxY, lastProcessMaxY, result_offset;
int my_rank, processors;
unsigned char (*resultBuffer)[3] = NULL;
unsigned char (*resultBufferTwo)[3] = NULL;
unsigned char (*finalResultBuffer)[3] = NULL;
MPI_Status stat;
/* color component ( R or G or B) is coded from 0 to 255 */
/* it is 24 bit color RGB file */
const int MaxColorComponentValue = 255;
// RGB color array
unsigned char color[3];
/* Z = Zx + Zy*i; Z0 = 0 */
double Zx, Zy;
double Zx2, Zy2; /* Zx2 = Zx*Zx; Zy2 = Zy*Zy */
/* */
int Iteration;
const int IterationMax = 1000; // default
/* bail-out value , radius of circle ; */
const double EscapeRadius = 400;
double ER2 = EscapeRadius * EscapeRadius;
MPI_Init(&argc, &argv);
MPI_Comm_rank(MPI_COMM_WORLD, &my_rank);
MPI_Comm_size(MPI_COMM_WORLD, &processors);
linePerProcess = iYmax / processors;
remainingLines = iYmax % processors;
processMinY = my_rank * linePerProcess;
processMaxY = processMinY + linePerProcess;
lastProcessMaxY = processMaxY + remainingLines;
if (my_rank == 0) {
finalResultBuffer = malloc(iXmax * iYmax * sizeof(color));
for(iY = processMinY; iY < processMaxY; iY++) {
Cy = CyMin + (iY * PixelHeight);
if (fabs(Cy) < (PixelHeight / 2))
{
Cy = 0.0; /* Main antenna */
}
for(iX = 0; iX < iXmax; iX++)
{
Cx = CxMin + (iX * PixelWidth);
/* initial value of orbit = critical point Z= 0 */
Zx = 0.0;
Zy = 0.0;
Zx2 = Zx * Zx;
Zy2 = Zy * Zy;
/* */
for(Iteration = 0; Iteration < IterationMax && ((Zx2 + Zy2) < ER2); Iteration++)
{
Zy = (2 * Zx * Zy) + Cy;
Zx = Zx2 - Zy2 + Cx;
Zx2 = Zx * Zx;
Zy2 = Zy * Zy;
};
/* compute pixel color (24 bit = 3 bytes) */
if (Iteration == IterationMax)
{
// Point within the set. Mark it as black
color[0] = 0;
color[1] = 0;
color[2] = 0;
}
else
{
// Point outside the set. Mark it as white
double c = 3*log((double)Iteration)/log((double)(IterationMax) - 1.0);
if (c < 1)
{
color[0] = 0;
color[1] = 0;
color[2] = 255*c;
}
else if (c < 2)
{
color[0] = 0;
color[1] = 255*(c-1);
color[2] = 255;
}
else
{
color[0] = 255*(c-2);
color[1] = 255;
color[2] = 255;
}
}
finalResultBuffer[(iY*iXmax)+iX][0] = color[0];
finalResultBuffer[(iY*iXmax)+iX][1] = color[1];
finalResultBuffer[(iY*iXmax)+iX][2] = color[2];
}
}
free(finalResultBuffer);
} else if (my_rank == processors - 1) {
resultBuffer = malloc((linePerProcess + remainingLines) * iXmax * sizeof(color));
for(iY = processMinY; iY < lastProcessMaxY; iY++) {
Cy = CyMin + (iY * PixelHeight);
if (fabs(Cy) < (PixelHeight / 2))
{
Cy = 0.0; /* Main antenna */
}
for(iX = 0; iX < iXmax; iX++)
{
Cx = CxMin + (iX * PixelWidth);
/* initial value of orbit = critical point Z= 0 */
Zx = 0.0;
Zy = 0.0;
Zx2 = Zx * Zx;
Zy2 = Zy * Zy;
/* */
for(Iteration = 0; Iteration < IterationMax && ((Zx2 + Zy2) < ER2); Iteration++)
{
Zy = (2 * Zx * Zy) + Cy;
Zx = Zx2 - Zy2 + Cx;
Zx2 = Zx * Zx;
Zy2 = Zy * Zy;
};
/* compute pixel color (24 bit = 3 bytes) */
if (Iteration == IterationMax)
{
// Point within the set. Mark it as black
color[0] = 0;
color[1] = 0;
color[2] = 0;
}
else
{
// Point outside the set. Mark it as white
double c = 3*log((double)Iteration)/log((double)(IterationMax) - 1.0);
if (c < 1)
{
color[0] = 0;
color[1] = 0;
color[2] = 255*c;
}
else if (c < 2)
{
color[0] = 0;
color[1] = 255*(c-1);
color[2] = 255;
}
else
{
color[0] = 255*(c-2);
color[1] = 255;
color[2] = 255;
}
}
resultBuffer[(iY*iXmax)+iX][0] = color[0];
resultBuffer[(iY*iXmax)+iX][1] = color[1];
resultBuffer[(iY*iXmax)+iX][2] = color[2];
}
}
free(resultBuffer);
} else {
resultBufferTwo = malloc(linePerProcess * iXmax * sizeof(color));
for(iY = processMinY; iY < processMaxY; iY++) {
Cy = CyMin + (iY * PixelHeight);
if (fabs(Cy) < (PixelHeight / 2))
{
Cy = 0.0; /* Main antenna */
}
for(iX = 0; iX < iXmax; iX++)
{
Cx = CxMin + (iX * PixelWidth);
/* initial value of orbit = critical point Z= 0 */
Zx = 0.0;
Zy = 0.0;
Zx2 = Zx * Zx;
Zy2 = Zy * Zy;
/* */
for(Iteration = 0; Iteration < IterationMax && ((Zx2 + Zy2) < ER2); Iteration++)
{
Zy = (2 * Zx * Zy) + Cy;
Zx = Zx2 - Zy2 + Cx;
Zx2 = Zx * Zx;
Zy2 = Zy * Zy;
};
/* compute pixel color (24 bit = 3 bytes) */
if (Iteration == IterationMax)
{
// Point within the set. Mark it as black
color[0] = 0;
color[1] = 0;
color[2] = 0;
}
else
{
// Point outside the set. Mark it as white
double c = 3*log((double)Iteration)/log((double)(IterationMax) - 1.0);
if (c < 1)
{
color[0] = 0;
color[1] = 0;
color[2] = 255*c;
}
else if (c < 2)
{
color[0] = 0;
color[1] = 255*(c-1);
color[2] = 255;
}
else
{
color[0] = 255*(c-2);
color[1] = 255;
color[2] = 255;
}
}
resultBufferTwo[(iY*iXmax)+iX][0] = color[0];
resultBufferTwo[(iY*iXmax)+iX][1] = color[1];
resultBufferTwo[(iY*iXmax)+iX][2] = color[2];
}
}
free(resultBufferTwo);
}
MPI_Finalize();
return 0;
}
当我 运行 这样做时，我得到分段错误 (11) 以及中止陷阱 (6)，这仅发生在排名 1 及以上的进程中。 0号进程没有问题，谁能帮我理解这里的问题？
对于高于 0 的排名，您对结果缓冲区的索引是错误的。
缓冲区始终从索引 0 开始，无论代码在哪个级别 运行。但是，对于高于 0 的排名，您将从 processMinY * iXmax 开始索引。然后这会覆盖任意内存，这可能会导致段错误。
要解决此问题，您应该在计算缓冲区中的索引时从 iY 中减去 processMinY，例如resultBufferTwo[((iY-processMinY)*iXmax)+iX][0] = color[0];.
我还建议您尽可能统一不同级别的代码。现在，您的大部分代码都是重复的，尽管实际上唯一不同的行是 malloc 调用和写入缓冲区。这使得代码难以遵循。您应该为传递给 malloc 的大小和缓冲区创建一个变量，然后您将在循环中使用它。这样只有这两个变量的初始化需要在等级之间有所不同。
我正在尝试并行化一些使用 OpenMPI 计算 Mandelbrot 集的串行代码。作为第一步，我试图像这样在不同的进程之间划分工作：
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <math.h>
#include <time.h>
#include "mpi.h"
// Main program
int main(int argc, char* argv[])
{
/* screen ( integer) coordinate */
int iX,iY,i,j;
const int iXmax = 5; // default
const int iYmax = 5; // default
/* world ( double) coordinate = parameter plane*/
double Cx, Cy;
const double CxMin = -2.5;
const double CxMax = 1.5;
const double CyMin = -2.0;
const double CyMax = 2.0;
/* */
double PixelWidth = (CxMax - CxMin)/iXmax;
double PixelHeight = (CyMax - CyMin)/iYmax;
int linePerProcess, remainingLines, processMinY, processMaxY, lastProcessMaxY, result_offset;
int my_rank, processors;
unsigned char (*resultBuffer)[3] = NULL;
unsigned char (*resultBufferTwo)[3] = NULL;
unsigned char (*finalResultBuffer)[3] = NULL;
MPI_Status stat;
/* color component ( R or G or B) is coded from 0 to 255 */
/* it is 24 bit color RGB file */
const int MaxColorComponentValue = 255;
// RGB color array
unsigned char color[3];
/* Z = Zx + Zy*i; Z0 = 0 */
double Zx, Zy;
double Zx2, Zy2; /* Zx2 = Zx*Zx; Zy2 = Zy*Zy */
/* */
int Iteration;
const int IterationMax = 1000; // default
/* bail-out value , radius of circle ; */
const double EscapeRadius = 400;
double ER2 = EscapeRadius * EscapeRadius;
MPI_Init(&argc, &argv);
MPI_Comm_rank(MPI_COMM_WORLD, &my_rank);
MPI_Comm_size(MPI_COMM_WORLD, &processors);
linePerProcess = iYmax / processors;
remainingLines = iYmax % processors;
processMinY = my_rank * linePerProcess;
processMaxY = processMinY + linePerProcess;
lastProcessMaxY = processMaxY + remainingLines;
if (my_rank == 0) {
finalResultBuffer = malloc(iXmax * iYmax * sizeof(color));
for(iY = processMinY; iY < processMaxY; iY++) {
Cy = CyMin + (iY * PixelHeight);
if (fabs(Cy) < (PixelHeight / 2))
{
Cy = 0.0; /* Main antenna */
}
for(iX = 0; iX < iXmax; iX++)
{
Cx = CxMin + (iX * PixelWidth);
/* initial value of orbit = critical point Z= 0 */
Zx = 0.0;
Zy = 0.0;
Zx2 = Zx * Zx;
Zy2 = Zy * Zy;
/* */
for(Iteration = 0; Iteration < IterationMax && ((Zx2 + Zy2) < ER2); Iteration++)
{
Zy = (2 * Zx * Zy) + Cy;
Zx = Zx2 - Zy2 + Cx;
Zx2 = Zx * Zx;
Zy2 = Zy * Zy;
};
/* compute pixel color (24 bit = 3 bytes) */
if (Iteration == IterationMax)
{
// Point within the set. Mark it as black
color[0] = 0;
color[1] = 0;
color[2] = 0;
}
else
{
// Point outside the set. Mark it as white
double c = 3*log((double)Iteration)/log((double)(IterationMax) - 1.0);
if (c < 1)
{
color[0] = 0;
color[1] = 0;
color[2] = 255*c;
}
else if (c < 2)
{
color[0] = 0;
color[1] = 255*(c-1);
color[2] = 255;
}
else
{
color[0] = 255*(c-2);
color[1] = 255;
color[2] = 255;
}
}
finalResultBuffer[(iY*iXmax)+iX][0] = color[0];
finalResultBuffer[(iY*iXmax)+iX][1] = color[1];
finalResultBuffer[(iY*iXmax)+iX][2] = color[2];
}
}
free(finalResultBuffer);
} else if (my_rank == processors - 1) {
resultBuffer = malloc((linePerProcess + remainingLines) * iXmax * sizeof(color));
for(iY = processMinY; iY < lastProcessMaxY; iY++) {
Cy = CyMin + (iY * PixelHeight);
if (fabs(Cy) < (PixelHeight / 2))
{
Cy = 0.0; /* Main antenna */
}
for(iX = 0; iX < iXmax; iX++)
{
Cx = CxMin + (iX * PixelWidth);
/* initial value of orbit = critical point Z= 0 */
Zx = 0.0;
Zy = 0.0;
Zx2 = Zx * Zx;
Zy2 = Zy * Zy;
/* */
for(Iteration = 0; Iteration < IterationMax && ((Zx2 + Zy2) < ER2); Iteration++)
{
Zy = (2 * Zx * Zy) + Cy;
Zx = Zx2 - Zy2 + Cx;
Zx2 = Zx * Zx;
Zy2 = Zy * Zy;
};
/* compute pixel color (24 bit = 3 bytes) */
if (Iteration == IterationMax)
{
// Point within the set. Mark it as black
color[0] = 0;
color[1] = 0;
color[2] = 0;
}
else
{
// Point outside the set. Mark it as white
double c = 3*log((double)Iteration)/log((double)(IterationMax) - 1.0);
if (c < 1)
{
color[0] = 0;
color[1] = 0;
color[2] = 255*c;
}
else if (c < 2)
{
color[0] = 0;
color[1] = 255*(c-1);
color[2] = 255;
}
else
{
color[0] = 255*(c-2);
color[1] = 255;
color[2] = 255;
}
}
resultBuffer[(iY*iXmax)+iX][0] = color[0];
resultBuffer[(iY*iXmax)+iX][1] = color[1];
resultBuffer[(iY*iXmax)+iX][2] = color[2];
}
}
free(resultBuffer);
} else {
resultBufferTwo = malloc(linePerProcess * iXmax * sizeof(color));
for(iY = processMinY; iY < processMaxY; iY++) {
Cy = CyMin + (iY * PixelHeight);
if (fabs(Cy) < (PixelHeight / 2))
{
Cy = 0.0; /* Main antenna */
}
for(iX = 0; iX < iXmax; iX++)
{
Cx = CxMin + (iX * PixelWidth);
/* initial value of orbit = critical point Z= 0 */
Zx = 0.0;
Zy = 0.0;
Zx2 = Zx * Zx;
Zy2 = Zy * Zy;
/* */
for(Iteration = 0; Iteration < IterationMax && ((Zx2 + Zy2) < ER2); Iteration++)
{
Zy = (2 * Zx * Zy) + Cy;
Zx = Zx2 - Zy2 + Cx;
Zx2 = Zx * Zx;
Zy2 = Zy * Zy;
};
/* compute pixel color (24 bit = 3 bytes) */
if (Iteration == IterationMax)
{
// Point within the set. Mark it as black
color[0] = 0;
color[1] = 0;
color[2] = 0;
}
else
{
// Point outside the set. Mark it as white
double c = 3*log((double)Iteration)/log((double)(IterationMax) - 1.0);
if (c < 1)
{
color[0] = 0;
color[1] = 0;
color[2] = 255*c;
}
else if (c < 2)
{
color[0] = 0;
color[1] = 255*(c-1);
color[2] = 255;
}
else
{
color[0] = 255*(c-2);
color[1] = 255;
color[2] = 255;
}
}
resultBufferTwo[(iY*iXmax)+iX][0] = color[0];
resultBufferTwo[(iY*iXmax)+iX][1] = color[1];
resultBufferTwo[(iY*iXmax)+iX][2] = color[2];
}
}
free(resultBufferTwo);
}
MPI_Finalize();
return 0;
}
当我 运行 这样做时，我得到分段错误 (11) 以及中止陷阱 (6)，这仅发生在排名 1 及以上的进程中。 0号进程没有问题，谁能帮我理解这里的问题？
对于高于 0 的排名，您对结果缓冲区的索引是错误的。
缓冲区始终从索引 0 开始，无论代码在哪个级别 运行。但是，对于高于 0 的排名，您将从 processMinY * iXmax 开始索引。然后这会覆盖任意内存，这可能会导致段错误。
要解决此问题，您应该在计算缓冲区中的索引时从 iY 中减去 processMinY，例如resultBufferTwo[((iY-processMinY)*iXmax)+iX][0] = color[0];.
我还建议您尽可能统一不同级别的代码。现在，您的大部分代码都是重复的，尽管实际上唯一不同的行是 malloc 调用和写入缓冲区。这使得代码难以遵循。您应该为传递给 malloc 的大小和缓冲区创建一个变量，然后您将在循环中使用它。这样只有这两个变量的初始化需要在等级之间有所不同。