TI工程师，您好
采用主从核的模式做多核分区计算矩阵乘法，0核为主核1~7核为从核，结果分区写入内存，代码如下：
bool stepCal(Uint32 COREID){
int i = 0;
int inputIndex2 = 0, outputIndex3 = 0;
int procRowNumber = 0;
double* s_step_loc = malloc(2 * (CHANNEL + 1) * sizeof(double));
double* trans_JR_loc = malloc(2 * (CHANNEL + 1) * sizeof(double));
double* inv_M_loc = malloc((CHANNEL + 1) * (CHANNEL + 1) * sizeof(double));
memset(s_step_loc, 0, 2 * (CHANNEL + 1) * sizeof(double));
memset(trans_JR_loc, 0, 2 * (CHANNEL + 1) * sizeof(double));
memset(inv_M_loc, 0, (CHANNEL + 1) * (CHANNEL + 1) * sizeof(double));
memcpy(trans_JR_loc, trans_JR, 2 * (CHANNEL + 1) * sizeof(double));
memcpy(inv_M_loc, inv_M, (CHANNEL + 1) * (CHANNEL + 1) * sizeof(double));
/*按核编号，分配处理地址范围（工作量）*/
if(COREID == 7){
inputIndex2 = (CHANNEL + 1 - 6 * (8 - COREID)) * (CHANNEL + 1);
outputIndex3 = (CHANNEL + 1 - 6 * (8 - COREID)) * 2;
procRowNumber = 6;
}
else if(COREID == 5 || COREID == 6) {
inputIndex2 = (CHANNEL + 1 - 6 * (8 - COREID)) * (CHANNEL + 1);
outputIndex3 = (CHANNEL + 1 - 6 * (8 - COREID)) * 2;
procRowNumber = 6;
}
else {
inputIndex2 = COREID * 4 * (CHANNEL + 1);
outputIndex3 = COREID * 4 * 2;
procRowNumber = 4;
}
/*矩阵求逆结束后，多核计算步长,由于inv_M矩阵的对称性，不需要转置*/
DSPF_dp_mat_mul_gemm((double*)trans_JR_loc, 1, 2, (CHANNEL + 1), (double*)(inv_M_loc + inputIndex2), procRowNumber, (double*)(s_step_loc + outputIndex3));
for(i = outputIndex3 / 2; i < outputIndex3 / 2 + procRowNumber; i++){
*(s_step + i) = *(s_step_loc + 2 * i);
}
Cache_wbInvAll();
free(s_step_loc);
free(trans_JR_loc);
free(inv_M_loc);
return true;
}
其中，trans_JR，inv_M，s_step 是全局指针变量
函数运行后，每个核用Memory Browser看到的结果一致，如下：
其中，s_step是double型指针，应该有37个非0数据。但是从Memory Browser看到的结果，每个核看到的前4个数据都是0（0核负责前4个核）。请问为什么会发生这个错误
以及，对于1~7核（从核），取消勾选L1D Cache或 L2Cache，Memory Browser看到的内容如下：
请问，Memory Browser看到的是哪一块存储空间的内容（L1D L2 MSMC）？
