static int localToRemoteCallback( const void *inputBuffer, void *outputBuffer, unsigned long framesPerBuffer, const PaStreamCallbackTimeInfo* timeInfo, PaStreamCallbackFlags statusFlags, void *userData )
{
callbackData *data = (callbackData*)userData;
SAMPLE *out = (SAMPLE*)outputBuffer;
SAMPLE *in = (SAMPLE*)inputBuffer;
(void) timeInfo; /* Prevent unused variable warnings. */
(void) statusFlags;
(void) userData;
SAMPLE downsampled[320];
unsigned char compressed[6];
vector<bool> dataBits;
vector<bool> transcodedBits;
if( inputBuffer == NULL )
{
for( int i=0; i<framesPerBuffer; i++ ) { *out++ = 0; }
}
else
{
// downsample 1280 samples to 320 samples by taking every 4th sample
Downsample::Perform(in, downsampled, 1280, 4);
// compress 320 samples to 6 bytes (48 bits)
codec2_encode(data->codec2, compressed, downsampled);
// add the bufferCounter to the dataBits
bitset<5> bC ( (unsigned long) data->bufferCounter );
for(int i=0;i<5;i++) { dataBits.push_back(bC[i]);}
// increment the Buffer Counter used for synchronization
data->bufferCounter = (data->bufferCounter + 1) % 32;
// convert and push unsigned chars onto dataBits
Convert::UnsignedCharToBits(compressed, dataBits, 6);
// record bits for testing
data->recordedBits.push_back(dataBits);
// transcode via IncDec algorithm.
Transcode::Perform(dataBits, transcodedBits);
// prep the transmitter
data->transmitter->setBits(transcodedBits);
// transmit
data->transmitter->emitSound(out);
}
return paContinue;
}
int main(int argc, char** argv)
{
if(argc != 4){
std::cout << "Error! Three arguments m, n and p are needed!" << std::endl;
return 1;
}
int m = atoi(argv[1]);
int n = atoi(argv[2]);
int p = atoi(argv[3]);
double elapseTime, elapseTime_single;
int myrank, numprocs;
MPI_Status status;
MPI_Init(&argc, &argv); // 并行开始
//std::cout << "Start to computing..." << std::endl;
MPI_Comm_size(MPI_COMM_WORLD, &numprocs); // 获取进程数
MPI_Comm_rank(MPI_COMM_WORLD, &myrank); // 获取本进城ID号
if(numprocs < 2){
std::cout << "Error! There must be more than two processors." << std::endl;
return 1;
}
int bm = m / numprocs;
int bn = n / numprocs;
SCMatrix A(1), B(1), C(1), C_true(1);
SCMatrix bA(bm,p), bB_send(bn,p), bB_recv(bn,p), bC(bm, bn), bC_send(bm, n);
if(myrank == 0){
SCMatrix A1(m,p), B1(n,p), C1(m,n), C_true1(m,n);
A = A1; B = B1; C = C1; C_true = C_true1;
A.init_with_RV();
B.init_with_RV();
}
MPI_Barrier(MPI_COMM_WORLD);
//A.print(); B.print();
// start to timing
clock_t start_time = clock();
//clock_t t_temp_start = clock();
MPI_Scatter(&A, bm * p, MPI_DOUBLE, &bA, bm * p, MPI_DOUBLE, 0, MPI_COMM_WORLD);
MPI_Scatter(&B, bn * p, MPI_DOUBLE, &bB_recv, bn * p, MPI_DOUBLE, 0, MPI_COMM_WORLD);
int sendTo = (myrank + 1) % numprocs;
int recvFrom = (myrank - 1 + numprocs) % numprocs;
int circle = 0;
bB_send = bB_recv;
do{
bC = bA * bB_recv.transpose();
int blocks_col = (myrank - circle + numprocs) % numprocs;
for(int i=0; i<bm; i++){
for(int j=0; j<bn; j++){
bC_send(i, blocks_col*bn + j) = bC(i, j);
}
}
if(myrank % 2 == 0){
bB_send = bB_recv;
MPI_Ssend(&bB_send, bn*p, MPI_DOUBLE, sendTo, circle, MPI_COMM_WORLD);
MPI_Recv(&bB_recv, bn*p, MPI_DOUBLE, recvFrom, circle, MPI_COMM_WORLD, &status);
}else{
MPI_Recv(&bB_recv, bn*p, MPI_DOUBLE, recvFrom, circle, MPI_COMM_WORLD, &status);
MPI_Ssend(&bB_send, bn*p, MPI_DOUBLE, sendTo, circle, MPI_COMM_WORLD);
bB_send = bB_recv;
}
circle++;
}while(circle < numprocs);
MPI_Barrier(MPI_COMM_WORLD);
MPI_Gather(&bC_send, bm * n, MPI_DOUBLE, &C, bm * n, MPI_DOUBLE, 0, MPI_COMM_WORLD);
if(myrank == 0){
int remainAStartId = bm * numprocs;
int remainBStartId = bn * numprocs;
for(int i=remainAStartId; i<m; i++){
for(int j=0; j<n; j++){
double temp=0;
for(int k=0; k<p; k++){
temp += A(i,k) * B(j,k);
}
C(i,j) = temp;
}
}
for(int i=0; i<remainAStartId; i++){
for(int j=remainBStartId; j<n; j++){
double temp = 0;
for(int k=0; k<p; k++){
temp += A(i,k)*B(j,k);
}
C(i,j) = temp;
}
}
}
// end timing
clock_t end_time = clock();
elapseTime = (double)(end_time-start_time) / CLOCKS_PER_SEC;
if(myrank == 0){
//std::cout << "[P_0] m = " << m << ", np = " << numprocs << ", time cost: " << elapseTime << std::endl;
std::cout << "[P_0] Totally cost " << elapseTime << " seconds in parallel procedure." << std::endl;
start_time = clock();
C_true = A * B.transpose();
end_time = clock();
elapseTime_single = double(end_time-start_time) / CLOCKS_PER_SEC;
std::cout << "[P_0] Totally cost " << elapseTime_single << " seconds with one processor." << std::endl;
bool b = (C==C_true);
if(b){
std::cout << "[P_0] Congradulations! The two results are equal." << std::endl;
}else{
std::cout << "[P_0] Bad news! The two results do not match." << std::endl;
A.print();
B.print();
C.print();
C_true.print();
}
}
MPI_Finalize(); // 并行结束
return 0;
}
