Parallel Programming

1. Parallel Programming AMANO, Hideharu

2. ParallelProgramming • Message Passing • PVM • MPI • Shared Memory • POSIX thread • OpenMP • CUDA/OpenCL • Automatic Parallelizing Compilers

3. Send Receive Send Receive Message passing（Blocking: randezvous）

4. Send Receive Send Receive Message passing（with buffer）

5. Send Receive Message passing（non-blocking） Other Job

6. PVM (ParallelVirtualMachine) • A buffer is provided for a sender. • Both blocking/non-blocking receive is provided. • Barrier synchronization

7. MPI(MessagePassingInterface) • Superset of the PVM for 1 to 1 communication. • Group communication • Various communication is supported. • Error check with communication tag.

8. Programming style using MPI • SPMD (Single Program Multiple Data Streams) • Multiple processes executes the same program. • Independent processing is done based on the process number. • Program execution using MPI • Specified number of processes are generated. • They are distributed to each node of the NORA machine or PC cluster.

9. Communication methods • Point-to-Point communication • A sender and a receiver executes function for sending and receiving. • Each function must be strictly matched. • Collective communication • Communication between multiple processes. • The same function is executed by multiple processes. • Can be replaced with a sequence of Point-to-Point communication, but sometimes effective.

10. Fundamental MPI functions • Most programs can be described using six fundamental functions • MPI_Init()… MPI Initialization • MPI_Comm_rank()… Get the process # • MPI_Comm_size()… Get the total process # • MPI_Send()… Message send • MPI_Recv()… Message receive • MPI_Finalize()… MPI termination

11. Other MPI functions • Functions for measurement • MPI_Barrier() … barrier synchronization • MPI_Wtime() … get the clock time • Non-blocking function • Consisting of communication request and check • Other calculation can be executed during waiting.

12. An Example 1: #include <stdio.h> 2: #include <mpi.h> 3: 4: #define MSIZE 64 5: 6: int main(int argc, char **argv) 7: { 8: char msg[MSIZE]; 9: int pid, nprocs, i; 10:MPI_Status status; 11: 12:MPI_Init(&argc, &argv); 13: MPI_Comm_rank(MPI_COMM_WORLD, &pid); 14:MPI_Comm_size(MPI_COMM_WORLD, &nprocs); 15: 16: if (pid == 0) { 17: for (i = 1; i < nprocs; i++) { 18: MPI_Recv(msg, MSIZE, MPI_CHAR, i, 0, MPI_COMM_WORLD, &status); 19:fputs(msg, stdout); 20:} 21:} 22:else { 23: sprintf(msg, "Hello, world! (from process #%d)\n", pid); 24:MPI_Send(msg, MSIZE, MPI_CHAR, 0, 0, MPI_COMM_WORLD); 25: } 26: 27: MPI_Finalize(); 28: 29: return 0; 30: }

13. Initialize and Terminate int MPI_Init( int *argc, /* pointer to argc */ char ***argv /* pointer to argv */ ); mpi_init(ierr) integer ierr! return code The attributes from command line must be passed directly to argc and argv. int MPI_Finalize(); mpi_finalize(ierr) integer ierr ! return code

14. Commincator functions It returns the rank (process ID) in the communicator comm. int MPI_Comm_rank( MPI_Comm comm, /* communicator */ int *rank /* process ID (output) */ ); mpi_comm_rank(comm, rank, ierr) integer comm, rank integer ierr ! return code It returns the total number of processes in the communicator comm. int MPI_Comm_size( MPI_Comm comm, /* communicator */ int *size /* number of process (output) */ ); mpi_comm_size(comm, size, ierr) integer comm, size integer ierr! return code • Communicators are used for sharing commnication space among a subset of processes. MPI_COMM_WORLD is pre-defined one for all processes.

15. MPI_Send It sends data to process “dest”. int MPI_Send( void *buf, /* send buffer */ int count, /* # of elements to send */ MPI_Datatype datatype, /* datatype of elements */ int dest, /* destination (receiver) process ID */ int tag, /* tag */ MPI_Comm comm /* communicator */ ); mpi_send(buf, count, datatype, dest, tag, comm, ierr) <type> buf(*) integer count, datatype, dest, tag, comm integer ierr ! return code • Tags are used for identification of message.

16. MPI_Recv int MPI_Recv( void *buf, /* receiver buffer */ int count, /* # of elements to receive */ MPI_Datatype datatype, /* datatype of elements */ int source, /* source (sender) process ID */ int tag, /* tag */ MPI_Comm comm, /* communicator */ MPI_Status /* status (output) */ ); mpi_recv(buf, count, datatype, source, tag, comm, status, ierr) <type> buf(*) integer count, datatype, source, tag, comm, status(mpi_status_size) integer ierr ! return code • The same tag as the sender’s one must be passed to MPI_Recv. • Set the pointers to a variable MPI_Status. It is a structure with three members: MPI_SOURCE, MPI_TAG and MPI_ERROR, which stores process ID of the sender, tag and error code.

17. datatype and count • The size of the message is identified with count and datatype. • MPI_CHAR char • MPI_INT int • MPI_FLOAT float • MPI_DOUBLE double … etc.

18. Compile and Execution % icc –o hello hello.c -lmpi % mpirun –np 8 ./hello Hello, world! (from process #1) Hello, world! (from process #2) Hello, world! (from process #3) Hello, world! (from process #4) Hello, world! (from process #5) Hello, world! (from process #6) Hello, world! (from process #7)

19. POSIX Thread • Standard API on Linux for controlling threads. • Portable Operating System Interface • Thread handling • pthread_create(); pthread_join(); pthread_exec(); • Synchronization • mutex • pthread_mutex_lock(); pthread_mutex_trylock(); pthread_mutex_unlock(); • Condition variable: Semaphore • pthread_cond_signal(); pthread_cond_wait(); etc.

20. OpenMP #include <stdio.h> int main() { pragma omp parallel { int tid, npes; tid = omp_get_thread_num(); npes = omp_get_num_threads(); printf(“Hello World from %d of %d\n”, tid, npes) } return 0; } • Multiple threads are generated by using pragma. • Variables declared globally can be shared.

21. Convenient pragma for parallel execution #pragma omp parallel { #pragma omp for for (i=0; i<1000; i++){ c[i] = a[i] + b[i]; } } • The assignment between i and thread is automatically adjusted in order that the load of each thread becomes even.

22. CUDA/OpenCL • CUDA is developed for GPGPU programming. • SPMD(Single Program Multiple Data) • 3-D management of threads • 32 threads are managed with a Warp • SIMD programming • Architecture dependent memory model • OpenCL is standard language for heterogeneous accelerators.

23. Heterogeneous Programming with CUDA Host Serial Code Device Parallel Kernel KernelA(args); … Host Serial Code Device … Parallel Kernel KernelB(args);

24. Threads and thread blocks each thread executes the same code Kernel = grid of thread blocks Thread Block N-1 Thread Block 1 Thread Block 0 threadID 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 … float x = input[threadID]; float y=func(x); output[threadID]=y; … … float x = input[threadID]; float y=func(x); output[threadID]=y; … … float x = input[threadID]; float y=func(x); output[threadID]=y; … … Threads in the same block may synchronize with barriers. _syncthreads(); Thread blocks cannot synchronize -> Execution is depending on machines.

25. Memory Hierarchy Thread Between host memory cudaMemcpy(); is used. Per-thread Local memory Block Per-block Shared Memory Kernel 0 Sequential Kernels … Per-device Global Memory Kernel 1 …

26. CUDA extensions • Declaration specifiers : • _global_ void kernelFunc(…); // kernel function, runs on device • _device_ int GlobalVar; //variable in device memory • _shared_ int sharedVar; //variable in per-block shared memory • Extend function invocation syntax for paralell kernel launch • KernelFunc<<<dimGrid, dimBlock>>> // launch dimGrid blocks with dimBlock threads each • Special variables for thread identification in kernels • dim3 threadIDx; dim3 blockIdx; dim3 block Dim; dim3 gridDim; • Barrier Synchronization between threads • _syncthreads();

27. CUDA runtime • Device menagement: • cudaGetDeviceCount(), cudaGetDeviceProperties(); • Device memory management: • cudaMalloc(), cudaFree(),cudaMemcpy() • Graphics interoperability: • cudaGLMapBufferObject(), cudaD3D9MapResources() • Texture management: • cudaBindTexture(), cudaBindTextureToArray()

28. Example: Increment Array Elements _global_ void increment_gpu(float *a, float b, int N) { int idx=blockidx.x*blockDim.x+threadIdx.x; if(idx<N) a[idx]=a[idx]+b; } void main() { … dim3 dimBlock(blocksize); dim3 dimGrid(ceil(N/(float)blocksize)); increment_gpu<<dimGrid,dimBlock>>(a,b,N); } void increment_cpu(float *a, float b, int N) { for(int idx=0;idx<N;idx++) a[idx]=a[idx]+b; } void main() { … increment_cpu(a,b,N); }

29. Example: Increment Array Elements Let’s assume N=16, blockDim=4 blockIdx.x=3 blockDim.x=4 threadIdx.x=0,1,2,3 idx=12,13,14,15 blockIdx.x=2 blockDim.x=4 threadIdx.x=0,1,2,3 idx=8,9,10,11 blockIdx.x=1 blockDim.x=4 threadIdx.x=0,1,2,3 idx=4,5,6,7 blockIdx.x=0 blockDim.x=4 threadIdx.x=0,1,2,3 idx=0,1,2,3 int idx = blockDim.x * blockId.x + threadldx.x; will map from local index threadIdx to global index. blockDim should be >= 32 in real code! Using more number of blocks hides the memory latency in GPU.

30. Host code // allocate host memory unsigned int numBytes = N*sizeof(float); float * h_A = (float *) malloc(numBytes); // allocate device memory float* d_A=0; cudaMalloc((void**)&d_a, numBytes); // copy data from host to device cudaMemcpy(d_A, h_A, numBytes, cudaMemcpyHostToDevice); // execute the kernel increment_cpu<<<N/blockSize, blockSize>>>(d_A,b); // copy data from device back to host cudaMemcpy(h_A, d_A, numBytes, cudaMemcpyDeviceToHost); // free device memory cudaFree(d_A);

31. GeForce GTX280 240 cores Host Input Assembler Thread Execution Manager Thread Processors Thread Processors Thread Processors Thread Processors Thread Processors … PBSM PBSM PBSM PBSM PBSM PBSM PBSM PBSM PBSM PBSM Load/Store Global Memory

32. Hardware Implementation: Execution Model Host Device Kernel 1 Grid1 Block (0,0) Block (1,0) Block (2,0) Block (0,1) Block (1,1) Block (2,1) Kernel 2 Grid2 Block (0,0) Block (1,0) Block (2,0) Block (1,1) Thread (0,0) Warp 0 … Thread (31,0) Thread (32,0) Warp 1 … Thread (63,0) Block (0,1) Block (1,1) Block (2,1) Thread (0,1) Warp 2 … Thread (31,1) Thread (32,1) Warp 3 … Thread (63,1) Thread (0,2) Warp 4 … Thread (31,2) Thread (32,2) … Warp 5 Thread (63,2) A multiprocessor executes the same instruction on a group of threads called a warp. Warp size= the number of threads in a warp

33. Automatic parallelizing Compilers • Automatically translating a code for uniprocessors into multiprocessors. • Loop level parallelism is main target of parallelizing. • Fortran codes have been main targets • No pointers • The array structure is simple • Recently, restricted C becomes a target language • Oscar Compiler (Waseda Univ.), COINS

34. Shared memory model vs．Message passing model • Benefits • Distributed OS is easy to implement. • Automatic parallelize compiler. • POSIX thread, OpenMP • Message passing • Formal verification is easy (Blocking) • No-side effect (Shared variable is side effect itself) • Small cost

35. Parallel Programming Contest • In this lecture, a parallel programming contest will be held. • All students who want to get the credit must join it. • At least, the program must correctly run. • For students with good achievement, the credit will be given unconditionally.