/* StarPU --- Runtime system for heterogeneous multicore architectures. * * Copyright (C) 2010-2020 Université de Bordeaux, CNRS (LaBRI UMR 5800), Inria * * StarPU is free software; you can redistribute it and/or modify * it under the terms of the GNU Lesser General Public License as published by * the Free Software Foundation; either version 2.1 of the License, or (at * your option) any later version. * * StarPU is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. * * See the GNU Lesser General Public License in COPYING.LGPL for more details. */ #include #include #include #include #include #include "../helper.h" #include /* * Mix submitting tasks and asynchronously acquiring the corresponding data. */ #define NBUFFERS_DEF 64 #define NITER_DEF 128 #define VECTORSIZE_DEF 1024 static unsigned nbuffers = NBUFFERS_DEF; static unsigned niter = NITER_DEF; static unsigned vectorsize = VECTORSIZE_DEF; float *buffer[NBUFFERS_DEF]; starpu_data_handle_t v_handle[NBUFFERS_DEF]; void dummy_codelet(void *descr[], void *_args) { (void)descr; (void)_args; } static struct starpu_codelet cl = { .modes = { STARPU_RW }, .cpu_funcs = {dummy_codelet}, #ifdef STARPU_USE_CUDA .cuda_funcs = {dummy_codelet}, #endif #ifdef STARPU_USE_OPENCL .opencl_funcs = {dummy_codelet}, #endif .cpu_funcs_name = {"dummy_codelet"}, .nbuffers = 1 }; static int use_handle(starpu_data_handle_t handle) { int ret; struct starpu_task *task; task = starpu_task_create(); task->cl = &cl; task->handles[0] = handle; ret = starpu_task_submit(task); return ret; } static starpu_pthread_mutex_t mutex = STARPU_PTHREAD_MUTEX_INITIALIZER; static starpu_pthread_cond_t cond = STARPU_PTHREAD_COND_INITIALIZER; static unsigned n_synced_buffers; static void callback_sync_data(void *arg) { (void)arg; STARPU_PTHREAD_MUTEX_LOCK(&mutex); n_synced_buffers++; if (n_synced_buffers == nbuffers) STARPU_PTHREAD_COND_SIGNAL(&cond); STARPU_PTHREAD_MUTEX_UNLOCK(&mutex); } int main(int argc, char **argv) { int ret; #ifdef STARPU_QUICK_CHECK nbuffers /= 4; niter /= 4; vectorsize /= 8; #endif ret = starpu_initialize(NULL, &argc, &argv); if (ret == -ENODEV) return STARPU_TEST_SKIPPED; STARPU_CHECK_RETURN_VALUE(ret, "starpu_init"); /* Allocate all buffers and register them to StarPU */ unsigned b; for (b = 0; b < nbuffers; b++) { ret = starpu_malloc((void **)&buffer[b], vectorsize); STARPU_CHECK_RETURN_VALUE(ret, "starpu_malloc"); starpu_vector_data_register(&v_handle[b], STARPU_MAIN_RAM, (uintptr_t)buffer[b], vectorsize, sizeof(char)); starpu_data_set_sequential_consistency_flag(v_handle[b], 0); } unsigned iter; for (iter = 0; iter < niter; iter++) { /* Use the buffers on the different workers so that it may not * be in main memory anymore */ for (b = 0; b < nbuffers; b++) { ret = use_handle(v_handle[b]); if (ret == -ENODEV) goto enodev; STARPU_CHECK_RETURN_VALUE(ret, "starpu_task_submit"); } ret = starpu_task_wait_for_all(); STARPU_CHECK_RETURN_VALUE(ret, "starpu_task_wait_for_all"); STARPU_PTHREAD_MUTEX_LOCK(&mutex); n_synced_buffers = 0; STARPU_PTHREAD_MUTEX_UNLOCK(&mutex); /* Grab the different pieces of data into main memory */ for (b = 0; b < nbuffers; b++) { ret = starpu_data_acquire_cb(v_handle[b], STARPU_RW, callback_sync_data, NULL); STARPU_CHECK_RETURN_VALUE(ret, "starpu_data_acquire_cb"); } /* Wait for all buffers to be available */ STARPU_PTHREAD_MUTEX_LOCK(&mutex); while (n_synced_buffers != nbuffers) STARPU_PTHREAD_COND_WAIT(&cond, &mutex); STARPU_PTHREAD_MUTEX_UNLOCK(&mutex); /* Release them */ for (b = 0; b < nbuffers; b++) starpu_data_release(v_handle[b]); } /* do some cleanup */ for (b = 0; b < nbuffers; b++) { starpu_data_unregister(v_handle[b]); starpu_free(buffer[b]); } starpu_shutdown(); return EXIT_SUCCESS; enodev: fprintf(stderr, "WARNING: No one can execute this task\n"); /* yes, we do not perform the computation but we did detect that no one * could perform the kernel, so this is not an error from StarPU */ starpu_shutdown(); return STARPU_TEST_SKIPPED; }