Searched refs:executions (Results 1 – 6 of 6) sorted by relevance
40 std::atomic_uint executions = 0; in TEST_F() local47 [&executions](const android::nn::telemetry::DiagnosticExecutionInfo*) { in TEST_F()48 executions++; in TEST_F()72 ASSERT_EQ(executions, 1u); in TEST_F()
3322 std::vector<ANeuralNetworksExecution*> executions = {nullptr, execution}; in TEST_F() local3330 for (auto e : executions) { in TEST_F()
366 * overhead between executions. See also {@link ANeuralNetworksMemoryDesc} and423 * once all executions using the compilation have completed.</li></ul></p>499 * executions to do new evaluations of the model.</p>527 * It is however safe to schedule and evaluate multiple computations on different executions530 * completion of executions. If it's important to the application, the533 * executions synchronously within a single thread, or by scheduling all534 * executions asynchronously and using {@link ANeuralNetworksEvent_wait} between555 * of a rapid sequence of executions. It will likely cause overhead if only used564 * lifetime of a rapid sequence of executions. For example, a driver may choose