More Quiz Questions Parallel Programming MPI Collective routines

More Quiz Questions Parallel Programming MPI Collective routines ITCS 4/5145 Parallel Programming, UNC-Charlotte, B. Wilkinson, 2013, Quiz. Questions 2 b. ppt Sept 23, 2013

What does the MPI routine MPI_Barrier() do? a) Waits for all messages to be sent and received. b) Will cause processes to wait for all processes within the specific communicator to call the routine. Then all processes send a message to the master process and continue. c) Makes a process to execute slower to allow debugging d) Waits for a specified amount of time. e) Will cause processes after calling MPI_Barrier() to wait for all processes within the specific communicator to call the routine. Then all processes are released and are allowed to continue.

The MPI collective routines are said to have the same semantics as using point-point MPI_send() and MPI_recv() routines separately in respect to when they return. If so, when does the MPI_Bcast() routine return? a) They all return when the message has been received by all processes. b) The root returns when its local actions are complete but the message may have not been received and each destination process returns when it receives the message. c) The root returns when its local actions are complete but the message may have not been received and all destination processes return when all have received the message. d) Never e) Immediately even before local actions arte complete. f) None or the other answers.

Name one MPI routine that does not have a named communicator as a parameter (argument). a) b) c) d) e) f) MPI_Send() MPI_Bcast() MPI_Init() MPI_Barrier() None - they all have a named communicator as a parameter. none of the other answers