Distributed Systems CS 15 440 Programming Models Part

Distributed Systems CS 15 -440 Programming Models- Part I Lecture 15, Nov 2, 2015 Mohammad Hammoud 1

Today… § Last Session: § Replication and Consistency (Management and Protocols) § Today’s Session: § Programming Models – Part I § Announcements: § P 3 will be posted by tonight. It is due on Monday, Nov 16 th by midnight § In the next recitation, we will practice on MPI 2

Objectives Discussion on Programming Models Why parallelism? Parallel computer architectures Traditional models of parallel programming Types of Parallel Programs Message Passing Interface (MPI) Map. Reduce, Pregel and Graph. Lab

Amdahl’s Law § We parallelize our programs in order to run them faster § How much faster will a parallel program run? § Suppose that the sequential execution of a program takes T 1 time units and the parallel execution on p processors takes Tp time units § Suppose that out of the entire execution of the program, s fraction of it is not parallelizable while 1 -s fraction is parallelizable § Then the speedup (Amdahl’s formula): 4

Amdahl’s Law: An Example § Suppose that 80% of your program can be parallelized and that you use 4 processors to run your parallel version of the program § The speedup you can get according to Amdahl’s law is: § Although you use 4 processors you cannot get a speedup more than 2. 5 times! 5

Real Vs. Actual Cases § Amdahl’s argument is too simplified to be applied to real cases § When we run a parallel program, there a communication overhead and a workload imbalance among processes (in general) Serial Parallel 20 20 80 Serial Parallel 20 Process 1 Process 2 20 20 80 20 Cannot be parallelized Process 3 Process 4 Cannot be parallelized Can be parallelized Process 3 Can be parallelized Process 4 Communication overhead Load Unbalance 1. Parallel Speed-up: An Ideal Case 2. Parallel Speed-up: An Actual Case

Guidelines § In order to efficiently benefit from parallelization, we ought to follow these guidelines: 1. Maximize the fraction of our program that can be parallelized 2. Balance the workload of parallel processes 3. Minimize the time spent for communication 7

Objectives Discussion on Programming Models Why parallelism? Parallel computer architectures Traditional models of parallel programming Types of Parallel Programs Message Passing Interface (MPI) Map. Reduce, Pregel and Graph. Lab

Parallel Computer Architectures § We can categorize the architecture of parallel computers in terms of two aspects: § Whether the memory is physically centralized or distributed § Whether or not the address space is shared Address Space Memory Shared Individual SMP (Symmetric Multiprocessor) Centralized UMA – SMP (Symmetric Multiprocessor) N/A Distributed NUMA (Non-Uniform Memory Access) MPP (Massively Parallel Processors) 9

Symmetric Multiprocessor § Symmetric Multiprocessor (SMP) architecture uses shared system resources that can be accessed equally from all processors Processor Cache Bus or Crossbar Switch Memory I/O § Usually, a single OS controls the SMP machine and it schedules processes and threads on processors so that the load is balanced 10

Massively Parallel Processors § Massively Parallel Processors (MPP) architecture consists of nodes with each having its own processor, memory and I/O subsystem Interconnection Network Processor Cache Bus Bus Memory I/O § An independent OS runs at each node 11 Memory I/O

Non-Uniform Memory Access § Non-Uniform Memory Access (NUMA) architecture machines are built on a similar hardware model as MPP § NUMA typically provides a shared address space to applications using a hardware/software directory-based coherence protocol § The memory latency varies according to whether the memory is accessed directly (locally) or through the interconnect (remotely) § As in an SMP machine, usually a single OS controls the whole system 12

Objectives Discussion on Programming Models Why parallelizing our programs? Parallel computer architectures Traditional Models of parallel programming Types of Parallel Programs Message Passing Interface (MPI) Map. Reduce, Pregel and Graph. Lab

Models of Parallel Programming § What is a parallel programming model? § A programming model is an abstraction provided by the hardware to programmers § It determines how easily programmers can specify their algorithms into parallel units of computations (i. e. , tasks) that the hardware understands § It determines how efficiently parallel tasks can be executed on the hardware § Main Goal: utilize all the processors of the underlying architecture (e. g. , SMP, MPP, NUMA, and Distributed Systems) and minimize the elapsed time of your program 14

Traditional Parallel Programming Models Shared Memory Message Passing 15

Shared Memory Model § In the shared memory programming model, the abstraction provided implies that parallel tasks can access any location of the memory § Accordingly, parallel tasks can communicate through reading and writing common memory locations § This is similar to threads in a single process (in traditional OSs), which share a single address space § Multi-threaded programs (e. g. , Open. MP programs) use the shared memory programming model 16

Shared Memory Model Single Thread S 1 Multi-Thread Time Si = Serial Pj = Parallel S 1 Spawn P 1 P 3 P 1 P 2 P 3 S 2 Shared Address Space P 2 P 3 Join P 4 S 2 Process 17

Shared Memory Example begin parallel // spawn a child thread private int start_iter, end_iter, i; shared int local_iter=4, sum=0; shared double sum=0. 0, a[], b[], c[]; shared lock_type mylock; for (i=0; i<8; i++) a[i] = b[i] + c[i]; sum = 0; for (i=0; i<8; i++) if (a[i] > 0) sum = sum + a[i]; Print sum; Sequential start_iter = getid() * local_iter; end_iter = start_iter + local_iter; for (i=start_iter; i<end_iter; i++) a[i] = b[i] + c[i]; barrier; for (i=start_iter; i<end_iter; i++) if (a[i] > 0) { lock(mylock); sum = sum + a[i]; unlock(mylock); } barrier; // necessary end parallel // kill the child thread Print sum; Parallel

Traditional Parallel Programming Models Shared Memory Message Passing 19

Message Passing Model § In message passing, parallel tasks have their own local memories § One task cannot access another task’s memory § Hence, tasks have to rely on explicit messages to communicate § This is similar to the abstraction of processes in traditional OSs, which do not share address spaces § MPI programs use the message passing programming model 20

Message Passing Model Message Passing Single Thread S 1 Time S = Serial P = Parallel S 1 S 1 P 1 P 2 S 2 S 2 P 3 P 4 Process 0 Process 1 Process 2 Process 3 S 2 Node 1 Node 2 Node 3 Node 4 Data transmission over the Network Process 21

Message Passing Example id = getpid(); local_iter = 4; start_iter = id * local_iter; end_iter = start_iter + local_iter; for (i=0; i<8; i++) a[i] = b[i] + c[i]; sum = 0; for (i=0; i<8; i++) if (a[i] > 0) sum = sum + a[i]; Print sum; Sequential if (id == 0) send_msg (P 1, b[4. . 7], c[4. . 7]); else recv_msg (P 0, b[4. . 7], c[4. . 7]); for (i=start_iter; i<end_iter; i++) a[i] = b[i] + c[i]; local_sum = 0; for (i=start_iter; i<end_iter; i++) if (a[i] > 0) local_sum = local_sum + a[i]; if (id == 0) { recv_msg (P 1, &local_sum 1); sum = local_sum + local_sum 1; Print sum; } else send_msg (P 0, local_sum); Parallel

Shared Memory Vs. Message Passing § Comparison between the shared memory and message passing programming models: Aspect Shared Memory Message Passing Communication Implicit (via loads/stores) Explicit Messages Synchronization Explicit Implicit (Via Messages) Hardware Support Typically Required None Development Effort Lower Higher Tuning Effort Higher Lower 23

Objectives Discussion on Programming Models Why parallelizing our programs? Parallel computer architectures Traditional Models of parallel programming Types of Parallel Program Message Passing Interface (MPI) Map. Reduce, Pregel and Graph. Lab

SPMD and MPMD § When we run multiple processes with message-passing, there are further categorizations regarding how many different programs are cooperating in parallel execution § We distinguish between two models: 1. Single Program Multiple Data (SPMD) model 2. Multiple Programs Multiple Data (MPMD) model 25

SPMD § In the SPMD model, there is only one program and each process uses the same executable working on different sets of data a. out Node 1 Node 2 Node 3 26

MPMD § The MPMD model uses different programs for different processes, but the processes collaborate to solve the same problem § MPMD has two styles, the master/worker and the coupled analysis a. out b. out c. out a. out= Structural Analysis, b. out = fluid analysis and c. out = thermal analysis Example Node 1 Node 2 Node 3 1. MPMD: Master/Slave Node 1 Node 2 Node 3 2. MPMD: Coupled Analysis

An Example A Sequential Program 1. 2. 3. 4. Read array a() from the input file Set is=1 and ie=6 //is = index start and ie = index end Process from a(is) to a(ie) Write array a() to the output file a a a is 1 2 3 4 5 ie 6 Colored shapes indicate the initial values of the elements Black shapes indicate the values after they are processed

An Example Under which category does this program lie? Process 0 Process 1 Process 2 1. Read array a() from the input file 2. Get my rank 3. If rank==0 then is=1, ie=2 If rank==1 then is=3, ie=4 If rank==2 then is=5, ie=6 4. Process from a(is) to a(ie) 5. Gather the results to process 0 6. If rank==0 then write array a() to the output file is ie 1 2 3 4 5 6 a a a

Concluding Remarks § To summarize, keep the following 3 points in mind: § The purpose of parallelization is to reduce the time spent on computation– A caveat, however, is that communication might increase as the cluster size is scaled up § Ideally, the parallel program is p times faster than the sequential program, where p is the number of processes involved in the parallel execution, but this is not always achievable § Message-passing is the tool to consolidate what parallelization has separated. It should not be regarded as the parallelization itself 30

Objectives Discussion on Programming Models Why parallelizing our programs? Parallel computer architectures Traditional Models of parallel programming Types of Parallel Programs Message Passing Interface (MPI) Map. Reduce, Pregel and Graph. Lab

Message Passing Interface § In this part, the following concepts of MPI will be described: § Basics § Point-to-point communication § Collective communication 32

Message Passing Interface § In this part, the following concepts of MPI will be described: § Basics § Point-to-point communication § Collective communication 33

What is MPI? § The Message Passing Interface (MPI) is a message passing library standard for writing message passing programs § The goal of MPI is to establish a portable, efficient, and flexible standard for message passing § By itself, MPI is NOT a library - but rather the specification of what such a library should be § MPI is not an IEEE or ISO standard, but has in fact, become the industry standard for writing message passing programs on HPC platforms 34

Reasons for using MPI Reason Standardization Portability Performance Opportunities Functionality Availability Description MPI is the only message passing library which can be considered a standard. It is supported on virtually all HPC platforms There is no need to modify your source code when you port your application to a different platform that supports the MPI standard Vendor implementations should be able to exploit native hardware features to optimize performance Over 115 routines are defined A variety of implementations are available, both vendor and public domain 35

The Programming Model § MPI is an example of a message passing programming model § MPI is now used on just about any common parallel architecture including MPP, SMP clusters, workstation clusters and heterogeneous networks § With MPI the programmer is responsible for correctly identifying parallelism and implementing parallel algorithms using MPI constructs 36

Communicators and Groups § MPI uses objects called communicators/groups to define which collection of processes may communicate with each other to solve a certain problem § Most MPI routines require you to specify a communicator as an argument § The communicator MPI_COMM_WORLD is often used in calling communication subroutines § MPI_COMM_WORLD is the predefined communicator that includes all of your MPI processes 37

Ranks § Within a communicator, every process has its own unique, integer identifier referred to as rank, assigned by the system when the process initializes § A rank is sometimes called a task ID. Ranks are contiguous and begin at zero § Ranks are used by the programmer to specify the source and destination of messages § Ranks are often also used conditionally by the application to control program execution (e. g. , if rank=0 do this / if rank=1 do that) 38

Multiple Communicators § It is possible that a problem consists of several sub-problems where each can be solved concurrently § This type of application is typically found in the category of MPMD coupled analysis § We can create a new communicator for each sub-problem as a subset of an existing communicator § MPI allows you to achieve that by using MPI_COMM_SPLIT 39

Example of Multiple Communicators § Consider a problem with a fluid dynamics part and a structural analysis part, where each part can be computed in parallel MPI_COMM_WORLD Comm_Fluid Comm_Struct Rank=0 Rank=1 Rank=4 Rank=5 Rank=2 Rank=3 Rank=6 Rank=7 Ranks within MPI_COMM_WORLD are printed in red Ranks within Comm_Fluid are printed with green Ranks within Comm_Struct are printed with blue

Next Class Discussion on Programming Models Why parallelizing our programs? Parallel computer architectures Traditional Models of parallel programming Types of Parallel Programs Message Passing Interface (MPI) Map. Reduce, Pregel and Graph. Lab Programming Models- Part II