Sunday, 16 January 2011

THREADS


In shared memory multiprocessor architectures, such as SMPs, threads can be used to implement parallelism. Historically, hardware vendors have implemented their own proprietary versions of threads, making portability a concern for software developers. For UNIX systems, a standardized C language threads programming interface has been specified by the IEEE POSIX 1003.1c standard. Implementations that adhere to this standard are referred to as POSIX threads, or Pthreads.
The tutorial begins with an introduction to concepts, motivations, and design considerations for using Pthreads. Each of the three major classes of routines in the Pthreads API are then covered: Thread Management, Mutex Variables, and Condition Variables. Example codes are used throughout to demonstrate how to use most of the Pthreads routines needed by a new Pthreads programmer. The tutorial concludes with a discussion of LLNL specifics and how to mix MPI with pthreads. A lab exercise, with numerous example codes (C Language) is also included.
Level/Prerequisites: Ideal for those who are new to parallel programming with threads. A basic understanding of parallel programming in C is assumed. For those who are unfamiliar with Parallel Programming in general, 

What is Parallel Computing?

  • Traditionally, software has been written for serial computation:
    • To be run on a single computer having a single Central Processing Unit (CPU);
    • A problem is broken into a discrete series of instructions.
    • Instructions are executed one after another.
  • Only one instruction may execute at any moment in time.



  • For example:
    Serial computing
  • In the simplest sense, parallel computing is the simultaneous use of multiple compute resources to solve a computational problem:
    • To be run using multiple CPUs
    • A problem is broken into discrete parts that can be solved concurrently
    • Each part is further broken down to a series of instructions
    • Instructions from each part execute simultaneously on different CPUs


    Parallel computing
    For example:
    Parallel computing
  • The compute resources might be:
    • A single computer with multiple processors;
    • An arbitrary number of computers connected by a network;
    • A combination of both.
  • The computational problem should be able to:
    • Be broken apart into discrete pieces of work that can be solved simultaneously;
    • Execute multiple program instructions at any moment in time;
    • Be solved in less time with multiple compute resources than with a single compute resource.
The Universe is Parallel:
  • Parallel computing is an evolution of serial computing that attempts to emulate what has always been the state of affairs in the natural world: many complex, interrelated events happening at the same time, yet within a sequence. For example:
    • Galaxy formation
    • Planetary movement
    • Weather and ocean patterns
    • Tectonic plate drift
    • Rush hour traffic
    • Automobile assembly line
    • Building a jet
    • Ordering a hamburger at the drive through.
    The Real World is Massively Parallel
    The real world is parallel
 Uses for Parallel Computing:
  • Historically, parallel computing has been considered to be "the high end of computing", and has been used to model difficult problems in many areas of science and engineering:
    • Atmosphere, Earth, Environment
    • Physics - applied, nuclear, particle, condensed matter, high pressure, fusion, photonics
    • Bioscience, Biotechnology, Genetics
    • Chemistry, Molecular Sciences
    • Geology, Seismology
    • Mechanical Engineering - from prosthetics to spacecraft
    • Electrical Engineering, Circuit Design, Microelectronics
    • Computer Science, Mathematics
    Computer simulations
  • Today, commercial applications provide an equal or greater driving force in the development of faster computers. These applications require the processing of large amounts of data in sophisticated ways. For example:
    • Databases, data mining
    • Oil exploration
    • Web search engines, web based business services
    • Medical imaging and diagnosis
    • Pharmaceutical design
    • Management of national and multi-national corporations
    • Financial and economic modeling
    • Advanced graphics and virtual reality, particularly in the entertainment industry
    • Networked video and multi-media technologies
    • Collaborative work environments
    Computer simulations



Overview

Why Use Parallel Computing?

 Main Reasons:
  • Save time and/or money: In theory, throwing more resources at a task will shorten its time to completion, with potential cost savings. Parallel computers can be built from cheap, commodity components.
  • Solve larger problems: Many problems are so large and/or complex that it is impractical or impossible to solve them on a single computer, especially given limited computer memory. For example:
    • "Grand Challenge" (en.wikipedia.org/wiki/Grand_Challenge) problems requiring PetaFLOPS and PetaBytes of computing resources.
    • Web search engines/databases processing millions of transactions per second
  • Provide concurrency: A single compute resource can only do one thing at a time. Multiple computing resources can be doing many things simultaneously. For example, the Access Grid (www.accessgrid.org) provides a global collaboration network where people from around the world can meet and conduct work "virtually".
  • Limits to serial computing: Both physical and practical reasons pose significant constraints to simply building ever faster serial computers:
    • Transmission speeds - the speed of a serial computer is directly dependent upon how fast data can move through hardware. Absolute limits are the speed of light (30 cm/nanosecond) and the transmission limit of copper wire (9 cm/nanosecond). Increasing speeds necessitate increasing proximity of processing elements.
    • Limits to miniaturization - processor technology is allowing an increasing number of transistors to be placed on a chip. However, even with molecular or atomic-level components, a limit will be reached on how small components can be.
    • Economic limitations - it is increasingly expensive to make a single processor faster. Using a larger number of moderately fast commodity processors to achieve the same (or better) performance is less expensive.
    • Current computer architectures are increasingly relying upon hardware level parallelism to improve performance:
      • Multiple execution units
      • Pipelined instructions
      • Multi-core

von Neumann Architecture

  • Named after the Hungarian mathematician John von Neumann who first authored the general requirements for an electronic computer in his 1945 papers.
  • Since then, virtually all computers have followed this basic design, differing from earlier computers which were programmed through "hard wiring".
    von Neumann architecture
    • Comprised of four main components:
      • Memory
      • Control Unit
      • Arithmetic Logic Unit
      • Input/Output
    • Read/write, random access memory is used to store both program instructions and data
      • Program instructions are coded data which tell the computer to do something
      • Data is simply information to be used by the program
    • Control unit fetches instructions/data from memory, decodes the instructions and thensequentially coordinates operations to accomplish the programmed task.
    • Aritmetic Unit performs basic arithmetic operations
    • Input/Output is the interface to the human operator

  • So what? Who cares? Well, parallel computers still follow this basic design, just multiplied in units. The basic, fundamental architecture remains the same.

Flynn's Classical Taxonomy

  • There are different ways to classify parallel computers. One of the more widely used classifications, in use since 1966, is called Flynn's Taxonomy.
  • Flynn's taxonomy distinguishes multi-processor computer architectures according to how they can be classified along the two independent dimensions ofInstruction and Data. Each of these dimensions can have only one of two possible states: Single or Multiple.
  • The matrix below defines the 4 possible classifications according to Flynn:

    S I S DSingle Instruction, Single Data

    S I M DSingle Instruction, Multiple Data

    M I S DMultiple Instruction, Single Data

    M I M DMultiple Instruction, Multiple Data

 Single Instruction, Single Data (SISD):
  • A serial (non-parallel) computer
  • Single Instruction: Only one instruction stream is being acted on by the CPU during any one clock cycle
  • Single Data: Only one data stream is being used as input during any one clock cycle
  • Deterministic execution
  • This is the oldest and even today, the most common type of computer
  • Examples: older generation mainframes, minicomputers and workstations; most modern day PCs.
SISD

    UNIVAC1
    		IBM 360
    		CRAY1
    CDC 7600
    		PDP1
    		Dell Laptop
 Single Instruction, Multiple Data (SIMD):
  • A type of parallel computer
  • Single Instruction: All processing units execute the same instruction at any given clock cycle
  • Multiple Data: Each processing unit can operate on a different data element
  • Best suited for specialized problems characterized by a high degree of regularity, such as graphics/image processing.
  • Synchronous (lockstep) and deterministic execution
  • Two varieties: Processor Arrays and Vector Pipelines
  • Examples:
    • Processor Arrays: Connection Machine CM-2, MasPar MP-1 & MP-2, ILLIAC IV
    • Vector Pipelines: IBM 9000, Cray X-MP, Y-MP & C90, Fujitsu VP, NEC SX-2, Hitachi S820, ETA10
  • Most modern computers, particularly those with graphics processor units (GPUs) employ SIMD instructions and execution units.


SIMD

    ILLIAC IV
    		MasPar

             
    Cray X-MP
    		Cray Y-MP
    		Thinking Machines CM-2
    		Cell Processor (GPU)
 Multiple Instruction, Single Data (MISD):
  • A type of parallel computer
  • Multiple Instruction: Each processing unit operates on the data independently via separate instruction streams.
  • Single Data: A single data stream is fed into multiple processing units.
  • Few actual examples of this class of parallel computer have ever existed. One is the experimental Carnegie-Mellon C.mmp computer (1971).
  • Some conceivable uses might be:
    • multiple frequency filters operating on a single signal stream
    • multiple cryptography algorithms attempting to crack a single coded message.


MISD

 Multiple Instruction, Multiple Data (MIMD):
  • A type of parallel computer
  • Multiple Instruction: Every processor may be executing a different instruction stream
  • Multiple Data: Every processor may be working with a different data stream
  • Execution can be synchronous or asynchronous, deterministic or non-deterministic
  • Currently, the most common type of parallel computer - most modern supercomputers fall into this category.
  • Examples: most current supercomputers, networked parallel computer clusters and "grids", multi-processor SMP computers, multi-core PCs.
  • Note: many MIMD architectures also include SIMD execution sub-components


MIMD

    IBM POWER5
    		HP/Compaq Alphaserver
    		Intel IA32
    AMD Opteron
    		Cray XT3
    		IBM BG/L

Some General Parallel Terminology

Like everything else, parallel computing has its own "jargon". Some of the more commonly used terms associated with parallel computing are listed below. Most of these will be discussed in more detail later.
Supercomputing / High Performance Computing (HPC)
Using the world's fastest and largest computers to solve large problems.
Node
A standalone "computer in a box". Usually comprised of multiple CPUs/processors/cores. Nodes are networked together to comprise a supercomputer.
CPU / Socket / Processor / Core
This varies, depending upon who you talk to. In the past, a CPU (Central Processing Unit) was a singular execution component for a computer. Then, multiple CPUs were incorporated into a node. Then, individual CPUs were subdivided into multiple "cores", each being a unique execution unit. CPUs with multiple cores are sometimes called "sockets" - vendor dependent. The result is a node with multiple CPUs, each containing multiple cores. The nomenclature is confused at times. Wonder why?
Task
A logically discrete section of computational work. A task is typically a program or program-like set of instructions that is executed by a processor. A parallel program consists of multiple tasks running on multiple processors.
Pipelining
Breaking a task into steps performed by different processor units, with inputs streaming through, much like an assembly line; a type of parallel computing.
Shared Memory
From a strictly hardware point of view, describes a computer architecture where all processors have direct (usually bus based) access to common physical memory. In a programming sense, it describes a model where parallel tasks all have the same "picture" of memory and can directly address and access the same logical memory locations regardless of where the physical memory actually exists.
Symmetric Multi-Processor (SMP)
Hardware architecture where multiple processors share a single address space and access to all resources; shared memory computing.
Distributed Memory
In hardware, refers to network based memory access for physical memory that is not common. As a programming model, tasks can only logically "see" local machine memory and must use communications to access memory on other machines where other tasks are executing.
Communications
Parallel tasks typically need to exchange data. There are several ways this can be accomplished, such as through a shared memory bus or over a network, however the actual event of data exchange is commonly referred to as communications regardless of the method employed.
Synchronization
The coordination of parallel tasks in real time, very often associated with communications. Often implemented by establishing a synchronization point within an application where a task may not proceed further until another task(s) reaches the same or logically equivalent point.Synchronization usually involves waiting by at least one task, and can therefore cause a parallel application's wall clock execution time to increase.
Granularity
In parallel computing, granularity is a qualitative measure of the ratio of computation to communication.
  • Coarse: relatively large amounts of computational work are done between communication events
  • Fine: relatively small amounts of computational work are done between communication events
Observed Speedup
Observed speedup of a code which has been parallelized, defined as: 
wall-clock time of serial execution
-----------------------------------
 wall-clock time of parallel execution
One of the simplest and most widely used indicators for a parallel program's performance.
Parallel Overhead
The amount of time required to coordinate parallel tasks, as opposed to doing useful work. Parallel overhead can include factors such as:

Shared Memory

 General Characteristics:
  • Shared memory parallel computers vary widely, but generally have in common the ability for all processors to access all memory as global address space.
  • Multiple processors can operate independently but share the same memory resources.
  • Changes in a memory location effected by one processor are visible to all other processors.
  • Shared memory machines can be divided into two main classes based upon memory access times: UMA and NUMA.
 Uniform Memory Access (UMA):
  • Most commonly represented today by Symmetric Multiprocessor (SMP) machines
  • Identical processors
  • Equal access and access times to memory
  • Sometimes called CC-UMA - Cache Coherent UMA. Cache coherent means if one processor updates a location in shared memory, all the other processors know about the update. Cache coherency is accomplished at the hardware level.
 Non-Uniform Memory Access (NUMA):
  • Often made by physically linking two or more SMPs
  • One SMP can directly access memory of another SMP
  • Not all processors have equal access time to all memories
  • Memory access across link is slower
  • If cache coherency is maintained, then may also be called CC-NUMA - Cache Coherent NUMA
Shared memory architecture
Shared Memory (UMA)


NUMA
Shared Memory (NUMA)
 Advantages:
  • Global address space provides a user-friendly programming perspective to memory
  • Data sharing between tasks is both fast and uniform due to the proximity of memory to CPUs
 Disadvantages:
  • Primary disadvantage is the lack of scalability between memory and CPUs. Adding more CPUs can geometrically increases traffic on the shared memory-CPU path, and for cache coherent systems, geometrically increase traffic associated with cache/memory management.
  • Programmer responsibility for synchronization constructs that ensure "correct" access of global memory.
  • Expense: it becomes increasingly difficult and expensive to design and produce shared memory machines with ever increasing numbers of processors.

Distributed Memory

 General Characteristics:
  • Like shared memory systems, distributed memory systems vary widely but share a common characteristic. Distributed memory systems require a communication network to connect inter-processor memory.Distributed memory architecture
  • Processors have their own local memory. Memory addresses in one processor do not map to another processor, so there is no concept of global address space across all processors.
  • Because each processor has its own local memory, it operates independently. Changes it makes to its local memory have no effect on the memory of other processors. Hence, the concept of cache coherency does not apply.
  • When a processor needs access to data in another processor, it is usually the task of the programmer to explicitly define how and when data is communicated. Synchronization between tasks is likewise the programmer's responsibility.
  • The network "fabric" used for data transfer varies widely, though it can can be as simple as Ethernet.
 Advantages:
  • Memory is scalable with the number of processors. Increase the number of processors and the size of memory increases proportionately.
  • Each processor can rapidly access its own memory without interference and without the overhead incurred with trying to maintain cache coherency.
  • Cost effectiveness: can use commodity, off-the-shelf processors and networking.
 Disadvantages:
  • The programmer is responsible for many of the details associated with data communication between processors.
  • It may be difficult to map existing data structures, based on global memory, to this memory organization.
  • Non-uniform memory access (NUMA) times

Hybrid Distributed-Shared Memory

  • The largest and fastest computers in the world today employ both shared and distributed memory architectures.
    Hybrid memory architectureHybrid memory architecture

  • The shared memory component can be a cache coherent SMP machine and/or graphics processing units (GPU).
  • The distributed memory component is the networking of multiple SMP/GPU machines, which know only about their own memory - not the memory on another machine. Therefore, network communications are required to move data from one SMP/GPU to another.
  • Current trends seem to indicate that this type of memory architecture will continue to prevail and increase at the high end of computing for the foreseeable future.
  • Advantages and Disadvantages: whatever is common to both shared and distributed memory architectures.
  • Task start-up time
  • Synchronizations
  • Data communications
  • Software overhead imposed by parallel compilers, libraries, tools, operating system, etc.
  • Task termination time
Massively Parallel
Refers to the hardware that comprises a given parallel system - having many processors. The meaning of "many" keeps increasing, but currently, the largest parallel computers can be comprised of processors numbering in the hundreds of thousands.
Embarrassingly Parallel
Solving many similar, but independent tasks simultaneously; little to no need for coordination between the tasks.
Scalability
Refers to a parallel system's (hardware and/or software) ability to demonstrate a proportionate increase in parallel speedup with the addition of more processors. Factors that contribute to scalability include:
  • Hardware - particularly memory-cpu bandwidths and network communications
  • Application algorithm
  • Parallel overhead related
  • Characteristics of your specific application and coding
Parallel Programming Models


  • There are several parallel programming models in common use:
    • Shared Memory (without threads)
    • Threads
    • Distributed Memory / Message Passing
    • Data Parallel
    • Hybrid
    • Single Program Multiple Data (SPMD)
    • Multiple Program Multiple Data (MPMD)
  • Parallel programming models exist as an abstraction above hardware and memory architectures.
  • Although it might not seem apparent, these models are NOT specific to a particular type of machine or memory architecture. In fact, any of these models can (theoretically) be implemented on any underlying hardware. Two examples from the past are discussed below.
    • SHARED memory model on a DISTRIBUTED memory machine: Kendall Square Research (KSR) ALLCACHE approach.Machine memory was physically distributed across networked machines, but appeared to the user as a single shared memory (global address space). Generically, this approach is referred to as "virtual shared memory".
    		KSR1
    • DISTRIBUTED memory model on a SHARED memory machine: Message Passing Interface (MPI) on SGI Origin 2000.The SGI Origin 2000 employed the CC-NUMA type of shared memory architecture, where every task has direct access to global address space spread across all machines. However, the ability to send and receive messages using MPI, as is commonly done over a network of distributed memory machines, was implemented and commonly used.
    		SGI Origin 2000
  • Which model to use? This is often a combination of what is available and personal choice. There is no "best" model, although there certainly are better implementations of some models over others.
  • The following sections describe each of the models mentioned above, and also discuss some of their actual implementations.

Shared Memory Model (without threads)

  • In this programming model, tasks share a common address space, which they read and write to asynchronously.
  • Various mechanisms such as locks / semaphores may be used to control access to the shared memory.
  • An advantage of this model from the programmer's point of view is that the notion of data "ownership" is lacking, so there is no need to specify explicitly the communication of data between tasks. Program development can often be simplified.
  • An important disadvantage in terms of performance is that it becomes more difficult to understand and manage data locality.
    • Keeping data local to the processor that works on it conserves memory accesses, cache refreshes and bus traffic that occurs when multiple processors use the same data.
    • Unfortunately, controlling data locality is hard to understand and beyond the control of the average user.
 Implementations:
  • Native compilers and/or hardware translate user program variables into actual memory addresses, which are global. On stand-alone SMP machines, this is straightforward.
  • On distributed shared memory machines, such as the SGI Origin, memory is physically distributed across a network of machines, but made global through specialized hardware and software.

Threads Model

  • This programming model is a type of shared memory programming.
  • In the threads model of parallel programming, a single process can have multiple, concurrent execution paths.
  • Perhaps the most simple analogy that can be used to describe threads is the concept of a single program that includes a number of subroutines:Threads Model
    • The main program a.out is scheduled to run by the native operating system. a.outloads and acquires all of the necessary system and user resources to run.
    • a.out performs some serial work, and then creates a number of tasks (threads) that can be scheduled and run by the operating system concurrently.
    • Each thread has local data, but also, shares the entire resources of a.out. This saves the overhead associated with replicating a program's resources for each thread. Each thread also benefits from a global memory view because it shares the memory space of a.out.
    • A thread's work may best be described as a subroutine within the main program. Any thread can execute any subroutine at the same time as other threads.
    • Threads communicate with each other through global memory (updating address locations). This requires synchronization constructs to ensure that more than one thread is not updating the same global address at any time.
    • Threads can come and go, but a.out remains present to provide the necessary shared resources until the application has completed.
 Implementations:
  • From a programming perspective, threads implementations commonly comprise:
    • A library of subroutines that are called from within parallel source code
    • A set of compiler directives imbedded in either serial or parallel source code
    In both cases, the programmer is responsible for determining all parallelism.
  • Threaded implementations are not new in computing. Historically, hardware vendors have implemented their own proprietary versions of threads. These implementations differed substantially from each other making it difficult for programmers to develop portable threaded applications.
  • Unrelated standardization efforts have resulted in two very different implementations of threads: POSIX Threads and OpenMP.
  • POSIX Threads
    • Library based; requires parallel coding
    • Specified by the IEEE POSIX 1003.1c standard (1995).
    • C Language only
    • Commonly referred to as Pthreads.
    • Most hardware vendors now offer Pthreads in addition to their proprietary threads implementations.
    • Very explicit parallelism; requires significant programmer attention to detail.
  • OpenMP
    • Compiler directive based; can use serial code
    • Jointly defined and endorsed by a group of major computer hardware and software vendors. The OpenMP Fortran API was released October 28, 1997. The C/C++ API was released in late 1998.
    • Portable / multi-platform, including Unix and Windows NT platforms
    • Available in C/C++ and Fortran implementations
    • Can be very easy and simple to use - provides for "incremental parallelism"
  • Microsoft has its own implementation for threads, which is not related to the UNIX POSIX standard or OpenMP.

Distributed Memory / Message Passing Model

  • This model demonstrates the following characteristics:Message Passing Model
    • A set of tasks that use their own local memory during computation. Multiple tasks can reside on the same physical machine and/or across an arbitrary number of machines.
    • Tasks exchange data through communications by sending and receiving messages.
    • Data transfer usually requires cooperative operations to be performed by each process. For example, a send operation must have a matching receive operation.
 Implementations:
  • From a programming perspective, message passing implementations usually comprise a library of subroutines. Calls to these subroutines are imbedded in source code. The programmer is responsible for determining all parallelism.
  • Historically, a variety of message passing libraries have been available since the 1980s. These implementations differed substantially from each other making it difficult for programmers to develop portable applications.
  • In 1992, the MPI Forum was formed with the primary goal of establishing a standard interface for message passing implementations.
  • Part 1 of the Message Passing Interface (MPI) was released in 1994. Part 2 (MPI-2) was released in 1996. Both MPI specifications are available on the web at http://www-unix.mcs.anl.gov/mpi/.
  • MPI is now the "de facto" industry standard for message passing, replacing virtually all other message passing implementations used for production work. MPI implementations exist for virtually all popular parallel computing platforms. Not all implementations include everything in both MPI1 and MPI2.

Data Parallel Model

  • The data parallel model demonstrates the following characteristics:Data Parallel Model
    • Most of the parallel work focuses on performing operations on a data set. The data set is typically organized into a common structure, such as an array or cube.
    • A set of tasks work collectively on the same data structure, however, each task works on a different partition of the same data structure.
    • Tasks perform the same operation on their partition of work, for example, "add 4 to every array element".
  • On shared memory architectures, all tasks may have access to the data structure through global memory. On distributed memory architectures the data structure is split up and resides as "chunks" in the local memory of each task.

 Implementations:
  • Programming with the data parallel model is usually accomplished by writing a program with data parallel constructs. The constructs can be calls to a data parallel subroutine library or, compiler directives recognized by a data parallel compiler.
  • Fortran 90 and 95 (F90, F95): ISO/ANSI standard extensions to Fortran 77.
    • Contains everything that is in Fortran 77
    • New source code format; additions to character set
    • Additions to program structure and commands
    • Variable additions - methods and arguments
    • Pointers and dynamic memory allocation added
    • Array processing (arrays treated as objects) added
    • Recursive and new intrinsic functions added
    • Many other new features
    Implementations are available for most common parallel platforms.
  • High Performance Fortran (HPF): Extensions to Fortran 90 to support data parallel programming.
    • Contains everything in Fortran 90
    • Directives to tell compiler how to distribute data added
    • Assertions that can improve optimization of generated code added
    • Data parallel constructs added (now part of Fortran 95)
    HPF compilers were relatively common in the 1990s, but are no longer commonly implemented.
  • Compiler Directives: Allow the programmer to specify the distribution and alignment of data. Fortran implementations are available for most common parallel platforms.
  • Distributed memory implementations of this model usually require the compiler to produce object code with calls to a message passing library (MPI) for data distribution. All message passing is done invisibly to the programmer.

Hybrid Model

  • A hybrid model combines more than one of the previously described programming models.Hybrid Model
  • Currently, a common example of a hybrid model is the combination of the message passing model (MPI) with the threads model (OpenMP).
    • Threads perform computationally intensive kernels using local, on-node data
    • Communications between processes on different nodes occurs over the network using MPI
  • This hybrid model lends itself well to the increasingly common hardware environment of clustered multi/many-core machines.
  • Another similar and increasingly popular example of a hybrid model is using MPI with GPU (Graphics Processing Unit) programming.
    • GPUs perform computationally intensive kernels using local, on-node data
    • Communications between processes on different nodes occurs over the network using MPI

What is a Thread?



What are Pthreads?



Why Pthreads?




Designing Threaded Programs

 Parallel Programming:
 Shared Memory Model:
  • All threads have access to the same global, shared memory
  • Threads also have their own private data
  • Programmers are responsible for synchronizing access (protecting) globally shared data.
    Shared Memory Model
 Thread-safeness:
  • Thread-safeness: in a nutshell, refers an application's ability to execute multiple threads simultaneously without "clobbering" shared data or creating "race" conditions.
  • For example, suppose that your application creates several threads, each of which makes a call to the same library routine:
    • This library routine accesses/modifies a global structure or location in memory.
    • As each thread calls this routine it is possible that they may try to modify this global structure/memory location at the same time.
    • If the routine does not employ some sort of synchronization constructs to prevent data corruption, then it is not thread-safe.




threadunsafe

  • The implication to users of external library routines is that if you aren't 100% certain the routine is thread-safe, then you take your chances with problems that could arise.
  • Recommendation: Be careful if your application uses libraries or other objects that don't explicitly guarantee thread-safeness. When in doubt, assume that they are not thread-safe until proven otherwise. This can be done by "serializing" the calls to the uncertain routine, etc.
Thread API

  • The original Pthreads API was defined in the ANSI/IEEE POSIX 1003.1 - 1995 standard. The POSIX standard has continued to evolve and undergo revisions, including the Pthreads specification.
  • Copies of the standard can be purchased from IEEE or downloaded for free from other sites online.
  • The subroutines which comprise the Pthreads API can be informally grouped into four major groups:
    1. Thread management: Routines that work directly on threads - creating, detaching, joining, etc. They also include functions to set/query thread attributes (joinable, scheduling etc.)
    2. Mutexes: Routines that deal with synchronization, called a "mutex", which is an abbreviation for "mutual exclusion". Mutex functions provide for creating, destroying, locking and unlocking mutexes. These are supplemented by mutex attribute functions that set or modify attributes associated with mutexes.
    3. Condition variables: Routines that address communications between threads that share a mutex. Based upon programmer specified conditions. This group includes functions to create, destroy, wait and signal based upon specified variable values. Functions to set/query condition variable attributes are also included.
    4. Synchronization: Routines that manage read/write locks and barriers.
  • Naming conventions: All identifiers in the threads library begin with pthread_. Some examples are shown below.
    Routine PrefixFunctional Group
    pthread_Threads themselves and miscellaneous subroutines
    pthread_attr_Thread attributes objects
    pthread_mutex_Mutexes
    pthread_mutexattr_Mutex attributes objects.
    pthread_cond_Condition variables
    pthread_condattr_Condition attributes objects
    pthread_key_Thread-specific data keys
    pthread_rwlock_Read/write locks
    pthread_barrier_Synchronization barriers

  • The concept of opaque objects pervades the design of the API. The basic calls work to create or modify opaque objects - the opaque objects can be modified by calls to attribute functions, which deal with opaque attributes.
  • The Pthreads API contains around 100 subroutines. This tutorial will focus on a subset of these - specifically, those which are most likely to be immediately useful to the beginning Pthreads programmer.
  • For portability, the pthread.h header file should be included in each source file using the Pthreads library.
  • The current POSIX standard is defined only for the C language. Fortran programmers can use wrappers around C function calls. Some Fortran compilers (like IBM AIX Fortran) may provide a Fortram pthreads API.
COMPILING THREADED PROGRAMS

  • Several examples of compile commands used for pthreads codes are listed in the table below.
    Compiler / PlatformCompiler CommandDescription
    INTEL
    Linux    		icc -pthreadC
    icpc -pthreadC++
    PathScale
    Linux    		pathcc -pthreadC
    pathCC -pthreadC++
    PGI
    Linux    		pgcc -lpthreadC
    pgCC -lpthreadC++
    GNU
    Linux, BG/L, BG/P    		gcc -pthreadGNU C
    g++ -pthreadGNU C++
    IBM
    BG/L and BG/P    		bgxlc_r  /  bgcc_rC (ANSI  /  non-ANSI)
    bgxlC_r, bgxlc++_rC++

Creating and Terminating Threads

 Routines:
Question: After a thread has been created, how do you know a)when it will be scheduled to run by the operating system, and b)which processor/core it will run on?




 Terminating Threads & pthread_exit():
  • There are several ways in which a thread may be terminated:
    • The thread returns normally from its starting routine. It's work is done.
    • The thread makes a call to the pthread_exit subroutine - whether its work is done or not.
    • The thread is canceled by another thread via the pthread_cancel routine.
    • The entire process is terminated due to making a call to either the exec() or exit()
    • If main() finishes first, without calling pthread_exit explicitly itself
  • The pthread_exit() routine allows the programmer to specify an optional termination status parameter. This optional parameter is typically returned to threads "joining" the terminated thread (covered later).
  • In subroutines that execute to completion normally, you can often dispense with calling pthread_exit() - unless, of course, you want to pass the optional status code back.
  • Cleanup: the pthread_exit() routine does not close files; any files opened inside the thread will remain open after the thread is terminated.
  • Discussion on calling pthread_exit() from main():
    • There is a definite problem if main() finishes before the threads it spawned if you don't call pthread_exit() explicitly. All of the threads it created will terminate because main() is done and no longer exists to support the threads.
    • By having main() explicitly call pthread_exit() as the last thing it does, main() will block and be kept alive to support the threads it created until they are done.

Example: Pthread Creation and Termination

  • This simple example code creates 5 threads with the pthread_create() routine. Each thread prints a "Hello World!" message, and then terminates with a call to pthread_exit().
#include <pthread.h>
#include <stdio.h>
#define NUM_THREADS     5

void *PrintHello(void *threadid)
{
   long tid;
   tid = (long)threadid;
   printf("Hello World! It's me, thread #%ld!\n", tid);
   pthread_exit(NULL);
}

int main (int argc, char *argv[])
{
   pthread_t threads[NUM_THREADS];
   int rc;
   long t;
   for(t=0; t<NUM_THREADS; t++){
      printf("In main: creating thread %ld\n", t);
      rc = pthread_create(&threads[t], NULL, PrintHello, (void *)t);
      if (rc){
         printf("ERROR; return code from pthread_create() is %d\n", rc);
         exit(-1);
      }
   }

   /* Last thing that main() should do */
   pthread_exit(NULL);
}
/*
OUPUT:
In main: creating thread 0
In main: creating thread 1
Hello World! It's me, thread #0!
In main: creating thread 2
Hello World! It's me, thread #1!
Hello World! It's me, thread #2!
In main: creating thread 3
In main: creating thread 4
Hello World! It's me, thread #3!
Hello World! It's me, thread #4!

*/

Passing Arguments to Threads

  • The pthread_create() routine permits the programmer to pass one argument to the thread start routine. For cases where multiple arguments must be passed, this limitation is easily overcome by creating a structure which contains all of the arguments, and then passing a pointer to that structure in thepthread_create() routine.
  • All arguments must be passed by reference and cast to (void *).
/* How can you safely pass data to newly created threads, given their non-deterministic start-up and scheduling?  */

Thread Argument Passing

Thread Argument Passing
This code fragment demonstrates how to pass a simple integer to each thread. The calling thread uses a unique data structure for each thread, insuring that each thread's argument remains intact throughout the program.



long *taskids[NUM_THREADS];

for(t=0; t<NUM_THREADS; t++)
{
   taskids[t] = (long *) malloc(sizeof(long));
   *taskids[t] = t;
   printf("Creating thread %ld\n", t);
   rc = pthread_create(&threads[t], NULL, PrintHello, (void *) taskids[t]);
   ...
}

/*

Creating thread 0
Creating thread 1
Creating thread 2
Creating thread 3
Creating thread 4
Creating thread 5
Creating thread 6
Creating thread 7
Thread 0: English: Hello World!
Thread 1: French: Bonjour, le monde!
Thread 2: Spanish: Hola al mundo
Thread 3: Klingon: Nuq neH!
Thread 4: German: Guten Tag, Welt!
Thread 5: Russian: Zdravstvytye, mir!
Thread 6: Japan: Sekai e konnichiwa!
Thread 7: Latin: Orbis, te saluto!

*/


/*
Thread Argument Passing (Incorrect)
This example performs argument passing incorrectly. It passes the address of variable t, which is shared memory space and visible to all threads. As the loop iterates, the value of this memory location changes, possibly before the created threads can access it.
*/
int rc;
long t;

for(t=0; t<NUM_THREADS; t++)
{
   printf("Creating thread %ld\n", t);
   rc = pthread_create(&threads[t], NULL, PrintHello, (void *) &t);
   ...
}
/*
Creating thread 0
Creating thread 1
Creating thread 2
Creating thread 3
Creating thread 4
Creating thread 5
Creating thread 6
Creating thread 7
Hello from thread 140737488348392
Hello from thread 140737488348392
Hello from thread 140737488348392
Hello from thread 140737488348392
Hello from thread 140737488348392
Hello from thread 140737488348392
Hello from thread 140737488348392
Hello from thread 140737488348392
*/

Joining and Detaching Threads

Routines:

pthread_join (threadid,status)
pthread_detach (threadid)

pthread_attr_setdetachstate (attr,detachstate)

pthread_attr_getdetachstate (attr,detachstate)

Joining:
  • "Joining" is one way to accomplish synchronization between threads. For example:Joining
  • The pthread_join() subroutine blocks the calling thread until the specified threadid thread terminates.
  • The programmer is able to obtain the target thread's termination return status if it was specified in the target thread's call to pthread_exit().
  • A joining thread can match one pthread_join() call. It is a logical error to attempt multiple joins on the same thread.
  • Two other synchronization methods, mutexes and condition variables, will be discussed later.
 Joinable or Not?
  • When a thread is created, one of its attributes defines whether it is joinable or detached. Only threads that are created as joinable can be joined. If a thread is created as detached, it can never be joined.
  • The final draft of the POSIX standard specifies that threads should be created as joinable.
  • To explicitly create a thread as joinable or detached, the attr argument in the pthread_create() routine is used. The typical 4 step process is:
    1. Declare a pthread attribute variable of the pthread_attr_t data type
    2. Initialize the attribute variable with pthread_attr_init()
    3. Set the attribute detached status with pthread_attr_setdetachstate()
    4. When done, free library resources used by the attribute with pthread_attr_destroy()
 Detaching:
  • The pthread_detach() routine can be used to explicitly detach a thread even though it was created as joinable.
  • There is no converse routine.
 Recommendations:
  • If a thread requires joining, consider explicitly creating it as joinable. This provides portability as not all implementations may create threads as joinable by default.
  • If you know in advance that a thread will never need to join with another thread, consider creating it in a detached state. Some system resources may be able to be freed.
/*Pthread Joining
Example Code - Pthread Joining
This example demonstrates how to "wait" for thread completions by using the Pthread join routine. Since some implementations of Pthreads may not create threads in a joinable state, the threads in this example are explicitly created
 in a joinable state so that they can be joined later.
*/
#include <pthread.h>
#include <stdio.h>
#include <stdlib.h>
#include <math.h>
#define NUM_THREADS 4

void *BusyWork(void *t)
{
   int i;
   long tid;
   double result=0.0;
   tid = (long)t;
   printf("Thread %ld starting...\n",tid);
   for (i=0; i<1000000; i++)
   {
      result = result + sin(i) * tan(i);
   }
   printf("Thread %ld done. Result = %e\n",tid, result);
   pthread_exit((void*) t);
}

int main (int argc, char *argv[])
{
   pthread_t thread[NUM_THREADS];
   pthread_attr_t attr;
   int rc;
   long t;
   void *status;

   /* Initialize and set thread detached attribute */
   pthread_attr_init(&attr);
   pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_JOINABLE);

   for(t=0; t<NUM_THREADS; t++) {
      printf("Main: creating thread %ld\n", t);
      rc = pthread_create(&thread[t], &attr, BusyWork, (void *)t); 
      if (rc) {
         printf("ERROR; return code from pthread_create() 
                is %d\n", rc);
         exit(-1);
         }
      }

   /* Free attribute and wait for the other threads */
   pthread_attr_destroy(&attr);
   for(t=0; t<NUM_THREADS; t++) {
      rc = pthread_join(thread[t], &status);
      if (rc) {
         printf("ERROR; return code from pthread_join() 
                is %d\n", rc);
         exit(-1);
         }
      printf("Main: completed join with thread %ld having a status   
            of %ld\n",t,(long)status);
      }
printf("Main: program completed. Exiting.\n");
pthread_exit(NULL);
}
/*
Main: creating thread 0
Main: creating thread 1
Thread 0 starting...
Main: creating thread 2
Thread 1 starting...
Main: creating thread 3
Thread 2 starting...
Thread 3 starting...
Thread 1 done. Result = -3.153838e+06
Thread 0 done. Result = -3.153838e+06
Main: completed join with thread 0 having a status of 0
Main: completed join with thread 1 having a status of 1
Thread 3 done. Result = -3.153838e+06
Thread 2 done. Result = -3.153838e+06
Main: completed join with thread 2 having a status of 2
Main: completed join with thread 3 having a status of 3
Main: program completed. Exiting.
*/

Stack Management

Routines:


pthread_attr_getstacksize (attr, stacksize)
pthread_attr_setstacksize (attr, stacksize)

pthread_attr_getstackaddr (attr, stackaddr)

pthread_attr_setstackaddr (attr, stackaddr)


Preventing Stack Problems:
  • The POSIX standard does not dictate the size of a thread's stack. This is implementation dependent and varies.
  • Exceeding the default stack limit is often very easy to do, with the usual results: program termination and/or corrupted data.
  • Safe and portable programs do not depend upon the default stack limit, but instead, explicitly allocate enough stack for each thread by using thepthread_attr_setstacksize routine.
  • The pthread_attr_getstackaddr and pthread_attr_setstackaddr routines can be used by applications in an environment where the stack for a thread must be placed in some particular region of memory.
 Some Practical Examples at LC:
  • Default thread stack size varies greatly. The maximum size that can be obtained also varies greatly, and may depend upon the number of threads per node.
  • Both past and present architectures are shown to demonstrate the wide variation in default thread stack size.
    Node
    Architecture    		#CPUsMemory (GB)Default Size
    (bytes)
    AMD Xeon 566012242,097,152
    AMD Opteron8162,097,152
    Intel IA644833,554,432
    Intel IA32242,097,152
    IBM Power5832196,608
    IBM Power4816196,608
    IBM Power3161698,304


/*Example Code - Stack Management
This example demonstrates how to query and set a thread's stack size.*/
#include <pthread.h>
#include <stdio.h>
#define NTHREADS 4
#define N 1000
#define MEGEXTRA 1000000

pthread_attr_t attr;

void *dowork(void *threadid)
{
   double A[N][N];
   int i,j;
   long tid;
   size_t mystacksize;

   tid = (long)threadid;
   pthread_attr_getstacksize (&attr, &mystacksize);
   printf("Thread %ld: stack size = %li bytes \n", tid, mystacksize);
   for (i=0; i<N; i++)
     for (j=0; j<N; j++)
      A[i][j] = ((i*j)/3.452) + (N-i);
   pthread_exit(NULL);
}

int main(int argc, char *argv[])
{
   pthread_t threads[NTHREADS];
   size_t stacksize;
   int rc;
   long t;

   pthread_attr_init(&attr);
   pthread_attr_getstacksize (&attr, &stacksize);
   printf("Default stack size = %li\n", stacksize);
   stacksize = sizeof(double)*N*N+MEGEXTRA;
   printf("Amount of stack needed per thread = %li\n",stacksize);
   pthread_attr_setstacksize (&attr, stacksize);
   printf("Creating threads with stack size = %li bytes\n",stacksize);
   for(t=0; t<NTHREADS; t++){
      rc = pthread_create(&threads[t], &attr, dowork, (void *)t);
      if (rc){
         printf("ERROR; return code from pthread_create() is %d\n", rc);
         exit(-1);
      }
   }
   printf("Created %ld threads.\n", t);
   pthread_exit(NULL);
}
/*

*/

Read more »
Posted by at No comments:
Subscribe to: Posts (Atom)