Threads

• Process: Owner of resources allocated for individual program execution, can encompass more than one thread of execution
• Thread: Unit of execution (unit of dispatching) and a collection of resources, with which the unit of execution is associated, characterize the notion of a process. A thread is the abstraction of a unit of execution. It is also referred to as a light-weight process LWP that share the same text (program code) and global data, but possess their own CPU register values and their own dynamic (or stack based) variables
• First look at the advantages of threads;
  • a program does not stall when one of its operations blocks.
  • save contents of a page to disk while downloading other page (for web server example)
  • Simplification of programming model
• Single process, single thread MS-DOS, old MacOS
• Single process, multiple threads OS/161
• Multiple processes, single thread traditional Unix
• Multiple processes, multiple threads modern Unices (Solaris, Linux), Windows2000
• As a basic unit of CPU utilization, a thread consists of an instruction pointer (also referred to as the PC or instruction counter), CPU register set and a stack. A thread shares its code and data, as well as system resources and other OS related information, with its peer group (other threads of the same process)

  Figure 2.6: Single and Multithreaded Processes

  
• Threads versus processes;
  • A thread operates in much the same way as a process:
    • can be one of the several states.
    • executes sequentially (within a process and shares the CPU).
    • can issue system calls.
  • Economy of Overheads - managing processes is considerably more expensive than managing threads so LWPs are better.
  • Responsiveness - less setup work means faster response to requests, and multiple thread of execution mean there can be response from some threads even if other threads are busy or blocked.
  • Resource Sharing - Threads within a process share resources (including the same memory address space) conveniently and efficiently compared to separate processes
  • Threads within a process are NOT independent and are NOT protected against each other
  • Multiprocessor Use - if multiple processors are available, a multithreaded application can have its threads run in parallel which means better utilization (especially if there are few other processes present so that, without a multithreaded application, some CPUs would be idle)

  Figure 2.7: Threads and Processes

  
• A process utilizing multithreading is a process with multiple points of execution-up to now, we have assumed that each process has only one point of execution
• similarity between an operating system supporting multiple processes and a process supporting multiple threads
• Figure 2.6 shows a traditional (or heavyweight) process, on the left, and 3 LWPs are drawn on the right in a way to emphasize their common text and global data (i.e. data and heap)
• In practice, an application such as a web server that can have considerable variations in the rate of requests can create additional threads in response to serving load, yet minimize process creation load on the host
• Less setup improves responsiveness, and shared text means more efficient memory use

Figure 2.8: A word processor with three threads, a multithreaded web server