Multicore and shared multi processor

Multicore and shared multi processor

• 1.
  MULTICORE AND SHAREDMULTI PROCESSOR S. Janani, AP/CSE KCET
• 2.
  • Purpose: Hardware multithreadingimproved the efficiency of processors at modest cost, still the big challenge was to deliver on the performance potential of Moore’s Law by efficiently programming the increasing number of processors per chip.
• 3.
  • While rewritingold programs to run well on parallel hardware, the question may be arise as what can computer designers do to simplify the task? • Solution #1 : To provide a single physical address space that all processors can share, so that programs need not concern themselves with where their data is. In this approach, all variables of a program can be made available at any time to any processor.
• 4.
  SMP A shared memorymultiprocessor (SMP) is one that offers the programmer a single physical address space across all processors (multicore chips) also known as shared-address multiprocessor. • Processors communicate through shared variables in memory, with all processors capable of accessing any memory location via loads and stores • Such systems can still run independent jobs in
• 5.
  Figure: Classical SharedMemory Multiprocessor
• 6.
  Types of SingleAddress space multi processor • Single address space multiprocessors come in two styles. • Type #1: Uniform Memory Access (UMA): the latency to a word in memory does not depend on which processor asks for it • Type #2: Non-Uniform Memory Access (NUMA): some memory accesses are much faster than others, depending on which processor asks for which word, typically because main memory is divided and attached to different microprocessors or to different memory controllers on the same chip • As you might expect, the programming challenges are harder for a NUMA multiprocessor than for a UMA multiprocessor, but NUMA machines can scale to larger sizes and NUMAs can have lower latency to nearby memory
• 7.
  • When sharingis supported with a single address space, there must be a separate mechanism for synchronization • Lock : For synchronization => Only one processor at a time can acquire the lock, and other processors interested in shared data must wait until the original processor unlocks the variable
• 8.
  Example
• 11.
  OpenMP • OpenMP: Itis just an Application Programmer Interface (API) along with a set of compiler directives, environment variables, and runtime library routines that can extend standard programming languages. • It offers a portable, scalable, and simple programming model for shared memory multiprocessors. Its primary goal is to parallelize loops and to perform reductions. • Command to use the OpenMP API with the
• 12.
  #pragma omp parallelfor for (Pn = 0; Pn < P; Pn += 1) for (i = 0; 1000*Pn; i < 1000*(Pn+1); i += 1) sum[Pn] += A[i]; /*sum the assigned areas*/ Reduction: #pragma omp parallel for reduction(+ : FinalSum) for (i = 0; i < P; i += 1) FinalSum += sum[i]; /* Reduce to a single
• 13.
  Advantages and Limitations •Note that it is now up to the OpenMP library to find efficient code to sum 64 numbers efficiently using 64 processors. • While OpenMP makes it easy to write simple parallel code, it is not very helpful with debugging, so many parallel programmers use more sophisticated parallel programming systems than OpenMP, just as many programmers today use more productive languages than C