Open main menu

CDOT Wiki β

Team Darth Vector

Revision as of 20:31, 30 November 2017 by Agodwin (talk | contribs)

TEAM, use this for formatting. Wiki Editing Cheat Sheet


Join me, and together we can fork the program as master and thread

Members


Alistair Godwin

Giorgi Osadze

Leonel Jara


TBB Background

Containers Comparison

List of TBB containers:

concurrent_queue : Multiple threads may simultaneously push and pop elements from the queue.

concurrent_vector :

concurrent_hash_map : hash table that permits concurrent accesses.

quotes: Intel Threaded Building Blocks book. Highly concurrent containers are very important because Standard Template Library (STL) containers generally are not concurrency-friendly, and attempts to modify them concurrently can easily corrupt the containers.

Algorithms

parallel_for

parallel_scan

parallel_reduce


Threads


STL Background

Containers Comparison

pair,vector,list,slist,

Algorithms

TBB Threads


Lock Convoying Problem

What is a Lock?

Quick background on mutex/lock

Parallelism Problems in STL

Put a picture and maybe some code here

Lock Convoying in TBB

Put another picture here


Study Ref: https://software.intel.com/en-us/blogs/2008/10/20/tbb-containers-vs-stl-performance-in-the-multi-core-age This leads into Concurrent_vector growing below..

Efficiency Comparison Parallel for and concurrent_vector

Concept: Fine-grained locking

Multiple threads operate on the container by locking only those portions they really need to lock.

Concept: Lock-free algorithms

Bits of knowledge:

STL interfaces are inherently not thread-safe.

Threading Building Blocks containers are not templated with an allocator argument.

Links

http://www.cs.northwestern.edu/~riesbeck/programming/c++/stl-summary.html

http://www.cplusplus.com/reference/stl/

https://www.inf.ed.ac.uk/teaching/courses/ppls/TBBtutorial.pdf


Coding Time Comparison for STL and TBB

Parallel Algorithms support

C++11 STL does not have much to offer for parallel algorithms natively unlike TBB

Though there are new algorithms for parallelism in the C++17 standard

More: http://www.bfilipek.com/2017/08/cpp17-details-parallel.html

Resource management

Overall partitioning, thread creation, and management is hidden in TBB

Thread Creation, terminating, and synchronizing is handled by TBB

The thread creation and overall resource for STL is managed by a combination of libraries;

-thread for creation and joining

-mutex for mutual exclusion and locks

-future for accessing results of asynchronous operations

Details: http://en.cppreference.com/w/cpp/thread