Differences
This shows you the differences between two versions of the page.
| Both sides previous revision Previous revision | Previous revision | ||
|
ffnamespace:about [2013/04/08 15:54] |
ffnamespace:about [2015/03/07 15:14] (current) aldinuc |
||
|---|---|---|---|
| Line 1: | Line 1: | ||
| + | ~~NOTOC~~ | ||
| + | {{description>Fastflow is a C++ pattern-based parallel programming framework based upon lock-free shared-memory multithreading}} | ||
| + | ===== FastFlow (FF) ===== | ||
| + | |||
| + | FastFlow (快速流) is a C++ parallel programming framework advocating high-level, pattern-based parallel programming. It chiefly supports streaming and data parallelism, targeting heterogenous platforms composed of clusters of shared-memory platforms, possibly equipped with computing accelerators such as NVidia GPGPUs, Xeon Phi, Tilera TILE64. | ||
| + | |||
| + | The main design philosophy of FastFlow is to provide application designers with key features for parallel programming (e.g. time-to-market, portability, efficiency and performance portability) via suitable parallel programming abstractions and a carefully designed run-time support. | ||
| + | |||
| + | ==== Application scenarios ==== | ||
| + | |||
| + | FastFlow is a general-purpose C++ programming framework for heterogenous parallel platforms. Like other high-level programming frameworks, such as Intel TBB and OpenMP, it simplifies the design and engineering of portable parallel applications. However, it has a clear edge in terms of expressiveness and performance with respect to other parallel programming frameworks in specific application scenarios, including, inter alia: | ||
| + | |||
| + | * fine-grain parallelism on cache-coherent shared-memory platforms; | ||
| + | * streaming applications; | ||
| + | * coupled usage of multi-core and accelerators. | ||
| + | |||
| + | In other cases FastFlow is typically comparable to (and is some cases slightly faster than) state-of-the-art parallel programming frameworks such as Intel TBB, OpenMP, Cilk, etc. More details may be found on the [[ffnamespace:performance|performance]] page. A number of both micro-benchmarks and real-world applications has been developed with FastFlow (or ported from other parallel libraries) in order to asses its usability and performance, among the others: | ||
| + | |||
| + | * **Bowtie2**: fast and sensitive read alignment //(multicore; porting of the original version developed with pthreads and spinlocks)// | ||
| + | * **Two-phase video/image restoration** for impulsive/Gaussian noise //(multicore, GPGPUs, heterogeneous; original algorithm)// | ||
| + | * **Block-based Cholesky & LU decomposition** for dense matrices //(multicore, original algorithms)// | ||
| + | * **Yadt C4.5 classifier** //(multicore; original algorithm)// | ||
| + | * **CWC Gillespie simulator** for systems biology //(multicore, distributed, GPGPUs; original algorithm)// | ||
| + | * SWPS3: vectorized **Smith-Waterman** local alignment software //(multicore; porting of the original version developed with pthreads)// | ||
| + | * **pbzip2**: Parallel BZIP2 //(multicore; porting of the original version developed with POSIX processes)// | ||
| + | * Fast networks **Deep Packet Inspection** //(multicore; original algorithm)// | ||
| + | * Several standard algorithms, such as nqueens, fibonacci, QT-mandebrot, matrix block multiplication, ... //(multicore, GPGPUs, distributed, heterogeneous)// | ||
| + | * over 100 micro-benchmarks testing individual features and patterns //(multicore, GPGPUs, distributed, heterogeneous)// | ||
| + | |||
| + | The source code of (almost) all applications can be found in the [[https://sourceforge.net/p/mc-fastflow/code/|FastFlow SVN repository]] under either LGPLv3 or GNU GPL license. The design of applications and their performance are described in [[http://alpha.di.unito.it/fastflow-papers|research papers]]. | ||
| + | ==== Design ==== | ||
| + | |||
| + | FastFlow comes as a C++ template library designed as a [[ffnamespace:architecture|stack of layers]] that progressively abstracts out the programming of parallel applications. The goal of the stack is threefold: portability, extensibility, and performance. For this, all the three layers are realised as thin strata of C++ templates that are 1) seamlessly portable; 2) easily extended via subclassing; and 3) statically compiled and cross-optimised with the application. The terse design ensures easy portability on almost all OSes and CPUs with a C++ compiler. The main development platform is Linux/x86_64/gcc, but it has been tested also on various combinations of x86, x86_64, PPC, ARM, Tilera, NVidia with gcc, icc, Visual Studio on Linux, Mac OS, and Windows XP/7. The FastFlow core has been ported to ARM/iOS. | ||
| + | |||
| + | The FastFlow [[ffnamespace:architecture|run-time support]] uses several techniques to efficiently support fine grain parallelism (and very high frequency streaming). Among these are: | ||
| + | |||
| + | * non-blocking multi-threading with lock-less synchronisations; | ||
| + | * zero-copy network messaging (via 0MQ/TCP and RDMA/Infiniband); | ||
| + | * asynchronous data feeding for accelerator offloading. | ||
| + | |||
| + | |||
| + | FastFlow has been adopted by a number of research projects and third-party development initiatives, and has thus been tested in a variety of application scenarios: from systems biology to high-frequency trading. | ||
| + | |||
| + | ==== Big Pictures ==== | ||
| + | |||
| + | ^FastFlow/C++11 in REPARA^FastFlow big picture (2014)^App: faster Bowtie2 (2013)^ | ||
| + | |[[http://calvados.di.unipi.it/storage/paper_files/2015_Artemis_REPARA_FF_poster.pdf|{{:ffnamespace:2015_artemis_repara_ff.png?220|}}]]|[[http://calvados.di.unipi.it/storage/paper_files/2014_ff_poster_hipeac.pdf|{{:ffnamespace:2014_ff_poster_hipeac.png?220}}]]|[[http://calvados.di.unipi.it/storage/paper_files/2013_ff_botie2_mem_affinity_acaces.pdf|{{:ffnamespace:2013_ff_botie2_mem_affinity_acaces.png?220}}]]| | ||
| + | |Artemis Co-Summit 2015| |HiPEAC-ACACES 2013| | ||
| + | |||
| + | ^FastFlow & its applications (2014)^Lock-less programming with FastFlow (2012)^ | ||
| + | |[[http://calvados.di.unipi.it/storage/paper_files/2014_ff_poster_openday_unito.pdf|{{:ffnamespace:2014_ff_poster_openday_unito.png?330|}}]]|[[http://calvados.di.unipi.it/storage/paper_files/2012_ACACES_poster.pdf|{{:ffnamespace:2012_acaces_poster.png?330}}]]| | ||
| + | |UniTO Industrial day 2014|HiPEAC-ACACES 2012| | ||
| + | |||
| + | |||
| + | ---- | ||
| + | <html> | ||
| + | <div id="eXTReMe"><a href="http://extremetracking.com/open?login=aldinuc2"> | ||
| + | <img src="http://t1.extreme-dm.com/i.gif" style="border: 0;" | ||
| + | height="38" width="41" id="EXim" alt="eXTReMe Tracker" /></a> | ||
| + | <script type="text/javascript"><!-- | ||
| + | EXref="";top.document.referrer?EXref=top.document.referrer:EXref=document.referrer;//--> | ||
| + | </script><script type="text/javascript"><!-- | ||
| + | var EXlogin='aldinuc2' // Login | ||
| + | var EXvsrv='s10' // VServer | ||
| + | EXs=screen;EXw=EXs.width;navigator.appName!="Netscape"? | ||
| + | EXb=EXs.colorDepth:EXb=EXs.pixelDepth;EXsrc="src"; | ||
| + | navigator.javaEnabled()==1?EXjv="y":EXjv="n"; | ||
| + | EXd=document;EXw?"":EXw="na";EXb?"":EXb="na"; | ||
| + | EXref?EXref=EXref:EXref=EXd.referrer; | ||
| + | EXd.write("<img "+EXsrc+"=http://e1.extreme-dm.com", | ||
| + | "/"+EXvsrv+".g?login="+EXlogin+"&", | ||
| + | "jv="+EXjv+"&j=y&srw="+EXw+"&srb="+EXb+"&", | ||
| + | "l="+escape(EXref)+" height=1 width=1>");//--> | ||
| + | </script><noscript><div id="neXTReMe"><img height="1" width="1" alt="" | ||
| + | src="http://e1.extreme-dm.com/s10.g?login=aldinuc2&j=n&jv=n" /> | ||
| + | </div></noscript></div> | ||
| + | </html> | ||
Page Tools
Except where otherwise noted, content on this wiki is licensed under the following license: CC Attribution-Share Alike 3.0 Unported
