Loading...

• This forum is read-only   Click here to view the Best Answer Next Thread Prev Thread

• Tags

  parallel computing parallel computing applications

• Posts   Search Threads

• sudharsan2020 January 3, 2010 6:29 PM PST    Candidates for parallel computing Sir, I am currently doing a project on "Parallel Computing " as my final year project. The main objective of the project is the "performance gain of an application in parallel mode,when compared to serial mode" I am currently looking for "applications are good candidates for parallel computing" and really confused! Can you suggest me some Algorithms-relevant to be taken as a project Can be efficiently(and easily) paralleleized Thanks in advance. Tamer Assad January 4, 2010 3:06 AM PST Rate   #1 Re: Candidates for parallel computing Quoting - sudharsan2020 Sir, I am currently doing a project on "Parallel Computing " as my final year project. The main objective of the project is the "performance gain of an application in parallel mode,when compared to serial mode" I am currently looking for "applications are good candidates for parallel computing" and really confused! Can you suggest me some Algorithms-relevant to be taken as a project Can be efficiently(and easily) paralleleized Thanks in advance. Sudharsan, download intel parallel studio and explore its usage, the installation includes number of useful sample applications code. further more, for the puropse of your research, you may utilize some test cases (ex. pixel drawing function with time mesurement) based on 4 approches: 1) serial programming 2) native threads (threads APIs) 3) Intel Threading Building Blocks (TBB) 4) Intel OpenMP chech this useful article: "Intel® Threading Building Blocks, OpenMP, or native threads?" http://software.intel.com/en-us/articles/intel-threading-building-blocks-openmp-or-native-threads/ regards, Claudio Ochoa (Intel) January 4, 2010 4:45 AM PST Rate   #2 Reply To #1 Re: Candidates for parallel computing Quoting - t_assad Sudharsan, download intel parallel studio and explore its usage, the installation includes number of useful sample applications code. further more, for the puropse of your research, you may utilize some test cases (ex. pixel drawing function with time mesurement) based on 4 approches: 1) serial programming 2) native threads (threads APIs) 3) Intel Threading Building Blocks (TBB) 4) Intel OpenMP chech this useful article: "Intel® Threading Building Blocks, OpenMP, or native threads?" http://software.intel.com/en-us/articles/intel-threading-building-blocks-openmp-or-native-threads/ regards, I strongly suggest you to follow t_assad advice, and then use our service to measure the scalability of the parallel code you wrote. Regards Claudio Tamer Assad January 4, 2010 7:26 AM PST Rate   #3 Re: Candidates for parallel computing Quoting - sudharsan2020 Sir, I am currently doing a project on "Parallel Computing " as my final year project. The main objective of the project is the "performance gain of an application in parallel mode,when compared to serial mode" I am currently looking for "applications are good candidates for parallel computing" and really confused! Can you suggest me some Algorithms-relevant to be taken as a project Can be efficiently(and easily) paralleleized Thanks in advance. sudharsan, for measurement concerns: - take a look at parallel studio documentation (composer, amplifier, and inspector) and you have a clear view of your implementation performance. - for additional measurement (could be beyond your paper scope but will help you figure measurment methods) check this article "Measure Intel® IPP Function Performance" http://software.intel.com/en-use/articles/measure-intel-ipp-function-performance/ regards, mami January 4, 2010 4:15 PM PST Rate   #4 Reply To #3 Re: Candidates for parallel computing Quoting - t_assad sudharsan, for measurement concerns: - take a look at parallel studio documentation (composer, amplifier, and inspector) and you have a clear view of your implementation performance. - for additional measurement (could be beyond your paper scope but will help you figure measurment methods) check this article "Measure Intel® IPP Function Performance" http://software.intel.com/en-use/articles/measure-intel-ipp-function-performance/ regards, I would suggest something on different aspects of the issue: - You should investigage first what kinds of paralelisms are possible in theory (Hint: Task parallelism, data parallelizm etc) - For each kind of parallelism pick up some reference. (Hint: For data parallelism you may use a disk file containing your data.) - Consider also the case where you have some constraints on the processing steps (i.e., dependence, orderness etc) . Such issues would lead you "pipelining" concept. Don't miss this in your project :) You can find multitudes of graphics drawin (are you interested in fractals for instance?) possibilities Or what about some data stream processing? CRC calculation? Letter/word frequency calculation? Once you cover such major aspects of parallelism issue, you may then (or acutally from the very beggining) consider more practical issues like scalibility etc. Hope these points are not already obvious to you. Wish you luck and enjoy the work! Gastón C. Hillar January 6, 2010 6:04 AM PST Rate   #5 Re: Candidates for parallel computing Quoting - sudharsan2020 Sir, I am currently doing a project on "Parallel Computing " as my final year project. The main objective of the project is the "performance gain of an application in parallel mode,when compared to serial mode" I am currently looking for "applications are good candidates for parallel computing" and really confused! Can you suggest me some Algorithms-relevant to be taken as a project Can be efficiently(and easily) paralleleized Thanks in advance. Hi sudharsan, You can find excellent and accurate examples of algorithms and techniques in Dr. Clay Breshears' book: "The Art of Concurrency". I think it can help you in your research. Cheers, Gaston Gastón C. Hillar

• = Best Answer

  Next Thread Prev Thread

Intel Software Network Forums Statistics

17,025 users have contributed to 48,319 threads and 172,758 posts to date.

For more complete information about compiler optimizations, see our Optimization Notice.