Articles by Intel ISN (Articles: 8)

Articles: 8

Articles
Technical Blogs
Tips and Tricks
Reference Articles
Projects

Average article rating: 4.80

Product Showcase
	Loop Modifications to Enhance Data-Parallel Performance 1 Apr 2010 Updated: 1 Apr 2010 Rating: 0.00/5 Votes: 0 Popularity: 0.00 Licence: CPOL Views: 21,931 Bookmarked: 12 Downloaded: 0 Please Sign up or sign in to vote. In data-parallel applications, the same independent operation is performed repeatedly on different data. Loops are usually the most compute-intensive segments of data parallel applications, so loop optimizations directly impact performance.
Programming Languages
C	Granularity and Parallel Performance 8 Apr 2010 Updated: 8 Apr 2010 Rating: 5.00/5 Votes: 3 Popularity: 2.39 Licence: CPOL Views: 21,365 Bookmarked: 7 Downloaded: 0 Please Sign up or sign in to vote. One key to attaining good parallel performance is choosing the right granularity for the application. The goal is to determine the right granularity (usually larger is better) for parallel tasks, while avoiding load imbalance and communication overhead to achieve the best performance.
	Load Balance and Parallel Performance 8 Apr 2010 Updated: 8 Apr 2010 Rating: 5.00/5 Votes: 5 Popularity: 3.49 Licence: CPOL Views: 35,464 Bookmarked: 18 Downloaded: 0 Please Sign up or sign in to vote. Load balancing an application workload among threads is critical to performance. However, achieving perfect load balance is non-trivial, and it depends on the parallelism within the application, workload, the number of threads, load balancing policy, and the threading implementation.
	Expose Parallelism by Avoiding or Removing Artificial Dependencies 8 Apr 2010 Updated: 8 Apr 2010 Rating: 5.00/5 Votes: 1 Popularity: 0.00 Licence: CPOL Views: 15,902 Bookmarked: 4 Downloaded: 0 Please Sign up or sign in to vote. Many applications and algorithms contain serial optimizations that inadvertently introduce data dependencies and inhibit parallelism. One can often remove such dependences through simple transforms, or even avoid them altogether through techniques such as domain decomposition or blocking.
	Exploiting Data Parallelism in Ordered Data Streams 2 Jun 2010 Updated: 2 Jun 2010 Rating: 0.00/5 Votes: 0 Popularity: 0.00 Licence: CPOL Views: 14,922 Bookmarked: 2 Downloaded: 0 Please Sign up or sign in to vote. Many compute-intensive applications involve complex transformations of ordered input data to ordered output data. While the algorithms employed in these transformations are often parallel, managing the I/O order dependence can be a challenge.
	Avoiding Heap Contention Among Threads 2 Jun 2010 Updated: 2 Jun 2010 Rating: 0.00/5 Votes: 0 Popularity: 0.00 Licence: CPOL Views: 20,242 Bookmarked: 4 Downloaded: 0 Please Sign up or sign in to vote. Allocating memory from the system heap can be an expensive operation due to a lock used by system runtime libraries to synchronize access to the heap. Contention on this lock can limit the performance benefits from multithreading. Learn how to solve this problem.
	Use Thread-local Storage to Reduce Synchronization 2 Jun 2010 Updated: 2 Jun 2010 Rating: 4.00/5 Votes: 1 Popularity: 0.00 Licence: CPOL Views: 19,642 Bookmarked: 6 Downloaded: 0 Please Sign up or sign in to vote. Synchronization is often an expensive operation that can limit the performance of a multithreaded program. Using thread-local data structures instead of data structures shared by the threads can reduce synchronization in certain cases, allowing a program to run faster.
	Avoiding and Identifying False Sharing Among Threads 2 Jun 2010 Updated: 2 Jun 2010 Rating: 5.00/5 Votes: 2 Popularity: 1.51 Licence: CPOL Views: 24,545 Bookmarked: 2 Downloaded: 0 Please Sign up or sign in to vote. In symmetric multiprocessor (SMP) systems, each processor has a local cache. The memory system must guarantee cache coherence. False sharing occurs when threads on different processors modify variables that reside on the same cache line. Learn methods to detect and correct false sharing.