Blogs
Blogs
Matching:
upc
X
Display: 
基于UPC(Unified Parallel C)的可扩展并行规约实现(第二部分) Community Blog
续 基于UPC(Unified Parallel C)的可扩展并行规约实现 原文 Implementing a Scalable Parallel Reduction in Unified Parallel C (part 2) 结果明显是错误的，但问题在哪里？眼尖的读者可能已经看出来了，该程序包含了一个竞态状态(race condition)。多个线程可以并发的写共享变量"sum"，这可能会重写掉之前已经写入的部分值。 ...
from Blog: IBM 编译器中国开发团队


IBM 100 Icons of Progress  or, Why I’m getting less done lately Community Blog
I challenge anyone working in the IT industry � or anyone with even a casual interest in technology � to visit the new IBM 100 Icons of Progress web site and not lose track of time. The site is intended as a retrospective of a century of IBM innovation, but in...
from Blog: IBM Redbooks Software Blog


Implementing a Scalable Parallel Reduction in Unified Parallel C Community Blog
NancyWang
Tags:
cppcafe
upc_programming
parallel
upc_forall
parallel_performance
reduction
parallel_computing
upc
4,315 Views
A reduction is the process of combining elements of a vector (or array) to yield a single aggregate element. It is commonly used in scientific computations. For instance the inner product of two ndimensional vectors x, y is given by: This computation requires...
from Blog: C/C++ Cafe


Implementing a Scalable Parallel Reduction in Unified Parallel C (part 3) Community Blog
NancyWang
Tags:
parallel
reduction
upc
upc_forall
cppcafe
parallel_computing
parallel_performance
parallel_programming
3,137 Views
continue from the second parallel reduction blog . To get better scalability (increased program performance as the number of threads increases), it is critical to remove the lock in the upc_forall loop. This can be done by accumulating the partial sum...
from Blog: C/C++ Cafe


Implementing a Scalable Parallel Reduction in Unified Parallel C (part 2) Community Blog
NancyWang
Tags:
parallel_computing
reduction
cppcafe
upc
parallel_programming
upc_forall
parallel
parallel_performance
1 Comment
3,965 Views
continue from the previous parallel reduction blog The result is obvious wrong, but what is the problem? The keen reader might point out that the program as written contains a race condition. Multiple threads can write into shared variable "sum" concurrently,...
from Blog: C/C++ Cafe
