Wavefront algorithm Optimization

接触到了一种新的依赖解决算法问题,英文叫做wavefront algorithm,中文好像叫波前算法吧。大体思路是,对于一个X*Y数组结构的数据,当运算依赖关系是从西北角开始运算,直到覆盖所有的表项,得到最终结果的时候。每一个运算单元,都必须依赖其西,西北和北边相邻的数据才能开始运算时,其运算方式如同水波一样,逐层前进。

如下图模拟,数字增长,表示随着时间推移,运算的顺序:从1-》2-》3-》 -》7结束,完成整个数据的运算。

1 2 3 4
2 3 4 5
3 4 5 6
4 5 6 7

继续阅读“Wavefront algorithm Optimization”

Summary of this week

After reading the essays about recent research, the main parallelization schemes for optimization on Cell Processor are: optimize the DMA transfer (single buffering, double buffering and multi-way buffering), optimize the implementation models (how to divide the whole data block into different piece and how to connect each program structures) and optimize the get/put methods (using some prefetch mechanism such as pseudo-result, asynchronous unsafe cache and software cache).

通过几天阅读的论文情况,基本上发现了几种加速并行处理的方式:优化DMA传输方式,缓冲,双缓冲和多重缓冲;优化实现的模型,拆分程序流到不同的SPE,不同的计算模型,会得到不同的实现效果;优化数据读取的方式,比如利用预读取,虚拟结果,和异步多流水线缓存和程序缓存的方式。

继续阅读“Summary of this week”

Programming with PS3

Now I have a desktop, a laptop, a sony play station 3 with 3 mouse and 3 keyboards.

Programming in Fedora 9 and Yellow Dog Linux with Cell BE SDK and IBM Cell simulator.

The main theme is develop a waiting mechanism for data driven machine and uate the mechanism.

I want to write a formal essay and publish it before I go back home.

Need to work hard now.