Thread Heirarchy

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi,

Just wanted to know what is the advantage of having an hierarchy of threads in Grid and Block. Since the blocks are executed independently of each other, can’t we simply always have a linear arrangement of Blocks inside a Grid and a linear arrangement of threads in a block. for example following declarations are valid:

DimGrid(100,1), DimGrid(500,1) and DimBlock(256,1,1), DimBlock(512,512,1) etc

Thanks in advance!!

2

Its mainly useful for representing your problem efficiently… :)
Look at the following thread for more details: [url=“http://forums.nvidia.com/index.php?showtopic=107759”]http://forums.nvidia.com/index.php?showtopic=107759[/url]