Coalesced Memory Access ?

Guys this might be a stupid question, but here it goes …

------> what is Coalesced Memory Access ???
-----------> in general terms ?
-----------> in CUDA term ?


please look at programming guide


It would be difficult for me to explain you in detail what Coalesced and non-coalesced memory access mean (in general terms and in CUDA terms), but I can address you to a very good tutorial on CUDA:

Download the ppt and/or the mp3 files from the lecture corresponding to

September 30, 2008 - “CUDA Memory Spaces and Access Overhead” and;
October 2, 2008 - “CUDA Memory Spaces and Access Overhead and Gauging Kernel Performance”.

You can start from slide 15 - Global Memory - on of the set of September and then continue with the first slides of the set of Octuber while listening to the explainations. Btw. Figures in CUDA Tutorial are very illustrative too.

Good luck,
Geovanny Giorgana