Loading Structure Data Efficiently Example

Accelerated Computing CUDA CUDA Programming and Performance
1

I want to learn more about Loading Data Structures in CUDA, especially loading 2D arrays examples and came across a PDF on loading structured data by Nvidia
“Loading Structured Data Efficiently With Cuda”

[url=“http://developer.download.nvidia.com/compute/cuda/sdk/website/projects/vectorLoads/doc/vectorLoads.pdf”]http://developer.download.nvidia.com/compu...vectorLoads.pdf[/url]

But I couldn’t find the vectorLoad examples in the SDK.

Does anyone know where i can find it,

Thanks

2

Hy!

it is in the 1.0 SDK examples, the last one!

regards.

3

btw,

When you want to load an array of structures from global memory to shared memory:

Just use integer pointers to cast the src and destination and load the array of structures as array of integers into shared memory(no matter what data-types the structure carries)! Now, u got all coalescing in place!

The only restriction is that “n*sizeof(struct)” has to be a multiple of sizeof(int) – you can always pad it to get it!

Once you are in shared memory - it does NOT matter much (as much as global memory)
:smile2:

4

what’s the name of that project example…?

My SDK is 1.1 … External Media .