I am newbies to CUDA but I have read and test a lot of Example provide in the SDK which are really valuable source of information.
But I still have some question which regarding the design of my application.
The first issue is that I would like to know if it is possible to call windows function (API) to rendere image directly in the cuda device?
The second issue is that I have three for loop.
Serial version for each Frame For earch Beam Rectify Scan data in Beam for each data make compute each data Approach Parallel version in CUDA map Frame to Grid map Beam to Block Rectify Scan data map data to sample used thread Idx to compute each data
dear “cuda guru” is this approach good with cuda architecture ?
if yes how can I implemented this in CUDA (pseudo code) I mean how can I get the index of thread.
Beam and Scan data size are variable and all loop are really independent.
Thanks a lot for your reply.