Inner workings of the OpenCL compiler How does it work?

Accelerated Computing CUDA CUDA Programming and Performance
#1

Hi,

I am working on my thesis and trying to find some information about what happens in the background process when program binaries are created and run. So far I have not found any document that would explain this.

First of all, the Nvidia OpenCL driver is the same as the CUDA Driver?
The developer guide only states that clBuildProgram() generates PTX code. Is the PTX generated in a similar matter as ATI IL using a LLVM framework that does syntax checking and linking of the built in functions?
As I have understod it PTX is a intermediate representation. When is it converted to device specific binary code?

#2

Hi,

I am working on my thesis and trying to find some information about what happens in the background process when program binaries are created and run. So far I have not found any document that would explain this.

First of all, the Nvidia OpenCL driver is the same as the CUDA Driver?
The developer guide only states that clBuildProgram() generates PTX code. Is the PTX generated in a similar matter as ATI IL using a LLVM framework that does syntax checking and linking of the built in functions?
As I have understod it PTX is a intermediate representation. When is it converted to device specific binary code?