The CDF of a normal distribution is a quantity that shows up frequently in the physical sciences and statistics. The function can be computed easily from erfc(), but the inverse function requires a special implementation. Nvidia currently offers normcdf() and normcdfinv() as part of CUDA. It would be great if AMD offered similar functions as an OpenCL extension, with the eventual hope that these will be standardized in a future version of OpenCL.
A little bit of background: I use OpenCL for statistics application for Mac called Wizard Pro. I would love to be able to simplify my code with high quality functions supplied by AMD. As statistics and "big data" are looking larger in the computational landscape, I think many users besides myself would benefit from these functions. An OpenCL version of normcdf() and normcdfinv() would also make it easier for statistics users to port existing CUDA code to over to AMD hardware.
Thanks for your consideration.
Welcome and thanks for posting.
I have white-listed you, so you should be able to directly post in the relevant developer forum.
As this post seems to be related to OpenCL, I am moving it to that forum.