Problem with learging Cuda

Hello all,

I’m a physicist who has eventually used programming tools for work, for instance simulate CMOS sensors in Matlab, or just medical x-ray models using Delphi in the past. This means I am not a software developer. Through > 20 years I have a number of scientific publications, book chapters, etc, that included work on programming like modifying algorithms or coding my own stuff in Matlab, but I have never been a programmer or devloper, just a user so to speak…

My company wants me to be a software developer in Cuda to develop ray trace based applications. They sent me to a 5 day C++ training that was a waste of time. My knowledge of C++ is basically limited to its similarities to Delphi or Matab, but do not know how it works. Now, I ahve to understand Cuda in a few weeks. At the moment I’m following a raytracing online page (https://raytracing.github.io/books/RayTracingInOneWeekend.html) but I do not understand most of what it is done. Furthermore, the fact that you cannot add breakpoint to Cuda files due to its construction, makes me quite lost.

Let me know if you could advise any online training for someone who doesn’t know C++, that could make me a software developer in Cuda for ray tracing in no more than 1 month.

Thanks a lot in advance and Best Regards!!

A next logical step after the Peter Shirley course might be this one:

https://devblogs.nvidia.com/accelerated-ray-tracing-cuda/

Thanks Robert, the problem is that I don’t get the Pter Shirley code, do ont understand it, I don’t see the logic…

Hi ,

Week 1: Know the basics of C programming. Probably spend the week coding problem statements understanding pointers, memory access/management, profiling.
Week 2: Read the basics of CUDA , probably I would suggest you to borrow a book:
https://developer.nvidia.com/cuda-books-archive
Week 3: Basic CUDA problem statements around matrix computations.
Week 4: Basics of CUDA for RayTracing.

I hope this helps.

-SKA