wiki.accu.org

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
Back to [[conference:committee:2015-proposals]]\\
\\
**Title**: GPU programming - a review\\
**Proposer**: [[2015:Pattabhi Raman]]\\
**Type**: Case Study\\
**Duration**: 90 mins\\
**Description**: \\
In my previous talk at ACCU 2014, I spoke about "Random number generation in C++ - present and potential future". The present talk will focus on its next domain, “GPU programming”. 
Although Graphics Processing Unit (GPU) was invented to boost image processing, it has proved to be a green solution to the super computation. This is so, because, a GPU contain hundreds of core and each core can do a little calculation.\\
\\
In a regular concurrent code, tasks are shared among cores of multicore CPU. Whereas in GPU programming, the code pass compute-intense functions from CPU to GPU and returns the results to CPU.
This helps time consuming simulations to run faster. Hence, super computers are being made with cluster of GPU, which consume less energy.\\
\\
Although simulation has benefited to be the best of the GPU programming, it is not much helpful in certain arithmetic calculations that lead to latency by frequent communication between CPU and GPU in such calculations.\\
\\
This talk will review a few fundamentals in GPU programming and it will discuss about its successes and limitations.\\
\\
\\
\\