<div>
I edit the question to be more direct !!!
</div>


<div>
Please look into CUDA SDK samples.<br />
Adding 32 elements of a vector should present no challenge for 12MHz 8086. You will spend much more time on initializing CUDA than operation takes inlined on normal CPU.<br />
You may try C++ offload, that should confirm indeed that CUDA is operational.
</div>


cuda.png

<div>
<div class="content wysiwyg-content">
I am new in CUDA programming I would like to know how to write a kernel that &nbsp;compute the average of array with 32 elements ?
</div>
</div>


I am new in CUDA programming I would like to know how to write a kernel that  compute the average of array with 32 elements ?

CUDA : Array Mean

C++ is an intermediate-level general-purpose programming language, not to be confused with C or C#. It was developed as a set of extensions to the C programming language to improve type-safety and add support for automatic resource management, object-orientation, generic programming, and exception handling, among other features.

Programming theory deals with the design, implementation, analysis, characterization, and classification of programming languages and their individual features, along with introductory programming concepts like program structure, variable declaration and conditional and looping constructs. Sub-disciplines include the formal semantics of programming languages, type theory, program analysis and transformation, comparative programming language analysis, metaprogramming, domain-specific languages, compiler construction and run-time systems.

Programming Theory

Programming includes both the specifics of the language you’re using, like Visual Basic, .NET, Java and others, but also the best practices in user experience and interfaces and the management of projects, version control and development. Other programming topics are related to web and cloud development and system and hardware programming.