<Group Member Names>
Final Project Report
Due: December 14th, 2007
Title of Final Project
I. Introduction
Reiterate your main points from your design document. Introduce the concept and
motivations, i.e. why is the potential improved performance worth the effort of
undertaking the project? Be sure to include any changes in your goals or motivations
over the course of the project, if applicable.
II. Design Overview
Draw a block diagram of the overview of the project, include the final flow charts of your
algorithms, and discuss how the parallelism is exploited in this algorithm.
III. Implementation
Break down the previous section further, and describe in more detail the kernels and
functions you created. Describe in more detail the data that has to flow through each
block, describing any data marshalling where applicable.
IV. Verification
What is your testing procedure to make sure its operation is robust and correct? For cases
where the accuracy may be compromised in order to use the single-precision hardware
and/or to achieve higher level of parallelism, provide an acceptance test for your output.
V. Performance
Did you achieve your performance goals? What went well with CUDA in terms of
performance gained? What hindered performance? Given enough developer time, what
is the best performance do you think you could ultimately achieve with this application
on this generation of cards? What are the features in the toolset or hardware that would
allow you to increase performance even more?
VI. Conclusion