Overview
Topic status: We're looking for students to study this topic.
Certain standard mathematical procedures (e.g. various Linear algebra operations, intrinsic functions such as exponential, trigonometric, etc) are repeatedly used in a wide variety of scientific and engineering computational tasks, but are relatively expensive to execute on standard CPU based computing platforms. In this project you will examine one or more of these procedures for implementation and execution in reconfigurable FPGA hardware. Alternatively, you can choose to investigate the implementation of a simplified Computational Fluid Dynamics (CFD) solver in reconfigurable hardware, which will involve the implementation of some standard linear algebra operations. You will start by understanding and profiling your chosen procedure(s) to identify parts suitable for FPGA co-processing, and also conduct a literature review to identify previous efforts in this area. Thereafter, you will attempt to design and simulate custom FPGA instructions using the VHDL programming language and Xilinx software tools, and implement your design on one of the available Xilinx FPGA computing platforms. If time permits, you will also explore the use of a higher level FPGA application development tool for this purpose (e.g. Mitrion-C, Impulse-C, DIME-C, ROCCC2.0, FpgaC, System Generator, ISE Schematic Design, or similar), and compare and contrast the use of a higher level tool with development in VHDL. Note that if your initially chosen procedure(s) is relatively complex, you will have the option of mainly using a higher level development tool in preference to VHDL. You will be supervised by both Faculty and High Performance Computing staff.
Find out more about this research project scholarship
- Study level
- Honours
- Supervisors
- QUT
- Organisational unit
Science and Engineering Faculty
- Research area
- Contact
-
Please contact the supervisor.
Dr Ross Hayward
Associate Professor Shlomo Geva
