Entry requirements

120 credits including 90 credits in computer science and mathematics with Computer Architecture or equivalent.

Learning outcomes

To pass, a student should be able to:

• Account and argue for how modern computer systems are designed, including details about pipeline, memory organisation, virtual and physical memory, and memory technology.
• Account and argue for techniques to create and use instruction-level parallelism, memory-level parallelism, and thread-level parallelism.
• Account and argue for the design and programming of multiprocessor systems and shared memory, especially taking coherence and memory modules into account.
• Construct effective synchronisation primitives.
• Account and argue for reasons that delimit computer systems with regard to bandwidth, energy consumption and cooling.
• Account and argue for design of multiprocessor systems of combination of SIMD and MIMD structures, especially with graphic processors and other types of accelerating devices, including programming them.
• Account and argue for desirable characteristics, such as energy efficient and reliable, and how they are achieved in embedded systems, PC and servers for commercial and scientific computing.

Content

The course has an emphasis on modern parallel computer systems (today multicores and graphic processors) and programming them. Areas covered are memory system organisation, principles for multiprocessor systems, and a detailed coverage of the design of a CPU.

Instruction

Lectures and labs.

Assessment

Assignments and written exam.