Lecturer:   Graham Ellis 
                  

Lectures: Thursdays 12-1pm, room C219

COURSE AIMS:
1) Introduce students to comcepts underlying concurrent and high performance computing.
2) Ensure that students are comfortable working in a UNIX environment.
3) Develop basic communication/reporting skills.

ASSESSMENT:
There will be four assignments during the course (10% each for the first three and 20% for the last) and a two-hour written exam (50%) at the end of the course.
An (old)  marking scheme for assignments can be found here.
 

Lectures

• Review of shell scripting. (Lecture 1)
• Review of fork related commands; writing your own shell. (Lectures 2 and 3)
• Fork versus pthreads. (Lecture 4)
  
• Concurrent programming; deadlock. (Lectures 5 and 6)
  
• Serializable concurrent transactions & two-phase-locking. (Lecture 7)
  
• Smith-Waterman algorithm (Lectures 8 and 9)
  
• Parallel computing: OpenMPI (Lecture 10)
• Case studies: ParaGAP and HAP. (Lecture 11)
  
• Topological networks (torus, hypercube etc.) and their properties (bisection width, number of nodes etc.). (Lecture 12,13,14)
  
• Amdahl's Law; some parallel algorithms. (Lecture 15)
  
• More parallel algorithms. (Lecture 16,17,18)
  
• Exam preparation
  
• Final assignment worth 20%