|
Fakultät für Informatik der Technischen Universität München Informatik X: Rechnertechnik und Rechnerorganisation / Parallelrechnerarchitektur Prof. Dr. Arndt Bode , Prof. Dr. Hans Michael Gerndt |
|
| This course is part of the CSE master program at TUM. | ||
Start: 25. October 2004
Parallel computers with a shared address space (Shared Memory Systems) simplify program parallelization since parallel processes or threads communicate by accessing the same shared memory locations. Systems with a physically shared memory do not scale beyond a small number of processors. Scalable shared memory systems are instead based on a physically distributed memory. The access time to memory locations varies in such systems for the processors. Therefore those machines are called non-uniform memory access systems (NUMA).
The lecture will cover concepts of system architecture as well as of programming those systems. It will introduce implementation techniques for hardware- and software-based scalable shared memory systems. Efficient programming requires careful program organisation to optimize data locality. Language concepts as well as compiler techniques helping the programmer in using those systems will be presented.
(More details to come)