multiprocessor systems presentation
TRANSCRIPT
Multiprocessor Systems
Objective
● Why Multiprocessors ? ● Basic Multiprocessor Architectures ● Multiprocessor Systems ● Crossbar Architecture ● Multistage Switching Networks ● Speedup in Multiprocessor Systems(MS) ● Performance characteristics of MS ● Limitations on Speedup of MS ● Multicore and Multithreading ● Summary
Semiconductor Transistor
Basic Multiprocessor Architecture
Multiprocessor Systems
•Loosely coupled
•Latency – milliseconds
•Closely coupled
•Multicomputer
•Latency – microseconds
•Tightly coupled
•Multiprocessor
•Latency – nanoseconds
Bus based architecture
•Bus contention limits the number of CPUs
• Lower bus contention
• Caches syn required
• Compiler places data and text in private or
shared memory
Crossbar
Can support a large number of CPUs Non-blocking network Cost/performance effective up to about 100 CPU – growing as n2
Multistage Switching Networks
● Lower cost, longer latency ● For N CPUs and N memories – log
2 n stages of n/2 switches
Speedup
Single Core
Multicore
Multithreading
Summary
● There are many forms of multiprocessor systems ● The system software to support them involves
substantial additional complexity over single processor systems
● The core OS must be carefully designed to fully
utilize the multiple resources ● Programming model support is essential to help
application developers
References: 1. Jeff Miller, Rolf Ulrich , Bettina Rolke: On the optimality of serial and
parallel processing in the psychological refractory period paradigm: Effects of the distribution of stimulus onset asynchronies, Cognitive Psychology 2009.
2. Jaejin Lee, Jung-Ho Park, Honggyu Kima, Changhee Jung, Daeseob Lima, SangYong Hana: Adaptive execution techniques of parallel programs for multiprocessors. In: Journal of Parallel and Distributed Computing, 2010
3. Ibrahim H. Önyüksel.: Throughput Analysis of Multiple-Bus Multiprocessor Systems with Simultaneous Possession of Common Resources.
In: Simulation May 1993. 4. Rajesh Sudarsan, Calvin J. Ribbens: Design and performance of a scheduling framework for resizable parallel applications. In: Parallel Computing January 2010. 5. ScienceDirect - Journal of Parallel and Distributed Computing #69 2009. 6. IBM Announcement Letters.