Distributed Computing under Linux page （在Linux下分布式计算页面）.pdf

Distributed Computing under Linux Lars Lindbom and E. Niclas Jonsson Division of Pharmacokinetics and Drug Therapy Department of Pharmaceutical Biosciences Uppsala University Overview • What is Distributed Computing? – Enabling distributed computation • What are the potential benefits of DC in population PK/PD? – Example • Bootstrap • Example of a small cluster • Conclusions What is Distributed Computing? • It is using two workstations to solve one task and gaining 100% in speed • It is using 1000 processors (CPUs) to solve a task that otherwise would have been (practically) impossible to solve. • It is using available resources more effeciently Enabling Distributed Computation Hardware Personal Network Computers Fileserver, Operating Tape backup System Cluster Computing Nodes Methods (Software) Enabling distributed computation There is more to distributed computing than connecting computers to a common network Hardware Operating • Making the computing nodes talk to each other System and cooperate Methods • Identifying the parts of a scientific problem that (Software) can be parallelized What are the potential benefits of DC in population PK/PD? • We rely on computers to fit models to data. – Preferably, this task should be written for the potential use of multiple processors – No software support this today • There are other tasks within a population analy