Supercomputers comprise more and more processors and these processors are increasingly heterogeneous, with differing performance characteristics. The conventional programming models assume that all nodes run in lockstep. Thus, applications run at the speed of the least powerful processor. This paper introduces DesyncFS, a new programming model based on the block abstraction of traditional file systems. It virtualizes the performance characteristics of processors; this allows their heterogeneity to be hidden. The paper shows that DesyncFS allows cluster throughput to scale with average processor throughput instead of being limited by the slowest processor.
Did you find this white paper useful?
25 out of 50 users found this white paper useful
Related white papers
Major Animation Studio Speeds to Finish Line on Latest Feature Film With IBRIX
Digital effects houses and film studios have leveraged the computational power of high performance clusters for rendering and other complex jobs. Over time, the size and makeup of the clusters...
Planning Considerations for HiperDispatch Mode
For all levels of z/OS, a TCB or SRB may be dispatched on any logical processor of the type required (standard, zAAP or zIIP). A unit of work starts on...
Premier IT Magazine: Reinvented Transistors
45-nm Manufacturing Creating the Next Wave of Quad-Core Processors
Live Migration With AMD-V Extended Migration Technology
Virtual Machine migration is a capability being increasingly utilized in today's enterprise environments. With live migration, a Virtual Machine Monitor (VMM) moves a running Virtual Machine (VM) nearly instantaneously from...
Tuning Symantec Brightmail AntiSpam on UltraSPARC T1 and T2 Processor-Powered Servers
Electronic mail is a business-critical function in virtually every enterprise, and it is also one that is under constant attack. Well-known viruses such as Melissa, and worms like SoBig have...
AMD Stream Computing: Software Stack
Advanced Micro Devices, Inc. (AMD) - a leading global provider of innovative computing solutions - is working with other leading companies and academic institutions worldwide to deliver a complete, accelerated...
Cache Optimization for Coarse Grain Task Parallel Processing Using Inter-Array Padding
The wide use of multiprocessor system has been making automatic parallelizing compilers more important. To improve the performance of multiprocessor system more by compiler, multigrain parallelization is important. In multigrain...
