The parallel effective I/O bandwidth benchmark (b_eff_io) is aimed at producing a characteristic average number of the I/O bandwidth achievable with parallel MPI-I/O applications exhibiting various access patterns and using various buffer lengths. It is designed so that 15 minutes should be sufficient for a first pass of all access patterns. First results of the b_eff_io benchmark are given for the IBM SP, Cray T3E, Hitachi SR 8000, and NEC SX-5 systems, and a discussion follows about problematic issues of the current approach. The paper shows how a redesign of the time-driven approach allows for rapid benchmarking of I/O bandwidth with various compute partition sizes.
Did you find this white paper useful?
Related white papers
Next-Generation Optical Transmission Fibers
The first part of this paper gives a brief introduction of modern optical transmission fiber technology. The second part of this paper illustrates detailed numerical analysis, lab tests, and field...
Layer 3 Malformed Packet Testing in MCI's Public IP Network
Malformed packets associated with routing protocols can cause undesirable consequences if not detected and contained properly as close to the originator as possible. This issue magnifies in scope and complexity...
Network Intrastructure: Getting Started with VoIP
IP/Ethernet networks are fast becoming an attractive and viable utility for supporting high-quality voice in a converged network. Bringing disparate voice and data systems together into a multi-service network can...
Internet Acceptable Use Policies: Navigating the Management, Legal and Technical Issues
Many organizations have discovered the hard way that unrestricted and unmanaged Internet access by employees can lead to dire consequences in the form of wasted time, lost productivity, misappropriation of...
The Real Cost of Spam
Unsolicited Commercial E-mail, or spam, is not just an annoying problem. It's a costly one. This paper examines the different cost impacts, both direct and indirect, that spam has on...
The Three Components of Optimizing WAN Bandwidth
One of the truths in the industry is that few organizations pay much attention to proactive network management of any type. However, the proactive optimization of WAN bandwidth will quickly...
Next-Generation Architectures
This paper summarizes a new application technology in advanced optical transmission amplifier development and a series of successful field-trials for "hut-skipping" line architecture using hybrid Raman/Erbium-Doped-Fiber-Amplifiers (EDFA). The trials were...
