We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cond-mat

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo

Condensed Matter > Disordered Systems and Neural Networks

Title: Ianus: an Adpative FPGA Computer

Abstract: Dedicated machines designed for specific computational algorithms can outperform conventional computers by several orders of magnitude. In this note we describe {\it Ianus}, a new generation FPGA based machine and its basic features: hardware integration and wide reprogrammability. Our goal is to build a machine that can fully exploit the performance potential of new generation FPGA devices. We also plan a software platform which simplifies its programming, in order to extend its intended range of application to a wide class of interesting and computationally demanding problems. The decision to develop a dedicated processor is a complex one, involving careful assessment of its performance lead, during its expected lifetime, over traditional computers, taking into account their performance increase, as predicted by Moore's law. We discuss this point in detail.
Subjects: Disordered Systems and Neural Networks (cond-mat.dis-nn); Other Condensed Matter (cond-mat.other); Computational Physics (physics.comp-ph)
Journal reference: Computing in Science & Engineering 8, 41-49 (2006)
Cite as: arXiv:cond-mat/0507270 [cond-mat.dis-nn]
  (or arXiv:cond-mat/0507270v1 [cond-mat.dis-nn] for this version)

Submission history

From: Andrea Maiorano [view email]
[v1] Tue, 12 Jul 2005 14:07:20 GMT (17kb)

Link back to: arXiv, form interface, contact.