The 1 Teraflops QCDSP computer
Robert D. Mawhinney
TL;DR
The paper presents the design, architecture, and current physics applications of the QCDSP computer, a cost-effective, massively parallel system optimized for lattice gauge theory simulations, achieving over 1 Teraflops in performance.
Contribution
It introduces a new massively parallel computer architecture tailored for lattice QCD simulations, demonstrating successful deployment and high performance.
Findings
Two large QCDSP machines are operational at Columbia and Brookhaven.
The systems achieve over 0.4 and 0.6 Teraflops performance respectively.
The paper details the design, software, and physics projects enabled by these computers.
Abstract
The QCDSP computer (Quantum Chromodynamics on Digital Signal Processors) is an inexpensive, massively parallel computer intended primarily for simulations in lattice gauge theory. Currently, two large QCDSP machines are in full-time use: an 8,192 processor, 0.4 Teraflops machine at Columbia University and an 12,288 processor, 0.6 Teraflops machine at the RIKEN-BNL Research Center at Brookhaven National Laboratory. We describe the design process, architecture, software and current physics projects of these computers.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsParticle physics theoretical and experimental studies · Advanced Data Storage Technologies · Scientific Computing and Data Management