Status of the 0.8 teraflops supercomputer at Columbia

Arsenin, Igor; Chen, D.; Christ, Norman H.; Edwards, Robert G.; Gara, Alan; Hanson, Sheila J.; Jung, Chulwoo; Kahler, Adrian; Kennedy, A.D.; Kilcup, Gregory W.; Luo, Yubing; Malureanu, C.; Mawhinney, Robert D.; Parsons, J. A.; Sexton, J.; Sui, C.; Vranas, P.

doi:10.1016/0920-5632(96)00179-x

Cited by 7 publications

(4 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition to the four-dimensional nearest neighbor network described in the introduction, each motherboard has a DSP serial network [3]. Figure 1 shows these two networks, plus an example of a SCSI tree connection which links motherboards to each other and the host workstation.…”

Section: Global Architecturementioning

confidence: 99%

“…The initial design for QCDSP was described at Lattice '93 [1], details of the processing nodes, NGA design and features at Lattice '94 [2] and final NGA design issues, the networks in QCDSP and some details of the global architecture at Lattice '95 [3]. Since Lattice '95, the design for our processing nodes, motherboards, backplanes and crates has been completed and the design translated into working hardware, demonstrated at Lattice '96 in the form of a working 64 processor machine.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

QCDSP: The first 64 nodes

Mawhinney

1997

Nuclear Physics B - Proceedings Supplements

View full text Add to dashboard Cite

show abstract

Section: Global Architecturementioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

QCDSP: The first 64 nodes

Mawhinney

1997

Nuclear Physics B - Proceedings Supplements

View full text Add to dashboard Cite

show abstract

“…In this paper, we discuss the work of the QCDSP collaboration to build an inexpensive Teraflop scale massively parallel computer suitable for computations in (QCD) (see references [1], [2], [3], [4], [5]). A 0.4 Teraflop 8192-node machine is being built at Columbia University for a cost of about $1.8 million.…”

Section: Introductionmentioning

confidence: 99%

QCDSP

Chen

Christ

et al. 1997

Proceedings of the 1997 ACM/IEEE Conference on Supercomputing (CDROM) - Supercomputing '97

View full text Add to dashboard Cite

We discuss the work of the QCDSP collaboration to build an inexpensive Teraflop scale massively parallel computer suitable for computations in Quantum Chromodynamics (QCD). The computer is a collection of nodes connected in a four dimensional toroidial grid with nearest neighbor bit serial communications. A node is composed of a Texas Instruments Digital Signal Processor (DSP), memory, and a custom made communications and memory controller chip. An 8192 node computer with a peak speed of 0.4 Teraflops is being constructed at Columbia University for a cost of $1.8 Million. A 12,288-node machine with a peak speed of 0.6 Teraflops is being constructed for the RIKEN Brookhaven Research Center. Other computers have been built including a 50 Gigaflop version for Florida State University.

show abstract

“…and at SCRI (Edwards, Kennedy, and Vranas). The intent of this project was to develop a machine capable of a peak performance of 0.8 Teraflops/s and a true sustained performance of around 35% for QCD [66]. DOE has funded a program to build a machine capable of a peak performance of 0.…”

Section: Computer Architecture and Designmentioning

confidence: 99%

[Support of theoretical high energy physics research at the Supercomputer Computations Research Institute]. Final report, September 30, 1992--July 31, 1997

Bitar¹,

Edwards²,

Heller³

et al. 1998

Self Cite

View full text Add to dashboard Cite

DISCLAIMERPortions of this document may be illegible in electronic image products. Images are produced from the best available original document. Executive SummaryThis is the final report on grant DE-FG05-92ER40742 from the U S . Department of Energy supporting the research of the Theoretical High Energy Physics group at the Supercomputer Computations Research Institute (SCRI) at Florida State University. The research primarily involved lattice field theory simulations such as Quantum Chromodynamics (QCD) and the Standard Model of electroweak interactions. Although QCD is generally accepted as the theory which describes the strong interactions responsible for nuclear binding, convincing computations of the properties of the elementary particles from first principles are still elusive. The problem lies in the fact that for data at the "low" energies involved (-1 GeV), the coupling constant in QCD is large, preventing application of the physicist's usual tool, perturbation theory. Nonperturbative computations are necessary and they appear possible only via large scale numerical simulations. Especially simulations of "full QCD" , including the effect of light dynamical quarks, are extremely CPU time consuming. The scientists in SCRI's lattice gauge theory group have been in the forefront of such numerical simulations since the inception of SCRI in 1985.SCRI scientists have developed highly efficient algorithms and codes, which are fairly easily portable to various supercomputers, as well as workstations (used mainly for code development). In some cases the most frequently used code segments have been hand optimized for a particular computer architecture. With these codes they have performed numerical simulations of many interesting physics problems. Most of these computations were carried out on the CM-2 a t SCRI.Among the works completed by the members of the lattice group and their outside collaborators in QCD simulations are extensive hadronic spectrum computations with both Wilson and staggered fermions, and calculations of hadronic matrix elements and wavefunctions. Studies of the QCD p function with two flavors of Wilson fermions, and the study of a possible flavor-parity breaking phase in QCD with two flavors of Wilson fermions have been completed. Studies of the finite temperature behavior of QCD have also been a major activity within the group.Studies of non-relativistic QCD, both for heavy-heavy mesons and for the heavy quark in heavy-light mesons have been done. The latter resulted in a computation of the B meson spectrum and its decay constant.Combining large N analytic computations within the Higgs sector of the standard model and numerical simulations at N = 4 have yielded a computation of the upper bound of the mass of the Higgs particle, as well as the energy scale above which deviations from the Standard Model may be expected.A major research topic during the second half of the grant period was the study of improved lattice actions, designed to diminish finite lattice spacing effects and thus accel...

show abstract

Status of the 0.8 teraflops supercomputer at Columbia

Cited by 7 publications

References 3 publications

QCDSP: The first 64 nodes

QCDSP: The first 64 nodes

QCDSP

[Support of theoretical high energy physics research at the Supercomputer Computations Research Institute]. Final report, September 30, 1992--July 31, 1997

Contact Info

Product

Resources

About