Blue Gene: A vision for protein science using a petaflop supercomputer

Abstract

, IBM announced the start of a five-year effort to build a massively parallel computer, to be applied to the study of biomolecular phenomena such as protein folding. The project has two main goals: to advance our understanding of the mechanisms behind protein folding via large-scale simulation, and to explore novel ideas in massively parallel machine architecture and software. This project should enable biomolecular simulations that are orders of magnitude larger than current technology permits. Major areas of investigation include: how to most effectively utilize this novel platform to meet our scientific goals, how to make such massively parallel machines more usable, and how to achieve performance targets, with reasonable cost, through novel machine architectures. This paper provides an overview of the Blue Gene project at IBM Research. It includes some of the plans that have been made, the intended goals, and the anticipated challenges regarding the scientific work, the software application, and the hardware design.

The Potential of On-Chip Multiprocessing for QCD Machines

“…In this paper, we investigate whether and when, with the evolving VLSI technology, increasing chip size (in square feature sizes) will yield diminishing returns 4 . We reach the following encouraging conclusions regarding QCD machines: Even at modest frequencies, say 20% of state of art, the above figures appear attractive 5 . The reminder of this paper is organized as follows.…”

“…Announced in 2005 are products such as the IBM/Sony/Toshiba Cell capable of 64 flop/cycle at about 4 GHz [24] and the ClearSpeed CSX600, capable of 192 flop/cycle, at 250MHz [16]. A number of research projects have also been active for a few years, such as BlueGene/Cyclops (64 flop/cycle at 500 MHz) [5] and TRIPS (targeting 5 Tflops in a 35 nm VLSI implementation) [36].…”

“…4 Beyond this point, further increases of computational power may still be achievable, by increasing clock frequency and by assembling systems with larger numbers of chips. 5 contribution of this section is the identification of schedules that reduce I(n, m). The analysis precisely determines constant factors to enable finite-horizon considerations, in addition to asymptotic ones.…”

Implications of a PIM architectural model for MPI

“…Other experimental systems have still relied on a conventional CPU for overall control or centralized processing (e.g. DIVA [11]), or looked to the outside world as an SMP on a chip with message-passing link interfaces between chips (Blue Gene [1]). This paper assumes a newly emerging view of PIMbased supercomputers (Figure 1) [13,15,7].…”

The BlueGene/L supercomputer

“…IBM has previously announced a multi-year initiative to build a petaflops scale supercomputer for computational biology research [1]. The BlueGene/L machine is a first step in this program [2].…”