Search NASASearch

NASA NTRS · 19980008108

Parallel supercomputing with commodity components

Abstract

We have implemented a parallel computer architecture based entirely upon commodity personal computer components. Using 16 Intel Pentium Pro microprocessors and switched fast ethernet as a communication fabric, we have obtained sustained performance on scientific applications in excess of one Gigaflop. During one production astrophysics treecode simulation, we performed 1.2 x 10(sup 15) floating point operations (1.2 Petaflops) over a three week period, with one phase of that simulation running continuously for two weeks without interruption. We report on a variety of disk, memory and network benchmarks. We also present results from the NAS parallel benchmark suite, which indicate that this architecture is competitive with current commercial architectures. In addition, we describe some software written to support efficient message passing, as well as a Linux device driver interface to the Pentium hardware performance monitoring registers.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Warren, M. S., Goda, M. P., Becker, D. J.. 1997-01-01. Parallel supercomputing with commodity components. https://ntrs.nasa.gov/citations/19980008108

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related reports

Performance and Facility Background Pressure Characterization Tests of NASAs 12.5-kW Hall Effect Rocket with Magnetic Shielding Thruster

NASA's Space Technology Mission Directorate (STMD) Solar Electric Propulsion Technology Demonstration Mission (SEP/TDM) project is funding the development of a 12.5-kW Hall thruster system to support future NASA missions. The thruster designated Hall Effect Rocket with Magnetic Shielding (HERMeS) is a 12.5-kW Hall thruster with magnetic shielding incorporating a centrally mounted cathode. HERMeS was designed and modeled by a NASA GRC and JPL team and was fabricated and tested in vacuum facility 5 (VF5) at NASA GRC. Tests at NASA GRC were performed with the Technology Development Unit 1 (TDU1) thruster. TDU1's magnetic shielding topology was confirmed by measurement of anode potential and low electron temperature along the discharge chamber walls. Thermal characterization tests indicated that during full power thruster operation at peak magnetic field strength, the various thruster component temperatures were below prescribed maximum allowable limits. Performance characterization tests demonstrated the thruster's wide throttling range and found that the thruster can achieve a peak thruster efficiency of 63% at 12.5 kW 500 V and can attain a specific impulse of 3,000 s at 12.5 kW and a discharge voltage of 800 V. Facility background pressure variation tests revealed that the performance, operational characteristics, and magnetic shielding effectiveness of the TDU1 design were mostly insensitive to increases in background pressure.

Performance

Performance of a Supersonic Over-Wing Inlet with Application to a Low-Sonic-Boom Aircraft

Development of commercial supersonic aircraft has been hindered by many related factors including fuel-efficiency, economics, and sonic-boom signatures that have prevented over-land flight. Materials, propulsion, and flight control technologies have developed to the point where, if over-land flight were made possible, a commercial supersonic transport could be economically viable. Computational fluid dynamics, and modern optimization techniques enable designers to reduce the boom signature of candidate aircraft configurations to acceptable levels. However, propulsion systems must be carefully integrated with these low-boom configurations in order that the signatures remain acceptable. One technique to minimize the downward propagation of waves is to mount the propulsion systems above the wing, such that the wing provides shielding from shock waves generated by the inlet and nacelle. This topmounted approach introduces a number of issues with inlet design and performance especially with the highly-swept wing configurations common to low-boom designs. A 1.79%-scale aircraft model was built and tested at the NASA Glenn Research Center's 8-by 6-Foot Supersonic Wind Tunnel (8x6 SWT) to validate the configuration's sonic boom signature. In order to evaluate performance of the top-mounted inlets, the starboard flow-through nacelle on the aerodynamic model was replaced by a 2.3%-scale operational inlet model. This integrated configuration was tested at the 8x6 SWT from Mach 0.25 to 1.8 over a wide range of angles-of-attack and yaw. The inlet was also tested in an isolated configuration over a smaller range of angles-of-attack and yaw. A number of boundary-layer bleed configurations were investigated and found to provide a substantial positive impact on pressure recovery and distortion. Installed inlet performance in terms of mass capture, pressure recovery, and distortion over the Mach number range at the design angle-of-attack of 4-degrees is presented herein and compared to that at 0- degrees, as well as the isolated inlet configuration to highlight installation effects. Performance of the installed inlet fell below that of the isolated inlet at Mach numbers of 1.4 and greater. The installed inlet demonstrated adequate operability over the expected range of angles-of-attack and yaw, but did exhibit definite angle-ofattack and yaw limits at supersonic conditions. At each supersonic flight Mach number, performance parameters near zero yaw angle were relatively insensitive to yaw, but in general the yaw angle yielding best performance was non-zero and varied with angle-of-attack. Performance of the installed inlet is also presented as functions of angle-of-attack and yaw to highlight these effects. Distortion at the aerodynamic interface plane ranged between 10 and 25% at the inlet critical points over the range of flight Mach numbers tested and did not decrease significantly for the isolated inlet. Although these distortion levels would be considered high for operation with a turbine engine, the over-wing installation is likely not as significant a contributor as the low test Reynolds number. This is demonstrated by comparing CFD analysis of the isolated inlet at test scale with that at intermediate and full scales.

Performance