Searcharxiv⌕ Search
Home/Archive/arXiv

arXiv · 0711.2034

Dynamic load-balancing on multi-FPGA systems: a case study

Volodymyr V. Kindratenko·Robert J. Brunner·Adam D. Myers
View original source ↗Read source document ↗Export metadata ↓

Abstract

In this case study, we investigate the impact of workload balance on the performance of multi-FPGA codes. We start with an application in which two distinct kernels run in parallel on two SRC-6 MAP processors. We observe that one of the MAP processors is idle 18% of the time while the other processor is fully utilized. We investigate a task redistribution schema which serializes the execution of the two kernels, yet parallelizes execution of each individual kernel by spreading the workload between two MAP processors. This implementation results in a near 100% utilization of both MAP processors and the overall application performance is improved by 9%.

Explore related subjects

astro-ph

Keep this discovery

BibTeXRIS

Volodymyr V. Kindratenko, Robert J. Brunner, Adam D. Myers. 2007-11-13. Dynamic load-balancing on multi-FPGA systems: a case study. https://arxiv.org/abs/0711.2034

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

KEEP EXPLORING

Related papers

arXiv2014

Interacting Phantom Energy and Avoidance of the Big Rip Singularity

Models of the universe with arbitrary (non gravitational) interaction between the components of the cosmic fluid: the phantom energy and the background, are investigated. A general form of the interaction that is inspired in scalar-tensor theories of gravity is considered. No specific model for the phantom fluid is assumed. We concentrate our investigation on solutions that are free of the coincidence problem. We found a wide region in the parameter space where the solutions are free of the big rip singularity also. Physical arguments, together with arguments based on the analysis of the observational evidence, suggest that phantom models without big rip singularity might be preferred by Nature.

astro-ph↗
arXiv2014

Exploring the Small Magellanic Cloud to the Faintest X-ray Fluxes: Source Catalog, Timing and Spectral Analysis

We present the results of a pair of 100 ksec Chandra observations in the Small Magellanic Cloud to survey high mass X-ray binaries (HMXBs), stars and LMXBs/CVs down to Lx = 4.3 x 10^32 erg/s The two SMC Deep Fields are located in the most active star forming region of the bar, with Deep Field-1 positioned at the most pulsar-rich location identified from previous surveys. Two new pulsars were discovered in outburst: CXOU J004929.7-731058 (P=892s), CXOU J005252.2-721715 (P=326s), and 3 new HMXB candidates were identified. Of 15 Be-pulsars now known in the field, 13 were detected, with pulsations seen in 9 of them. Ephemerides demonstrate that 6 of the 10 pulsars known to exhibit regular outbursts were seen outside their periastron phase, and quiescent X-ray emission at Lx=10^33 - 10^34 is shown to be common. Comparison with ROSAT, ASCA, XMM-Newton catalogs resulted in positive identification of several previously ambiguous sources. Bright optical counterparts exist for 40 of the X-ray sources, of which 33 are consistent with early-type stars Mv<-2, B-V<0.2), and are the subject of a companion paper. The results point to an underlying HMXB population-density up to double that of active systems. The full catalog of 394 point-sources is presented along with detailed analyses of timing and spectral properties.

astro-ph↗
arXiv2014

A Computer Program to Visualize Gravitational Lenses

Gravitational lenses are presently playing an important role in astrophysics. By means of these lenses the parameters of the deflector such as its mass, ellipticity, etc. and Hubble's constant can be determined. Using C, Xforms, Mesa and Imlib a computer program to visualize this lens effect has been developed. This program has been applied to generate sequences of images of a source object and its corresponding images. It has also been used to visually test different models of gravitational lenses.

astro-ph↗