Information Technology Reference
In-Depth Information
5 Experimental Analysis
This subsection reports the experimental analysis of the proposed parallel algo-
rithm for distributed fluorescence analysis.
5.1 Development and Execution Platform
Fig. 5.
OurGrid platform is used to create a federation of volunteer computing nodes.
Nodes from Argentina, Brazil, Mexico, and Uruguay were assembled to support the
execution of the distributed version of MCell+Fernet.
The experimental evaluation was performed on a volunteer infrastructure with
computing resources from four Latin America research institutions: Universidad
de Buenos Aires (UBA) in Argentina, Universidade Federal de Campina Grande
(UFCG) in Brazil, Universidad Veracruzana (UV) in Mexico, and Universidad
de la Republica (UdelaR), in Uruguay, as it is shown in Fig. 5.
The volunteer computing infrastructure gathers a heterogeneous collection of
computing resources, including:
-
UBA, Argentina: IBM Express x3650 M4, Xeon E5-2620 12 cores, (2.0GHz),
128GB RAM, Gigabit Ethernet.
-
UFCG, Brazil: : Intel i5-3470S, 4 cores (2.9 GHz) and i7-2600, 8 cores
(3.40GHz), 8 GB RAM, Gigabit Ethernet.
-
UV, Mexico: Intel i5-3470S, 4 cores (2.3 GHz), 48 GB RAM, Ethernet.
-
UdelaR, Uruguay: AMD Opteron 6172, 24 cores (2.1 GHz), 24 GB RAM,
Gigabit Ethernet, from Cluster FING.
5.2 Problem Instances
The problem instances are inspired in the study of the dynamics of the binding
of a special molecule called
transcription factor
to a specific site (i.e.
binding
site
). Thus, three types of molecules exists in the system:
i
) TF: transcription
factor (most of molecules correspond to this type);
ii
) B (Binding site); and
iii
)
[TF-B] (binded molecule).
