In SMP mode, each compute node executes a single task with a maximum of four threads, with node resources including memory and network bandwidth shared by all processes.
This solution is useful in SMP structures design and optimization (e.g., actuators made of SMP beams) where a large number of simulations is required.
It was found that carbohydrates in SMP, proteins in SMP and humics in SMP were decreased in the dual-MBR compared to the control-MBR.
The PDJDS/CM-RCM reordering method provides excellent vector and parallel performance in SMP nodes.
Figure 3a shows the speedup observed with the 34 k ZINC library in SMP and VN modes (gm = reuse).
Gel filtration chromatography (GFC) analysis showed that the substances with macromolecular weight in SMP were reduced in the dual-MBR.
Similarly, average log file writing time was reduced by 92% with grid map reuse, and over 99% in SMP mode (Table 1).
Multithreaded execution in SMP mode further reduced docking time by 10%, for an overall improvement of 25% over VN OMP = 1, gm = reload) (Figure 2a).
It is indicated that the recovery stress of CNT/SMP nanocomposites with only 3.3% weight fraction of carbon nanotubes will reach almost twice of that in SMP bulk.
Three-dimensional excitation emission matrix (EEM) fluorescence spectra demonstrated that the aromatic protein-like substances in SMP were also lowered in the dual-MBR.
This study proposes a unique approach to mechanically drive recovery in SMP networks using external forces to facilitate shape change in a material with stored strain.
