Computational Drug Discovery and Design

(backadmin) #1

2.1 Hardware To enhance the computing power, one of the means is to increase
the number of central processing units (CPUs) used for the calcu-
lation. Several possibilities exist, from Desktop station, containing
about 4–16 CPUs, to supercomputer with thousands of CPUs
[20]. Nowadays, the arrival in the market of the graphics processing
unit (GPU), a very accessible computing technology mainly
designed for gaming performances, has revolutionized the field of
computational sciences and especially numerical simulations. GPU
cards have accelerated calculations by several orders of magnitude
compared to CPU and have similar performances to supercom-
puter [21]. Moreover, several CPU cluster nodes can be replaced
by one GPU card, which allows for extensive and parallel calcula-
tions, and has the advantage to decrease power consumption com-
pared to CPUs. Desktop workstations equipped with a GPU card
have become a useful tool to carry out molecular dynamics simula-
tions, and clusters containing GPUs (allowing multiple GPUs to
carry out parallel calculations) open the way toward the simulation
of larger biomolecular systems. However, an additional effort
should be made in writing the code of the software in specific
programing language for GPU cards, called CUDA. A very recent
new Intel hardware, called the Xeon Phi, which is a Many-
Integrated-Core [MIC] in a single chip, has been recently pro-
posed. These multicore processors have the advantage to perform


Table 1
Various molecular dynamics methods, software implementation, hardware support


MD method Software implementation

Hardware
support
Conventional MD
(cMD)

AMBER [13], CHARMM[14], NAMD [15], Gromacs [16],
Desmond [17], ACEMD [18]

CPU/GPU

Accelerated MD
(aMD)

AMBER, NAMD CPU/GPU

Scaled MD AMBER, BiKi [19] CPU/GPU
Random accelerated
MD (RAMD)

AMBER, CHARMM, NAMD CPU

Steered MD (SMD) AMBER, CHARMM, NAMD, Gromacs CPU/GPU
Desmond CPU
Targeted MD (TMD) AMBER, CHARMM, NAMD CPU
Metadynamics (MTD) AMBER, CHARMM, NAMD, Gromacs, Desmond CPU/GPU
Replica Exchange MD
(REMD)

AMBER, CHARMM, NAMD, Gromacs, Desmond CPU/GPU

Simulating Annealing
(SA)

AMBER, CHARMM, NAMD, Gromacs, Desmond CPU/GPU

406 Sonia Ziada et al.

Free download pdf