White Papers

Extreme GPU Computing
Page 16
Figure 15: GROMACS power and relative power consumption compared to CPU-only
Power consumption is another critical factor to consider when using performance-optimized servers as
dense as the Dell PowerEdge C4130 with 4 x 300 Watt accelerators. Figure 15 answers questions about
how much power these platforms consume.
GROMACS shows the order of power consumption is as follows. B >> A >= C > D > E. In configuration B
(switched configuration) has an extra CPU more than Configuration A, and takes up considerably more
power than A and C when compared LAMMPS scores. This is because GROMACS uses the extra CPU in B,
while LAMMPS does not. Configuration A incurs a slight overhead of the switch and thus takes up
slightly more power than C. Configuration D is a dual GPU/dual CPU configuration and takes up more
power than E, which is a single CPU/dual GPU configuration
Overall, the acceleration in GROMACS is less than in LAMMPS and NAMD. This is due to the nature and
implementation of the underlying code and algorithms. The best result we see in GROMACS is a 3.3X
improvement in performance while talking up 2.6X more power. GROMACS seems to be sensitive to the
number of GPUs and CPUs and is able to utilize all available compute resources.
Four K80 Boards
Two K80 Boards