Table Of Contents

Benchmarks
- Simple Lennard-Jones fluid in 3 dimensions
- Supercooled binary mixture (Kob-Andersen)

Previous topic

Next topic

Formulae and units

Benchmarks¶

The benchmark results were produced by the scripts in examples/benchmarks, e.g.:

examples/benchmarks/generate_configuration.sh lennard_jones
examples/benchmarks/run_benchmark.sh lennard_jones

Simple Lennard-Jones fluid in 3 dimensions¶

Parameters:

64,000 particles, number density $\rho = 0.4\sigma^3$

force: lennard_jones ( $r_c = 3\sigma, r_\text{skin} = 0.7\sigma$ )

integrator: verlet (NVE, $\delta t^* = 0.002$ )

Hardware	time per MD step and particle	steps per second	FP precision	compilation details
Intel Xeon E5620	1.40 µs	11.2	double	GCC 4.4.1, -O3
NVIDIA Tesla S1070	58.6 ns	267	double-single	CUDA 4.2, -arch compute_12
	54.3 ns	288	single	CUDA 4.2, -arch compute_12
NVIDIA Tesla C2050	40.5 ns	386	double-single	CUDA 4.2, -arch compute_12
	34.6 ns	452	single	CUDA 4.2, -arch compute_12
NVIDIA Tesla S2050	46.4 ns	337	double-single	CUDA 4.2, -arch compute_20
	39.6 ns	395	single	CUDA 4.2, -arch compute_12
NVIDIA Tesla M2090	37.8 ns	414	double-single	CUDA 4.2, -arch compute_20
	33.2 ns	470	single	CUDA 4.2, -arch compute_12

Results were obtained from 1 independent measurement based on release version 0.2.0. Each run consisted of NVT equilibration at $T^*=1.2$ over $\Delta t^*=100$ (10⁴ steps), followed by benchmarking 5 times 10⁴ NVE steps in a row.

Supercooled binary mixture (Kob-Andersen)¶

Parameters:

256,000 particles, number density $\rho = 1.2\sigma^3$

force: lennard_jones with 2 particle species (80% $A$ , 20% $B$ )

( $\epsilon_{AA}=1$ , $\epsilon_{AB}=.5$ , $\epsilon_{BB}=1.5$ , $\sigma_{AA}=1$ , $\sigma_{AB}=.88$ , $\sigma_{BB}=.8$ , $r_c = 2.5\sigma$ , $r_\text{skin} = 0.5\sigma$ )

integrator: verlet (NVE, $\delta t^* = 0.001$ )

Hardware	time per MD step and particle	steps per second	FP precision	compilation details
Intel Xeon E5620	1.96 µs	2.00	double	GCC 4.4.1, -O3
NVIDIA Tesla S1070	68.7 ns	56.9	double-single	CUDA 4.2, -arch compute_12
	68.4 ns	57.3	single	CUDA 4.2, -arch compute_12
NVIDIA Tesla C2050	41.9 ns	93.3	double-single	CUDA 4.2, -arch compute_12
	35.0 ns	112	single	CUDA 4.2, -arch compute_12
NVIDIA Tesla S2050	44.4 ns	88.0	double-single	CUDA 4.2, -arch compute_20
	38.0 ns	103	single	CUDA 4.2, -arch compute_12
NVIDIA Tesla M2090	35.8 ns	109	double-single	CUDA 4.2, -arch compute_20
	29.6 ns	132	single	CUDA 4.2, -arch compute_12

Results were obtained from 1 independent measurement and are based on release version 0.2.0. Each run consisted of NVT equilibration at $T^*=0.7$ over $\Delta t^*=100$ (2×10⁴ steps), followed by benchmarking 5 times 10⁴ NVE steps in a row.