System Thumbnail

Recommended Systems for Molecular Dynamics

Share:

We've taken the time to perform testing to determine what hardware not just runs your molecular dynamics applications, but gives you the peak performance that you can fit in a single box. Based on this testing, we have come up with our own list of recommended hardware - as well as specific configurations tailored with these recommendations in mind.

Overview of Recommended Hardware for Molecular Dynamics:

Processor (CPU)

The CPU and GPU are, by far, the most important components for these programs. Because these programs are so well threaded and parallel, the most important thing you can do is pick a CPU with a high core count. The more cores, the better performance, and the shorter time to run simulations. 

It is also worth noting that the more GPUs that are added to the system, the more CPU cores you will need as well. [More Information]

Video Card (GPU)

The ability of molecular dynamics programs to utilize GPU acceleration massively impacts our recommendation for GPUs in these systems. The release of the GeForce GTX 100 Series has made our GPU recommendations very simple.

Due to the speed of the GTX 1070 and 1080 video cards, the CPU has become the bottleneck for molecular dynamics applications. This means that, on a Single CPU system, there is no performance increase beyond a GTX 1070 and therefore we only recommend this card on the single CPU system.

On dual CPU systems, the program is able to squeeze out more performance from the GPUs and therefore we recommend either two GTX 1070s or two GTX 1080s.

[More Information]

Memory (RAM)

NAMD, LAMMPS, and GROMACS can be fairly memory hungry, so we are defaulting the single and dual CPU systems to 32GB and 64GB of RAM, respectively. If you know that the simulations you will be running are very large and/or complex, it is likely you will need upwards of 128-256GB. [More Information]

Storage (Hard Drives)

Storage is pretty straightforward for these molecular dynamics programs. To improve boot and load times, we highly recommend an SSD as your primary drive for your OS and NAMD installation. We also recommend a larger secondary SSD for storage of past projects and if you know you will be using VMD to create videos, animations, or movies of your simulations. [More Information]


Detailed Hardware Information:

Processor (CPU)

When it comes to CPUs there are two main specifications that define the capability of a CPU:

  • The frequency directly affects how many operations a single CPU core can complete in a second (how fast it is).
  • The number of cores is how many physical cores there are within a CPU (how many operations it can run simultaneously).

Whether a high frequency or high core count CPU is better depends on how well a program is designed to take advantage of multiple CPU cores (often referred to as multi-threading).

Fortunately, these molecular dynamics codes are incredibly well threaded. This means they can practically utilize as many CPUs / CPU cores as you can throw at them. Because of this, we recommend a CPU configuration that contains as many CPUs and/or cores as your budget allows. In the two graphs below (click on them to expand them), you can see how dramatically the time to perform a simulation is cut down as more CPU cores are added.

Single Xeon Performance by Core Count

(8-core E5 Xeon; 16-cores w/ hyper threading)

Dual Xeon Performance by Core Count

(10-core E5 Xeon; 40-cores w/ hyper threading)

It is worth noting that it may seem as though CPU performance begins to plateau around 16 total cores, but this begins to change as multiple GPUs are added because the GPUs need CPU resources. In fact, we recommend that you have at least 4-cores available for every GPU in the system.

Additional Resources:


Video Card (GPU)

This is where things get really interesting due to the fact that NAMD, LAMMPS, and GROMACS support GPU acceleration. So, as you would expect with GPU acceleration, performance increases nearly 6 fold when adding multiple GPUs to the system. What you wouldn't necessarily expect though, is which GPU we would recommend. Let's take a look at the GPU performance by card: 

Single Xeon Workstation -- Xeon E5 2690 V4 14-Core @ 2.6GHz

   apoa1 wall time  day/ns f1atpase wall time  day/ns  stmv wall time day/ns
CPU 31.3 0.629 88.0 1.83 297.2 6.33

Titan X

5.48 0.0896 17.0 0.278 52.5 0.944
(2) Titan X 5.29 0.0749 15.8 0.237 47.7 0.808
GTX 1070 5.25 0.0785 16.3 0.261 51.4 0.908
(2) GTX 1070 5.19 0.0708 16.1 0.238 47.9 0.811
GTX 1080 5.11 0.0731 16.1 0.243 48.5 0.831
(2) GTX 1080 5.13 0.0716 16.2 0.239 47.7 0.809

Dual Xeon Workstation -- 2x Xeon E5 2687W V4 12-Core @ 3.0GHz

   apoa1 wall time  day/ns f1atpase wall time  day/ns  stmv wall time day/ns
CPU 19.7 0.370 55.7 1.09 181.4 3.73

Titan X

5.67 0.0954 16.4 0.289 48.3 0.548
(2) Titan X 4.27 0.0593 11.9 0.168 36.7 0.548
GTX 1070 4.94 0.0757 14.7 0.246 56.0 0.796
(2) GTX 1070 4.19 0.0477 11.7 0.154 36.4 0.532
GTX 1080 4.45 0.0653 13.1 0.207 40.3 0.652
(2) GTX 1080 4.08 0.0472 11.7 0.147 35.4 0.504

You'll notice, on the Single Xeon Workstation, that there is essentially no performance increase beyond a single GTX 1070. On the Dual Xeon Workstation, you can see that there is a significant speedup going from a single 1070 to two 1070s but only a nominal speedup from two GTX 1070s to two GTX 1080s (for twice the cost).

This is why we recommend a single GTX 1070 for the Single Xeon Workstation and two GTX 1070s for the Dual Xeon Workstation.

For more information:


Memory (RAM)

Even moderate simulations require fairly large amounts of RAM. Because of this, we are starting RAM recommendations at 32GB for the small workstation and 64GB for the larger system. If you are working with very complex and/or large simulations, you may need as much as 128GB or 256GB of RAM.


Storage (Hard Drives)

With the falling costs associated with SSDs, we almost always recommend using an SSD for the primary drive that will host your OS and programs. The high speed of SSDs allows your system to boot, launch applications, and load files many times faster than any traditional hard drive.

We also highly recommend another, larger, SSD for storage of past projects as well as for any recordings and/or movies of simulations using programs such as Visual Molecular Dynamics (VMD).

 

See which Molecular Dynamics Workstation is right for you!