X
  • June 17, 2019

Accelerating DEM Simulations with Rocky on Oracle Cloud and NVIDIA

Arnaud Froidmont
HPC Solution Architect

It’s a sunny afternoon, you’re mowing your lawn, and the grass buildup in your mower disrupts your smooth progress. This disruption could have been avoided if the design process for your mower involved airflow modeling with particles. Similarly, you’d hope that the discrete element method (DEM) was used to simulate the flow of beans through the coffee machines that you trust to brew your coffee.

 

DEM simulation packages like Rocky DEM from ESSS include particles and can be coupled with computational fluid dynamics or the finite element method to improve results. However, they add a layer of complexity that results in increased simulation time. To speed up the simulation, Rocky DEM provides the option to parallelize to a high number of CPUs or gain even more speed by unleashing multiple NVIDIA GPUs in Oracle Cloud Infrastructure.

No special setup or driver is needed to run Rocky on Oracle Cloud Infrastructure. Import your model, choose the number of CPUs or NVIDIA GPUs, and start working. Using Oracle Cloud Infrastructure removes the wait time for resources in your on-premises cluster. It also avoids having people battle for high-end GPUs at peak times and then having those GPUs sit idle for the rest of the week.

“Oracle Cloud Infrastructure and Rocky DEM have collaborated to provide a scalable experience to customers with performance similar to on-premises clusters. The bare metal NVIDIA GPU servers, without hypervisor overhead, further help to tackle very large problems in a reasonable amount of time,” said Marcus Reis, Vice President of ESSS.

Depending on the simulation, Oracle Cloud Infrastructure provides different machine shapes to stay cost-effective without compromising on compute power. The following table shows the machine shapes suited for Rocky. Explore all the different storage options, remote direct memory access (RDMA) capabilities, and the composition of NVIDIA GPUs on our Compute service page.

 

Shape

CPU

GPU

VM.Standard2.4

4

-

BM.Standard2.52

52

-

BM.HPC2.36

36

-

VM.GPU2.1

12

1 x P100

VM.GPU3.1

6

1 x V100

BM.GPU2.2

28

2 x P100

BM.GPU3.8

52

8 x V100

 

“NVIDIA and Oracle Cloud Infrastructure are collaborating to help customers reduce their computation time from days to hours by providing GPUs for HPC applications. The Tesla P100 and advanced V100 GPUs increase customer productivity while reducing cost,” said Paresh Kharya, Director of Product Marketing, Accelerated Compute, NVIDIA.

The following chart shows that NVIDIA GPUs offer up to 6X better price-performance than CPU-based instances for this simulation. It also shows faster results at a similar price when switching from P100 to V100 GPUs or increasing the core count of CPUs.

 

Oracle Cloud Infrastructure provides bare metal instances with up to 8 Tesla V100 GPUs, and it’s making a difference. Companies can start thinking about the engineering and the design of their next product rather than worrying about simulation runtimes or compute resource availability.

Get started on Oracle Cloud Infrastructure, and run your Rocky DEM workloads today!

Be the first to comment

Comments ( 0 )
Please enter your name.Please provide a valid email address.Please enter a comment.CAPTCHA challenge response provided was incorrect. Please try again.