Article

Harnessing the Power of the Cloud: Running RS3 Models Faster Than Ever

Published on: Apr 18, 2024 Updated on: Apr 24, 2024 Read: 4 minutes
Authors:
  • Jim Zhao, Software Developer at Rocscience
  • Kien Dang, FE Group Manager at Rocscience

At Rocscience’s core, is our desire to innovate and develop features that will improve the user experience of our programs. One of our latest developments will deliver on this promise by allowing RS3 models to be run on the cloud.

What benefits does the Cloud offer RS3 users?

RS3 models can often be large and detailed, and as a result be time consuming to run. This is why powerful computer hardware is always recommended to speed up the process, however, this hardware comes at a cost. Even with powerful on-premises machines, could cloud computing offer a method to further improve the speed of running of RS3 models?

RS3 & Cloud Computing

We have created a platform in the cloud that runs RS3 engine models to drastically reduce computation time.

The two main benefits of leveraging the Cloud for computing RS3 models:

  1. Performance Improvements: Leveraging AWS’s powerful servers with Intel Xeon Processors and up to 512 GB of RAM, we have seen performance gains of 1.5-3x compared to some extremely powerful desktop machines. Since the Cloud also has a huge pool of machines, multiple models could be run simultaneously if desired.
  2. Convenience: You don’t have to purchase and maintain powerful hardware to run your RS3 models. Simply upload your compute file and the Cloud will take care of the rest.

Performance Improvements of the Cloud

To highlight the performance improvements of the cloud for RS3 computations, a test model was run with 2,383,317 elements and 9,737,631 degrees of freedom. This test compared the run time of on-premises machines with cloud virtual machines (VMs).

The results for the on-premises machines were:

Machine Type

Processor

Cores

Memory

Run Time

Dell XPS 15 9500

Intel I7-10875H

8

64 GB

9hrs, 37mins, 57secs

Dell XPS 8950

Intel i9-12900K

16

128 GB

5hrs, 52mins, 30secs

Custom

Intel i9-10980XE

18

256 GB

4hrs, 39mins, 15secs


The results for cloud VMs were:

Machine Type

Processor

Cores

Memory

Run Time

r6i.2xlarge

Intel Xeon 8375C

8

64 GB

6hrs, 10mins, 7secs

m6i.4xlarge

Intel Xeon 8375C

16

64 GB

5hrs, 35mins, 1sec

r6i.8xlarge

Intel Xeon 8375C

32

256 GB

3hrs, 2mins, 13secs


As you can see, one of the cloud machines ran the model in just over 3 hours, which was:

  • 1.5x faster than the custom desktop machine
  • 2x faster than the Dell XPS 8950, a powerful desktop machine
  • 3x faster than the Dell XPS 15 9500, a powerful laptop machine

The future of the Cloud for Geotechnical Analysis

This development showcases the power that cloud computing functionality can bring to the Rocscience software ecosystem. We will be starting to test out this functionality with RS3 users as 3D FEM analysis stands to gain the most from this technology, however, the possibilities are endless, and this capability could be extended to other programs in the future.

Are you an RS3 user that is interested in leveraging cloud computing for your models? If you have got a large model and would like to see if this functionality could help speed up your analysis, contact us here, and we would be happy to talk with you.

    Back to top