Which GPU card is best for InfoWorks ICM?

GPU cards come into play with the simulation of 2D meshes in InfoWorks ICM.  The biggest speed gain is when the majority of mesh elements are wet.  As the mesh dries out, the influence of the 2D calculations becomes less significant in the overall simulation, thus reducing the effect of the GPU card.

When an ICM simulation starts on a PC with a compatible GPU card, all the 2D calculations are pushed onto the GPU.  We have developed the Simulation Engine so it will work with all nVidia GPU cards that support CUDA technology and have a Compute Level of 2.0 or higher. Desktop GPU cards are significantly better than those found in Laptops/Mobile devices (those in a laptop typically have ‘M’ at the end of the number, so Quadro 4000M for example).  A list of current nVidia GPU cards that support CUDA technology can be found at -> http://developer.nvidia.com/cuda-gpus

GPU Meter is a very useful gadget for monitoring your GPU (graphics processing unit) or graphics card.

GPU Meter is a very useful gadget for monitoring your GPU (graphics processing unit) or graphics card.

Because of the vast range of cards available, we can’t be 100% sure that the whole range of Compute Level 2.0 GPU cards from nVidia do actually work with InfoWorks ICM, or indeed what speed gains they offer. However, we have conducted tests using TESLA K40, K20c, C2075, C2070 and C2050, Quadro K5200, M1000M, K1101M, 4000 and 2000, and GeForce GTX1080, GTX970, GTX960, GTX780, GTX690, GTX580, GTX560Ti, GTX460M, GT610 and GT440 cards, so we know for sure that they all work with InfoWorks ICM.The NVIDIA® Tesla® companion processors are built for GPU computing.

It’s the speed, memory bandwidth and number of cores the device has which determines how quickly your 2D simulation will complete on the GPU. The Tesla family is more thoroughly quality-controlled by NVidia and allows use of the “Tesla Compute Cluster” (TCC) driver which runs ~20% faster than other drivers and allows the Simulation Agent to be run as a Windows Service (i.e. allows GPU based Simulations to be run on a remote host).

The Tesla cards are the only ones that support double-precision computing.  The 2D engine does not currently support double-precision but were this to be introduced in the future, only Tesla cards would support it. The Quadro or GeForce cards are a lot less expensive than the Tesla cards while still offering good performance, but they can only be used when your 2D Simulation is run on the local PC and don’t support access via a Windows Remote Desktop Connection (Windows RDP connection).

Please note that your PC/Laptop must have the latest, and official, NVIDIA drivers installed for the GPU to be able to process your 2D models.  Generic drivers (commonly employed in corporate environments, where everyone is given a ‘standard’ install of Windows) will render the GPU card invisible to InfoWorks ICM.

We don’t currently support any of the ATi GPU cards.

It’s very important to note that the GPU card plays no part at all in regular 1D calculations. Also, these cards give no advantage for 2D calculations in InfoWorks CS or RS and we will not be updating either of these applications to detect or utilise GPU technology.

Share this post!

    About Andrew Walker

    Andrew Walker is a Senior Client Services Manager with Innovyze in the United Kingdom, specializing in the computerized analysis of drainage and flooding. He has over 30 years experience of modeling the key hydraulic processes involved in urban drainage design and analysis. He is one of the key members of staff tasked with supervising the roll-out and adoption of InfoWorks ICM throughout the UK and Europe.

    This entry was posted in Information Technology, InfoWorks ICM and tagged . Bookmark the permalink.