What cards for Ansys? Hardware questions

Hi everyone,

I am totally new to CUDA and parallel processing but am interested in it for a program I use. I am looking to build a new Workstation for use with Solidworks and CATIA design software, as well as Ansys simulation and analysis software.

Next year’s release of Ansys 13, boasts that it supports GPU based Parallel processing, and I am hoping someone here can tell me if that means I need specifically the Tesla cards, or could I use several Geforce cards as processing mules?

Thanks,
Jim

Hi everyone,

I am totally new to CUDA and parallel processing but am interested in it for a program I use. I am looking to build a new Workstation for use with Solidworks and CATIA design software, as well as Ansys simulation and analysis software.

Next year’s release of Ansys 13, boasts that it supports GPU based Parallel processing, and I am hoping someone here can tell me if that means I need specifically the Tesla cards, or could I use several Geforce cards as processing mules?

Thanks,
Jim

From a technical perspective, I don’t see why a GeForce would not work as the compute features are very similar to a Telsa. You should check with Ansys to make sure they haven’t restricted it to Tesla and/or Quadro for some marketing reason.

From a technical perspective, I don’t see why a GeForce would not work as the compute features are very similar to a Telsa. You should check with Ansys to make sure they haven’t restricted it to Tesla and/or Quadro for some marketing reason.

Jim,

I’m at NVIDIA and managing our alliance with ANSYS and was involved in their R13 GPU develpments. ANSYS Mechanical is available today for GPU acceration but one canot use Geforce cards because memory and (essiential) double precision performance are insufficent. ANSYS recommends either a Tesla C2050 (3GB) or C2070 (6GB) if you already have an existing GPU for pre/post tasks – or a Quadro 5000 (2.5GB) or 6000 (6GB) if you are purchasing a new system and want a single GPU for both compute and visualization (pre/post + CAD) tasks. Please contact me at sposey@nvidia.com if you wish to learn more about the performance on Tesla and Quadro.

Stan

Hi,

I just wanted to ask which Quadro Produkts are supportet to Ansys 13. I’m working at university with the Release 13 and we do have several Quadro cards. The biggist available for me is one of the 3000 Series. Are those cards sufficient for Cuda in Ansys 13?

Thanks,

Stefan

Stefan,

Unfortunately Quadro 3000 does not have enough memory or double precision performance to provide a meaningful solution for anything computational such as finite element analysis. I would characterize the Quadro 3000 as a very good solution for high-end visualization.

Good ANSYS model candidates for GPU computing would include on average ~1M DOF and above – although DOF alone is not the proper guide. Below that level, a model may not require enough “computational work” to benefit from the computational costs to move the work from the CPU to the GPU for acceleration.

Stan

Hi everyone.
Does anybody know, are mobile Quadro GPUs (5000M, K5000M and similar 6000) supported for ANSYS computing?

Hi,

As you know that Ansys gives out HPC packs for upscaling the usage of CPU cores. So can anyone tell me if there is anything similar with regard to the GPU cores?

Thanks