Cantera on accelerators

[Esteban Meneses-Rojas]

   Hi all.

   I was wondering if there is somebody looking at accelerate Cantera through CUDA kernels, OpenMP, or OpenAcc directives. I am planning on using Cantera on computers with either NVIDIA GPUs or Intel Xeon Phi coprocessors. Having Cantera extended to use accelerators would be very useful for me.

   Thanks.

   

[James Sutherland]

We have an implementation of a subset of cantera (ideal gas thermodynamics, mixture-averaged transport, and homogeneous kinetics) running on GPU (CUDA at the moment) as well as multicore.  It is tuned for performance on PDE solvers (e.g. it is vectorized).  We will be porting it to Xeon Phi in the coming 1-2 years.

We are currently seeing at least 10x speedups (some cases as high as 100x) on K20 cards over CPU speeds, and our vectorized CPU implementation is slightly faster than stock Cantera.

We don't yet have these results published (this work is very new), and are not yet ready to publicly release the code, but it will be open source (MIT license) and should be available in the coming year.  It is strictly a C++ implementation, geared toward HPC applications with PDEs.

Best wishes,

James

--

James C. Sutherland
Associate Professor of Chemical Engineering
The University of Utah
50 S. Central Campus Dr, 3290 MEB
Salt Lake City, UT 84112-9203
+1-801-585-1246
www.che.utah.edu/~sutherland

Schedule an appointment

--
You received this message because you are subscribed to the Google Groups "Cantera Users' Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cantera-user...@googlegroups.com.
To post to this group, send email to canter...@googlegroups.com.
Visit this group at http://groups.google.com/group/cantera-users.
For more options, visit https://groups.google.com/d/optout.