WIPL-D software is supported on Windows platforms: Windows 7, 8.x, 10 and Windows Server 2008, 2012 and 2016. The kernel can run on Linux platforms as well.
All products run on 64-bit and 32-bit systems, except for GPU add-on, which is supported only on 64-bit systems. Processors with multiple threads are supported by default, while cluster versions are custom-made upon request.
For more information about platforms and cluster options, contact our sales department.
Basic requirements for installation and run, in terms of memory, are determined according to the configuration of WIPL-D license:
- Basic - 1GB RAM and 300-1000MB HD space (depending on included add-ons)
- Standard - 2GB RAM and 300-1000MB HD space (depending on included add-ons)
- Premium - 4GB RAM and 300-1000MB HD space (depending on included add-ons)
When running a simulation, certain amount of memory is required depending on the problem size. To solve a problem with N-thousand unknowns, 8*N*N [MB] is required for single precision, and twice as much for double precision. For example, running a simulation with 15,000 unknowns takes up 1,800MB. If the computer on which the simulation is run has enough RAM to fit all the unknowns, simulation is performed in-core, otherwise the out-core simulation uses hard-disk space to store unknowns. The largest problem solvable by current version of WIPL-D Pro requires about 2TB of hard disk space.
To take advantage of GPU acceleration, one or more Nvidia's CUDA-enabled GPU cards must be used. We recommend GeForce, Tesla and Quadro products. For simulations performed in single precision our recommendation is usage of GeForce GPUs, such as Titan XP or GTX 1080 Ti since their price is lower than the price of Tesla GPUs. For simulations done in double precision Tesla-series GPUs should be used (Tesla K80 or P100).