A simple formula to determine the number of tokens required to run Abaqus on N CPU's or cores is provided by:
int(5 x N^0.422). Enter the number of desired cores in the form below and the number of tokens required will be returned:
GPU (also referred to as GPGPU) behaves as a coprocessor to a CPU and therefore requires a minimum of one CPU core. When determining token usage for Abaqus utilizing a GPU, count the number of desired CPU cores and then count a single-GPU as a single core.
1 GPU = 1 CPU core (5 tokens) + 1 GPU (1 token) = 6 tokens
As an illustration, to run Abaqus Standard on a single-GPU, the total token count required is 6 which is equivalent to 2 cores. Minimum GPU requirement is CUDA Compute Version 3.0 or higher. Abaqus Certified GPU List
GPU's have revolutionized high-performance computing (HPC) for Abaqus simulations. A significant economic advantage of GPU computing for Abaqus is that a single-GPU which contains hundreds or thousands of CUDA cores will consume one token while a CPU with only a few cores will each consume a token. With fewer tokens required to achieve the same simulation throughput when compared to a pure CPU-based system, overall cost can be reduced. For more information including the benefits of GPU computing, visit Abaqus performance on Nvidia GPU's.