| CPC G06F 12/08 (2013.01) | 20 Claims |

|
1. A computer-implemented method of virtual memory pooling, the method comprising:
identifying a plurality of graphic processing units (GPUs) of respective information handling systems, wherein each of the GPUs is associated with a respective internal memory allocation;
partitioning, for each GPU, the respective internal memory allocation associated with the GPU into a first memory allocation and a second memory allocation;
allocating, for each GPU, the first memory allocation of the respective internal memory allocation associated with the GPU as accessible only by the associated GPU;
pooling, for each GPU, the second memory allocation of the respective internal memory allocation associated with the GPU to define a virtual memory pool, the virtual memory pool accessible by each GPU of the plurality of GPUs;
processing, at a first GPU of the plurality of GPUs, a computational task, including:
accessing the first memory allocation associated with the first GPU;
determining that processing of the computational task exceeds a capacity of the first memory allocation of the first GPU; and
in response to determining that processing of the computational task exceeds the capacity of the first memory allocation of the first GPU, requesting access to the virtual memory pool.
|