Programming with Computational Storage

With the exponential growth of stored data, there is a need for applications to process more data faster than before. Traditional CPU based architectures may run into resource limitations with such applications. Computational Storage solutions try to address these limitations in traditional models with innovative data processing paths. In this new model, compute is performed near data thereby overcoming existing CPU, memory and fabric limitations. The SNIA Computational Storage TWG is leading the way towards an application programming model while the NVMe CS TG is defining the device programming model at physical protocol layer. This presentation will provide details on how SNIA Computational Storage APIs may be used to connect and utilize a Computational Storage capable device. Learn how to discover compute capabilities, transfer data internally to compute and execute compute operations directly in device.

Download PDF