White Papers

7 Deep Learning Inferencing with Mipsology using Xilinx ALVEO™ on Dell EMC Infrastructure
the performance and free the CPU. To accommodate the progress of ML technology, Mipsology
R&D expands the acceleration to new layers on a regular basis.
When in the course of a project a CNN evolves, it can be processed on Zebra on-the-spot once
trained, drastically simplifying the deployment of new CNN versions in data centers, at the edge,
on a desktop, or in embedded applications.
Once a CNN has been trained on GPUs or CPUs, it can be processed on Zebra as is;
eliminating the need for re-training and for new tools to migrate the neural network.
Zebra Works with Most Popular Neural Frameworks
Zebra is integrated in Caffe, Caffe2, MXNet, TensorFlow, PyTorch and ONNX to accelerate any
application without changing the sources code, without adding a proprietary API, and most likely
without even recompiling the application.
Zebra Accelerates CNN in Datacenters, on The Edge, in the Desktop, in Embedded
Applications
Whether installed on PCIe-boards designed for Data Centers or for desktops, or encapsulated
in devices for edge computing, the combination of Zebra with FPGAs accelerates any CNN
application.
The board configured with Zebra can compute a CNN in a data center when processing
massive amounts of data, or in the field for local processing. One-size-fits-all so there is no
need to duplicate the efforts designing ML for data center and for the edge.