Skip to content
Josh Blum edited this page Jul 4, 2014 · 7 revisions

Users can create processing blocks for various kinds of DMA devices such as DSPs, PCIe boards, and GPUs. These devices tend to provide their own device-allocated memory. Pothos supports the integration of this device memory directly with the framework to minimize overhead and uncessary memory copies.

Details:

A interconnections of blocks can be distributed across multiple processes and remote hosts. The coding burden to the user to distibute blocks on a remote node is no different in practice than instantiating the block locally. The topology will automatically add network ingress and outgress blocks to the design to support the distributed topology.

Every feature available in the Pothos API is available in the PothosGui. With PothosGui, users can design processing topologies, deploy those topologies on local and remote nodes, monitor and debug the design, author new processing blocks within the GUI...

Pothos supports language bindings through its Proxy interface. Processing IP can be written in any language with the provided bindings. This IP can be connected and interfaced with in any other supported binding.

Current bindings are:

The following is a list of some of the advanced API features that Pothos can offer to authors of processing IP: