This PR adds a test demonstrating concurrent host<->mc communication. The goal is not maximizing throughput, but instead streaming data in a consistent and predictable fashion while also tolerating fluctuations in network latency. The test added here streams data from host->mc->mc->mc->mc->host, moving over circular buffers several times.
This required a few changes to the libraries.
First, we needed to implement the timeout functionality in bsg_manycore_plaform_receive.
Second, we need to handle the timeout error in hb_mc_device_podv_wait_for_tile_group_finish_any
Last, we need to expose hb_mc_device_podv_wait_for_tile_group_finish_any as a public function
WIP -- not ready for review yet.
This PR adds a test demonstrating concurrent host<->mc communication. The goal is not maximizing throughput, but instead streaming data in a consistent and predictable fashion while also tolerating fluctuations in network latency. The test added here streams data from host->mc->mc->mc->mc->host, moving over circular buffers several times.
This required a few changes to the libraries.
This relies on #808