Currently, deployment of hardware to F1 instances is disabled, so
reco run will always fail. This is due to an underlying framework we use, SDAccel, that is not yet supported on F1 instances. We’ve been working closely with Amazon in debugging their SDAccel support, but it requires a large revision to their FGPA tools. We expect the v1.3.0 release of their FPGA tools to provide support for SDAccel, which is already in preview.
Once the latest version of the FPGA tools are released, we will run our internal integration suite against them. Once this suite passes, we will move our build process over to the new version and enable deployments. We expect this to happen in our release scheduled for 7/27.