This video documents a weekly meeting of the Open Research Institute, specifically focusing on progress and roadblocks within their FPGA development projects. Key discussion points include overcoming issues with a no-OS build approach, potential clock-related problems with the hardware, and the slow performance of their virtual machine build environments. Efforts to debug and improve the build process, including disk upgrades and code optimization, are detailed. The meeting also touches upon the "Opulent Voice" project, its receiver node, and challenges with RFIC configuration. Updates on the firmware build, PRBS generator testing, and the Xilinx AXI bus implementation are provided. Power output issues with the Pluto SDR are also discussed, along with preparations for Defcon, including a presentation on open-source synthetic aperture radar.
Summary:
00:00:13 Meeting Kickoff: The meeting begins with introductions and the standard agenda: discussing past work, planning for the next week, identifying resource needs, and addressing roadblocks.
00:01:01 No-OS Build Stumbling Block: Ken reports a significant issue with their no-OS approach, encountering script parsing problems during the build process that halt progress on a new server instance. He suspects a clock-related issue.
00:02:11 Unpredictable Hardware State: Ken highlights that some errors are not consistently repeatable and disappear after rebooting the card, suggesting a deeper issue, potentially with the clocking, which manifests as a lock status error for C-TSI lanes.
00:03:46 Opulent Voice Receiver Progress: The team aims to get the no-OS build working to test the receiver for the central node/spacecraft for the Opulent Voice transceiver. The polyphase filter bank logic is working well in simulation, but configuring the RFIC in the payload remains a challenge.
00:05:27 GitHub Repository and Firmware: Matthew reports good progress on moving code to the GitHub repository, making it easier to manage commits and build the project. Firmware builds for the Pluto are slow, but everything is generally looking good. Testing with a PRBS generator and digital loopback within the FPGA is yielding data, but not perfect results yet.
00:06:26 Xilinx AXI Bus Iterations: The main holdup has been the Xilinx AXI bus, with read/write issues in previous iterations. A third iteration using System RDL is showing promise, with reads and writes appearing to work correctly.
00:07:48 Next Steps for Opulent Voice: Future steps include achieving RF output and seeing a spectrum, then looping back to achieve zero bit errors. In parallel, DMA transmit and receive functionality will be worked on, leveraging the modem's loopback feature to avoid needing a complete RF path.
00:10:15 Virtual Machine Performance Issues: The performance of virtual machines hosting builds is significantly slower than expected, impacting build times for bitstreams, firmware, and XSA files.
00:11:44 Disk-Bound Bottleneck Theory: Paul suggests the VMs are likely disk-bound rather than CPU-bound, based on profiling. Additional disks are ready to be installed to improve throughput, but this requires a system downtime window.
00:12:53 CPU Speed and Cache Concerns: The possibility of CPUs not running at full speed due to VM environment issues is raised, but without concrete suggestions for improvement. Paul proposes proceeding with disk upgrades.
00:13:49 Disk Upgrade Proposal: Michelle proposes two options:
Option 1 (Downtime): Wait until after Defcon, take down Remote Lab West, perform disk upgrades, and test.
Option 2 (No Downtime): Move equipment from Remote Lab South to another location for upgrades, allowing continuous use of Remote Lab West but with continued slow performance there.
00:16:18 Defcon Preparations: Work is underway to finalize materials for Defcon (August 8-11), including a booth and a speaking slot on open-source synthetic aperture radar.
00:18:08 Pluto SDR Power Output Issues: Measurements indicate the Pluto SDR is outputting less than its rated +7 dBm, with observed levels around +3 dBm or +0 dBm. Efforts to maximize software output and ensure zero attenuation have not fully resolved this.
00:20:34 Register Map for Testing Software: Everist asks Michelle for a link to her testing software, as he cannot find it on GitHub. Michelle explains it's rough code adapted from an Analog Devices example and was shared on Slack. She acknowledges using a stale register map for Pluto msk and the need for updates.
00:21:15 Goals for Testing Software: Michelle aims to use the code for register access via I/O and DMA, exercise the Pluto over the air, integrate remote access functions (MQTT), and eventually develop a web interface with UI elements.
00:23:55 Defcon Details: The Defcon event will be August 8-11 at the Las Vegas Convention Center. The team will have a booth and a speaking slot on open-source SAR, addressing legal aspects and experimentation.
00:25:39 Meeting Wrap-up: The meeting concludes with thanks for the hard work and an announcement for a follow-up meetup later that day.