10 April 2021 - 16 April 2021
Table of Contents
Overcoming frustration
1. Santa Cruz Weather Station
Mast sensors still broken but started a test jig. Using a subset of 'HHEM' as a test program.
1.1. Next step
1.1.1. Construct DB9M test cable.
1.1.2. Test sensors.
2. EPICS
No progress. Now that the Brini Tracker A/D is working (which was the original motivation for getting U3 stuff working), this is no longer "critical path" - not like anything is critical in this list.
2.1. Next steps
2.1.1. u3scpi network port testing
2.1.2. initial EPICS interface
2.1.3. u3scpi analog/digital port reconfiguration.
3. Brini Tracker II
3.1. Fixed A/D
Readings were pegged because RIOT by default uses the 1V internal reference, but the Adafruit M0 board assumes an external 3.3V reference so it uses a 100k:100k voltage divider for VBAT measurements, which I in turn used for the VBUS measurement.
I don't want to change the RIOT default A/D configuration, so I am using a 470k:100k divider for both VBAT and VBUS and ignore the internal VBAT divider.
3.2. Fixed GPS NMEA reader
Essentially the ringbuffer/message passing algorithm between the UART RX and the NMEA parser was broken. Using tsrb (Thread Safe Ring Buffer!) instead of ringbuffer. Not tracking line lengths but explicitly using '*' and '\n' start and end line markers.
Time display now consistently updates.
3.3. Next steps
3.3.1. Validate remaining standard GPS messages
3.3.2. Investigate GPS chipset data logging
3.3.3. Investigate LoRA radio interface
4. Weber Grill Handles
Made new birch handles for Weber grill.
5. wktcolor
No progress.
5.1. Next steps
5.1.1. finish wktlib c++ export.
5.1.2. create vertex map for geometry importer
7. AMDGPU
For (likely) business reasons, AMD has chosen to prioritize non-APU devices for ROCm support. Such devices also require ACPI BIOS support, which may or may not be available. In my case, the CRAT table is incorrect according to the AMDGPU driver.
Either I need to wait to see if APU support will catch up, or construct a HSA compatible runtime based on the AMDGPU driver, or hack AMDGPU for APUs until it works with the released HSA runtime.
In either case, this will be a major time commitment if it goes forward.
7.1. Next steps
7.1.1. Reverse engineer HSAKMT programming.
8. GRASS geopackage import
Maureen asked me to evaluate the feasibility of importing one of her GIS projects to GRASS. It's possible if the database gets cleaned up as there were about 20 points with illegal coordinates.
DONE
9. Tessellation
Playing with 2x2 and 2x3 blocks, inspired by the back patio brickwork.
Questions:
What is the algorithm to fill a rectangular area with random tile placement?
What are the number of unique arrangements not counting rotations and reflections?
10. Ninja v Make
LLVM can use both Ninja and Make build systems with a preference for ninja. Started a simple measurement, but ran out of disk space(!).
10.1. Next Step
10.1.1. clean simple measurements
11. Plumbing
Santa Cruz kitchen sink has been slow draining, so I tried snaking it, and made it worse - now it doesn't drain at all. A real plumber has been called and he has been finding it a challenge. Trish called out her assumption that the plumber was going to be a man.
Likely, but maybe things will change eventually:
https://www.servicetitan.com/blog/women-in-plumbing
Hopefully Done