Repository navigation
Replies: 1 comment
|
Summary and transcript are now in the top post. |
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Summary generated by AI from the meeting recording.
Post-Meeting Summary and Transcript
Attendees: Ryan (host), Tyler, Eco, Dylan, Susan, and Wandia. Referenced but absent: Jayr, Jeremiah, Jolade, Ronald, Johnny, Skot.
Summary
Rust Bitcoin Summit, Mining Illustrated, and a NYC workshop (00:00)
Ryan may represent Mujina at the Rust Bitcoin Summit in Nashville next week. Mujina uses rust-bitcoin for its basic types, so it is a user of the libraries the summit covers, along with others such as BDK and LDK. Ryan will also run a Mining Illustrated at the Chicago Bitcoin Collective in November; the collective will announce the date.
Tyler described the hashrate heating workshop in New York in November. PizzAndy is working with Harrison, who runs a hacker space there called Freedom Labs, on an educational event built around the 3D printer and filament dryer that PizzAndy and Aadhi heat with hashrate, followed by a meetup at PubKey (Telegram). The organizers asked Ryan and Tyler to help tell how the open mining stack fits in, and they may film the event. Ryan may bring the foundation's demo, depending on the travel budget.
Stratum V2 review and the prep PR (04:26)
Ryan posted his review of Jayr's SV2 support, #65, and will make sure Jayr has seen it. With a few changes, the PR is close to merging. SV2 will be marked as new when it first merges, in case it has unhandled corner cases. Tyler asked to be told when it merges so he can announce it. Ryan's fix to the SV2 working group's sv2-apps, stratum-mining/sv2-apps#922, was approved today.
Some changes in PR 65 touched code that SV2 shares with the rest of Mujina. Ryan found it easier to make those changes himself than to describe them, in #125. Jayr will rebase PR 65 onto it, and the two together will bring SV2 into the main line. Ryan asked for reviews of PR 125, Jayr's first, and for testing of Jayr's branch with a Bitaxe against an SV2 pool such as Braiins, CKPool, DMND, or the SRI pool in stratum-mining/sv2-apps.
Dependency removal and the Python prerequisite (07:59)
Susan's #118 follows her earlier CRC change, #103, and the dependency reduction discussion, #8. It replaces bitvec with Mujina's own code. Ryan will review it and thanked those who review others' PRs.
Wandia started on Mujina two weeks ago.
just checksfailed with her older Python, and the docs did not mention Python, so she added it to the prerequisites in #119. Ryan will review it. He suggested ajustrecipe that checks the prerequisites, as he recalls having in another repository.Fan control and the Bitaxe firmware docs (12:01)
Jayr's fan speed control for the Bitaxe on RHAP-D, #33, looks complete, but Ryan wants more testing before he merges it. Heatsinks and fan controller hardware vary between Bitaxe makers, and a bad fan controller can overheat a board.
Ryan merged #112, which points new users at the standalone RHAP-D firmware for the Bitaxe Gamma, rhapd-bitaxe-gamma (see also Set up a Bitaxe Gamma). His PR on the old repository, bitaxeorg/bitaxe-raw#19, does the same for anyone who follows old instructions there; that repository belongs to Skot and Johnny.
Malformed pool input and a backplane report (14:18)
#113 came from a red team finding about malformed input from pools. Jolade offered to take part of it, and Ryan will reply to welcome the PR. Jolade also reported in #122 that the backplane stops boards at startup when the only transport closes. Ryan will convert the discussion into an issue.
BZM2: an RDS for the foundation and the Bitaxe Bonanza (16:17)
Ronald revised his BZM2 PRs (#117, #120, and #121). He offered to send the 256 Foundation an Intel RDS if the foundation pays the shipping. Ryan will follow up, since running the code on the hardware will help him review and merge it.
Johnny posted #123 with his plan for running the Bitaxe Bonanza on the same BZM2 driver, as discussed on the last call, #115. The largest task is to combine Johnny's and Ronald's work into one driver for the chip. That cannot start in earnest until the next round of Bonanza boards reaches Johnny, Ryan, Skot, and perhaps Ronald. The Bonanza can then serve as a development board, as the Bitaxe Gamma does now.
S19 installers and test hardware (19:59)
The next milestones are installers for the S19 and the S19 board support merged into the main line. Once they merge, Tyler can announce them and call for testing. The first targets are the S19j Pro and S19k Pro with the Amlogic control board. Other hash board models on the same control board should need small changes. The Xilinx and BeagleBone control boards each need their own image (Xilinx work has started in mujina-xilinx-platform), and each image supports any miner that uses that control board.
Eco offered his S19a Pro (110 TH nominal, stock firmware, what looks like a Xilinx control board). Dylan has Xilinx and BeagleBone control boards on hand, and Ryan has a BeagleBone. Tyler saw pallets of machines marked for scrap at Compass's distribution center in Denver and offered to ask for boards from a wish list. Ryan will list the models and control boards, how common each is, and who has which.
RHAP-D instructions (26:22)
Dylan asked for instructions to install and run RHAP-D, like those for bitaxe-raw. Ryan pointed to Mujina's board documentation, which names the firmware for each board and links to the firmware's own build and install instructions, such as those in rhapd-bitaxe-gamma. He simplified those instructions, but the build depends on the local environment and on upstream servers, so he asked Dylan to report how it goes.
What to test now (29:08)
Tyler asked where people can help. Ryan named the SV2 branch and the Bitaxe fan control. After the S19 support merges, the new API and persistent configuration files from MIP-0001, 256foundation/mujina-mips#1, and Home Assistant integration will follow and need testing too.
BZM2 hardware for Kenya (31:05)
Susan asked whether Ronald could ship some of his BZM2 hardware to Kenya. Ryan suggested asking him in the 256 Foundation Telegram group, and Tyler asked Ronald there during the call (Telegram). Ryan suggested the open source lab that Gridless plans in Kenya as a permanent home for one or two units, since they take a lot of setup and power, and raised it in the same thread (Telegram). Bitaxe Bonanzas may follow.
Why
just checksfails on older Python (33:06)Susan and Wandia traced the Python failure behind PR 119 to Ryan's #109, whose crate age check needs a newer Python. Ryan asked them to add that to PR 119.
Dev kits and LibreBoard v3 (34:15)
Ryan asked about 256 Foundation dev kits. Eco said they are still in planning, and he wants to support them. He wants LibreBoards in a dev kit and would like to build them himself, but his pick-and-place machine must move out of its current warehouse first. Meanwhile, a third-party vendor will build a few LibreBoard v3 prototypes for validation, expected by the end of October or early November.
Action items
Transcript
Full transcript on the wiki: https://github.com/256foundation/mujina/wiki/dev-call-20261012
Agenda (as planned)
Agenda
Meeting: Monday 2026-10-12 at 1700 UTC (noon US Central), https://meet.jit.si/mujina-dev-call
Note: These meetings are recorded and transcribed by AI.
Purpose: keep everyone aware of what's happening, roughly following the development workflow. Not a replacement for the other communication channels.
Announcements & host topics
PRs
Jayr's SV2 support. Ryan's review asks for changes, mostly in the job source code that SV2 shares with the rest of Mujina. Schnitzel was to run it against an SV2 pool. (follow-up from the last call)
Ryan's prep for PR 65. Renames the coinbase and extranonce types to SV2's terms and lets an extranonce fill a slot of 1 to 32 bytes. Changes the extranonce bytes sent to V1 pools. PR 65 is to rebase onto it. Review wanted, Jayr's first.
Jeremiah's fix for the scheduler panic in issue 114. Jolade, new to the project, tested it and found that excess threads kept hashing stale work on a replacement job; Jeremiah fixed that. Ready for a final review?
Susan removes the bitvec dependency from the BM13xx protocol code, following discussion 8. winterrdog asked for a shorter commit message.
Wandia's first PR. Adds Python 3.11 or newer to the prerequisites in CONTRIBUTING.md, after
just checksfailed on macOS.Jayr's PI fan controller, still waiting on a test with a Gamma and the RHAP-D firmware. (follow-up from the last call) Ronald's RDS work (PR 121) has its own fan control; see issue 9.
Points the docs at rhapd-bitaxe-gamma. Merged. (follow-up from the last call)
Retires bitaxe-raw for the Gamma. Still open; Skot and Johnny were to settle the default branch. (follow-up from the last call)
Issues
How a job source reports upstream errors and latches faults. Jolade offers to take the job-gap watchdog part, so the connection recycles when no jobs convert, and has a fix with a regression test.
Jolade found that the backplane stops boards at startup when the only transport closes, as with the CPU miner in the Quick Start. A regression from the startup enumeration change, and it blocks PR 56. A fix with a test is ready to open as a PR.
BZM2 support
The BZM2 chip driver, independent of any board, now 22 commits that each pass CI. Jeremiah reviewed the single-commit version; Ronald folded in the fixes.
Four general fixes found along the way: a serial descriptor leak, fan targets routed through the API, a scheduler refactor, and an observer mode. Review only the last 4 commits.
Intel RDS platform support: the hashboard MCU, rail sequencing, self-test, calibration, and fan control. Review only the last 27 commits.
Johnny's plan for Bitaxe Bonanza support through RHAP-D, and how his BZM2 driver will align with PR 117. (follow-up from the last call)
Open floor
All reactions