JAM A-Z
summary
Proposes JAM (Join‑Accumulate Machine) as a domain‑specific successor to the relay chain that accepts rollup outputs, runs a two‑stage service model (refine then accumulate), and removes transactions entirely. Refine accepts up to 5 megabyte refine input per work item and compacts it to about 4 kilobyte refine output; accumulate is stateful and gets ~10 ms per output within a 6 s slot, with pipelining yielding ~3.5 s effective compute. Uses a RISC‑V based PVM, saffr consensus, five extrinsic types, a 1,200+ node “Jam toaster” test rig, and targets deployment in ~8–20 months.
AI-written summary · about
chapters (31)
- 0:00Introduction to Jam
- 1:30Jam computation model
- 5:37Service-based architecture
- 9:27Refine and accumulate functions
- 12:37No transaction model
- 14:18Security apparatus
- 16:59Pre-images and tickets
- 18:57Reducing opinionation
- 21:01Service examples and details
- 23:33Data processing limitations
- 26:26Contextual information
- 29:07Stateful accumulate functions
- 31:32Asynchronous service communication
- 33:31PVM based technology
- 36:44Advantages of PVM
- 39:26Block production algorithm
- 41:11Simplicity and implementations
- 43:20Polkadot one compatibility
- 45:55Authorization and optimization
- 47:44Network optimizations
- 50:01Pipelining block execution
- 52:36Prior state route advantages
- 54:34Chain upgradeability
- 57:36Token economics
- 1:00:06Jam toaster test rig
- 1:02:25Interaction with substrate
- 1:04:36XCM and Accords
- 1:06:10Exotic use cases
- 1:08:25Mixed resource consumption
- 1:11:21Future timeline
- 1:15:06Q&A and conclusion
