Connected data, agreed state

Grape’s DAG organizes signed vertices with parent references and transactions. The graph supports network dissemination and deterministic ordering. Grape’s native agreement implementation then establishes the committed history that nodes use to update state.

Smart-contract execution uses Geth-based EVM components. The result is one canonical EVM state for accounts and contracts. Multiple graph paths do not mean conflicting account balances can all become final.

Agreement and finality

The selected consensus design is in the QBFT family, implemented through Grape’s own graph, wire, signer, driver, and persistence code. Commit certificates require strictly more than two-thirds of authorized voting power, under the design’s Byzantine-fault assumptions.

CometBFT experiments in repository history are comparison material, not the selected Grape consensus engine. Geth is a retained open-source execution dependency; Grape-owned consensus does not mean the complete software stack has no external dependencies.

What scalability means

Dissemination, ordering, execution, storage, and indexing have different bottlenecks. More nodes can improve distribution and resilience, but their advertised TPS cannot simply be added together to calculate total canonical execution throughput.

The public 2,700 TPS-per-node figure is a design specification, not a measurement of this live network. Use reproducible workload benchmarks, hardware, software revision, and finality criteria when comparing performance.