Informations pratiques sur cette ressource

Description de la ressource pédagogique

Description (résumé)

Modern cryptocurrency systems, such as Ethereum, permit complex financial transactions through scripts called *smart contracts*. These smart contracts are executed many, many times, always without concurrency. First, smart contracts are serially executed before being recorded on the blockchain, and once recorded, they are serially re-executed whenever it is necessary to check that the blockchain is correct. Such serial execution limits system throughput and fails to exploit today's concurrent multicore and cluster architectures. This talk presents a novel way to permit miners and validators to execute smart contracts in parallel. A party attempting to add a smart contract to the blockchain executes multiple smart contracts speculatively, in parallel, thus ``discovering'' a serializable concurrent schedule for those transactions. This schedule is captured and encoded as a parallel but deterministic fork-join program used by later validators to re-execute those transactions concurrently and securely. Experiments show non-trivial speedup with even a small number of cores. Joint work with Thomas Dickerson, Eric Koskinen, and Paul Gazzillo.

"Domaine(s)" et indice(s) Dewey

• Distributed processing (004.36)
• informatique distribuée (005.276)

Thème(s)