Crypto glossary
Parallel execution (parallel EVM)
Parallel execution lets a blockchain process transactions that don't conflict at the same time on several CPU cores, instead of strictly one after another.
The original EVM processes a block's transactions sequentially, so one powerful machine still uses a single core for execution. Parallel execution changes that. Some designs require transactions to declare in advance which accounts or objects they touch, as Solana and Sui do, so the runtime can schedule non-overlapping ones together. Others execute optimistically in parallel and re-run any transaction that turns out to conflict, the approach behind Aptos's Block-STM and several parallel EVM chains.
Parallelism is one of the main ways newer layer 1s and layer 2s claim higher throughput while staying compatible with existing tools. Chains marketed as parallel EVMs aim to run Ethereum contracts unchanged but faster, and Ethereum itself has explored block-level access lists, which record what each transaction touches, to let clients execute and validate blocks in parallel.
Speed-ups depend on the workload. When many users hit the same contract, such as a popular token launch or a single trading pool, transactions conflict and fall back to sequential processing. Parallelism also raises hardware requirements, and headline TPS figures rarely reflect real usage.
Example
During a quiet day, a parallel EVM chain spreads thousands of unrelated token transfers across cores; during a hyped mint, most transactions hit one contract and run in order.