First, generate a serialized cachedb for in memory reads
cargo run --example load_db --features="example_utils"
Then, to execute the test, run the following:
cargo run --example parallel --features="example_utils"
To compare execution of sequential vs parallel for correctness:
cargo run --example compare --features="example_utils"
To test performance, add the --release
flag like such
cargo run --example compare --features="example_utils" --release
Profiling:
CARGO_PROFILE_RELEASE_DEBUG=true cargo flamegraph --root --release --example simple --features "example_utils"
Compared to a single-threaded REVM instance, Concerto performs 4x faster when running with 6 threads