I'm not sure where you get that from? The benchmark code clones one read handle for each read thread at the beginning, which is then used for the entire benchma

I'm not sure where you get that from? The benchmark code clones one read handle for each read thread at the beginning, which is then used for the entire benchmark: https://github.com/jonhoo/rust-evmap/blob/d307999c1ad78d10ecb6ea911b6e94d36bc841ca/benchmark/src/main.rs#L142 https://github.com/jonhoo/rust-evmap/blob/d307999c1ad78d10ec...