Currently upodesh's Trie implementation uses a huge amount of heap allocation (around ~145MB) for only 3.8MB of words (159427 words in total). This is a huge overhead in terms of memory consumption.

Heaptrack Analysis

Running the Heaptrack analysis

Enable Debug symbols in release mode in Cargo.toml:

[profile.release]
debug = true

Compile the example in release mode:

cargo run --release --example example -- ami

Run the example inside heaptrack:

heaptrack ./target/release/examples/example ami