I recommend creating a slightly larger array that mirrors actual workloads (e.g. 4k or 8k values, maybe just repeating these values multiple times).

Using a small array like may amplify the overheads of function invocation compared to the actual work being done

I think we should put this along side the other benchmarks for cast in https://github.com/apache/arrow-rs/blob/main/arrow/benches/cast_kernels.rs#L1-L0

and it turns out that we already have tests for this that I can easily modify 🙄

    c.bench_function("cast decimal128 to decimal128 512", |b| {
        b.iter(|| cast_array(&decimal128_array, DataType::Decimal128(30, 5)))
    });