bench(snapshots): mmap reader

[rkrasiuk]

Description

Tests for clarifying the functionality of memmap2 crate.

Benches for comparing the speed of simple file & mmap readers.

Benchmarks will be included soon

[codecov]

Codecov Report

Merging #4585 (caee69d) into joshie/nippy-jar (c8f6307) will decrease coverage by 52.15%.
Report is 2 commits behind head on joshie/nippy-jar.
The diff coverage is 0.00%.

Files Changed	Coverage Δ
crates/storage/nippy-jar/src/lib.rs	0.00% <ø> (-90.88%)	⬇️
crates/storage/nippy-jar/src/mmap/mod.rs	0.00% <0.00%> (ø)

... and 400 files with indirect coverage changes

Flag	Coverage Δ
integration-tests	16.62% <0.00%> (-0.04%)	⬇️
unit-tests	?

Flags with carried forward coverage won't be shown. Click here to find out more.

Components	Coverage Δ
reth binary	6.69% <ø> (-24.15%)	⬇️
blockchain tree	0.00% <ø> (-83.59%)	⬇️
pipeline	0.00% <ø> (-90.57%)	⬇️
storage (db)	16.66% <0.00%> (-58.97%)	⬇️
trie	0.00% <ø> (-94.85%)	⬇️
txpool	22.38% <ø> (-25.13%)	⬇️
networking	27.12% <ø> (-50.43%)	⬇️
rpc	25.73% <ø> (-31.64%)	⬇️
consensus	0.99% <ø> (-62.42%)	⬇️
revm	1.36% <ø> (-30.32%)	⬇️
payload builder	6.34% <ø> (ø)
primitives	22.50% <ø> (-64.01%)	⬇️

[joshieDo]

On the read and read_multiple we are doing .to_vec() to each value. Could we add/replace for one where we just return the reference or the decoded value from the ref? (either for Mmap and/or FileReader). Similar to read_mmaped_block_snapshots() test we're not actually decoding on the bench

Interesting to know how does it compare overall to querying them from mdbx

[rkrasiuk]

closing as we never intended to merge this bench