feat/36/add benchmarks

[cernicc]

Description

Added benchmarks for mater create.

Important points for reviewers

Used: https://docs.rs/criterion/latest/criterion/

Checklist

• Are there important points that reviewers should know?
  • If yes, which ones?
• Make sure that you described what this change does.
• If there are follow-ups, have you created issues for them?
  • If yes, which ones? / List them here
• Have you tested this solution?
• Were there any alternative implementations considered?
• Did you document new (or modified) APIs?

[th7nder]

benchmark setup looks great! got a question though, what do we do with it and when do we run it, should it be on the pipeline? what's the process?

[cernicc]

benchmark setup looks great! got a question though, what do we do with it and when do we run it, should it be on the pipeline? what's the process?

Good question. I am not sure. @serg-temchenko what is usually the process?

[serg-temchenko]

benchmark setup looks great! got a question though, what do we do with it and when do we run it, should it be on the pipeline? what's the process?

Good question. I am not sure. @serg-temchenko what is usually the process?

Considering that our proposal didn't contain a lot of information about this, I was thinking that the purpose is to run benchmarks with several types of files, such as small, medium, and large files. We should run each case multiple times, gather performance output, and include it in the documentation/README file of the crate.

Is it possible to do with this setup?

[th7nder]

looks solid! let's generate those benchmarks and add them to the docs.

[jmg-duarte]

If I try to run the benchmarks using cargo bench all tests are ignored. Tried a bunch of flags and nothing.

Please document how to run the benchmarks in the crate's README (and take the opportunity to make the specs list, an actual list, I f'd up when I wrote it 😅).

Do also run the benchmarks and document your findings in your PC — maybe @th7nder could also run them on his 24-core machine? 👀

[cernicc]

If I try to run the benchmarks using cargo bench all tests are ignored. Tried a bunch of flags and nothing.

cargo bench is correct. You have to wait for a few seconds at the beginning because the files are being generated. On my machine it takes ~5s for them to be generated before running the first benchmark

Please document how to run the benchmarks in the crate's README (and take the opportunity to make the specs list, an actual list, I f'd up when I wrote it 😅).

The spec list for the storage provider? Uff :D

[jmg-duarte]

cargo bench is correct. You have to wait for a few seconds at the beginning because the files are being generated. On my machine it takes ~5s for them to be generated before running the first benchmark

I'd wait, but the tests are ignored, what gives?

The spec list for the storage provider? Uff :D

I meant this:

-CARv1 specification: https://ipld.io/specs/transport/car/carv1/
-CARv2 specification: https://ipld.io/specs/transport/car/carv2/
-UnixFS specification: https://github.com/ipfs/specs/blob/e4e5754ad4a4bfbb2ebe63f4c27631f573703de0/UNIXFS.md
+* CARv1 specification: https://ipld.io/specs/transport/car/carv1/
+* CARv2 specification: https://ipld.io/specs/transport/car/carv2/
+* UnixFS specification: https://github.com/ipfs/specs/blob/e4e5754ad4a4bfbb2ebe63f4c27631f573703de0/UNIXFS.md

[jmg-duarte]

Looks good to me. From looking at the benchmarks I'm fairly sure that the filestore has a bug — i.e. doesn't de-duplicate blocks