Skip to content
This repository has been archived by the owner on Feb 1, 2019. It is now read-only.

Write benchmarks for existing code, e.g. blob serialization #84

Open
jamesray1 opened this issue May 25, 2018 · 3 comments
Open

Write benchmarks for existing code, e.g. blob serialization #84

jamesray1 opened this issue May 25, 2018 · 3 comments

Comments

@jamesray1
Copy link
Member

jamesray1 commented May 25, 2018

Is your feature request related to a problem? Please describe.

geth-sharding had results (see the below link) of 80000 ns/op roundtrip serializing/deserializing a blob to a collation body and back for their benchmark, while other results ranged from 100th to less than half of this. Note that we will need to do #78 first before this is done, so this might be a task suited for @tcsiwula when done with #78, if he's up for it?

Describe the solution you'd like

Write bench tests and compare with other results e.g. in the geth-sharding issue below, to decide whether further optimizations are needed. Writing bench tests are a nightly feature, see https://doc.rust-lang.org/stable/unstable-book/library-features/test.html FMI. If results indicate a large improvement is needed, e.g. worse than XML, then fixing this would be a more immediate priority, otherwise it could be postponed until e.g. post-production.

Describe alternatives you've considered

Additional context
prysmaticlabs/prysm#139

Cute Animal Picture

put a cute animal picture link inside the parentheses

@jamesray1
Copy link
Member Author

jamesray1 commented May 25, 2018

@samparsky, interested? On second thought, as noted above, this task needs #78 first which @tcsiwula is working on, so it would be best if he handled this as well.

@samparsky
Copy link
Collaborator

@jamesray1 Yeah sure but if @tcsiwula wants to handle it, okay

@jamesray1
Copy link
Member Author

Requires work on a bintrie and #78.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants