---
title: "Comment by Simonas Urbelis on \"NodeJS: 4.8x faster if we go back to callbacks!\" | Hashnode"
description: "I've modified the tests a little to be more like-for-like, added some alternative async/await implementations, and put them up on my github (can't link because"
url: "https://hashnode.com/posts/nodejs-347x-faster-if-we-go-back-to-callbacks/65c7473a9a6382054ca5b138/comment/65cb99a07d8386f7804806d8"
---

[Discussion](https://hashnode.com/posts/nodejs-347x-faster-if-we-go-back-to-callbacks/65c7473a9a6382054ca5b138)

Comment [NodeJS: 4.8x faster if we go back to callbacks!](https://hashnode.com/posts/nodejs-347x-faster-if-we-go-back-to-callbacks/65c7473a9a6382054ca5b138)

[Simonas Urbelis](https://hashnode.com/@Vez)

[Feb 13, 2024](https://hashnode.com/posts/nodejs-347x-faster-if-we-go-back-to-callbacks/65c7473a9a6382054ca5b138/comment/65cb99a07d8386f7804806d8)

I've modified the tests a little to be more like-for-like, added some alternative async/await implementations, and put them up on my github (can't link because I'm newly registered)

My main takeaways are:

-   The blocking promise version is now ~30% slower than the parallel one.
-   Parallel promises and async/await appear to be about 3.4-3.7x slower than the parallel callback version on my machine. Take the results with a massive pinch of salt since runtime variance is about 200ms for some reason.

This is a much older machine running Windows and doing 20% less iterations (due to open file limits on Windows) so the numbers are not quite directly comparable, but hopefully relative scaling remains mostly intact.

[Gemma Black](https://hashnode.com/@GemmaBlack)

[Feb 13, 2024](https://hashnode.com/posts/nodejs-347x-faster-if-we-go-back-to-callbacks/65c7473a9a6382054ca5b138/comment/65cb99a07d8386f7804806d8#reply-65cb9d8c77cedc99dbd59867)

Nice! Thanks [Simonas Urbelis](https://hashnode.com/@Vez). Thanks for checking through it too. That's a really interesting find.

I would love to be able to run the same code across different machines with different specs. My hunch was that the number of cores on a machine had a big effect on parallel performance too.

It's a shame you can't share your GitHub implementation. That would be lovely to see.

My aim, unless someone does it before me is to create a GitHub repo and be able to take advantage of their machines to run the tests and benchmark them too. That'll make the tests more fair. True.

Also, I have to decide whether I want to test just blocking or truly sequential. So as someone mentioned the `node:test` overhead, I may have to remove `node:test` completely so there are no promises or async awaits being used in the callback tests at all.

Thank you again for your comment. I truly appreciate it.
