วีดีโอ

Because it's been a while since I made the video.

Thanks man. I inherited the codebase. Your points though are valid and useful.

That's a good idea. Thanks man.

@@Vagelis_Prokopiou My assumption is that in real world scenarious .NET Core can potentially be even sometimes faster when more advanced optimizations are applied, like code generators, like Spans, like FrozenDictionary etc. Not saying that it's not possible to do with Rust, but some things, like for example Regex and Json code generators is something that you likely won't try to manually implement if it's not provided via the standard means already.

Hi man. Maybe. But my personal preference is Rust. I have used both languages and the feeling of robustness I get from Rust is golden. And it is backed up by the result.

@@Vagelis_Prokopiou Sure, just saying that MS is putting a lot of effort into optimizing .NET Core for web APIs usage scenarios. When in really performance critical scenarios we now have .NET Core Minimal APIs, we have data structures optimized for reads (like FrozenDictionary), we have fast stack-based Spans which are now used for common substring operations, we have code generators and so on. Rust is not as widely used for the web API scenarios, so it may not have the same kind of tooling as of yet (correct me if I am wrong).

Well, I am kind of sceptical, since .net core still has the overhead of the runtime and the GC. Rust uses LLVM which is very good at optimizing code. In any case, a benchmark can provide insight. If you want you get setup a GitHub repo with some API, and I will reproduce it in Rust. And then we can check.

It will be the best decision of your life.

@@Vagelis_Prokopiou It would make sense tbh, because my API's serves as a Backend for multiplayer game, where speed is obviously cruical. Thanks for great comparison, I will start re-writing my API asap

Cool. Thanks for the feedback

Τhanks for the feedback man.

Thanks for the feedback man.

You are welcome man.

🤔 I am not sure you would ever do that if you used Rust, so why do it in the benchmark?

I don't think we want the overhead of the network. We want pure code performance. In any case, the setup is the same for both APIs, so we have the same comparison baseline.

Thanks for your valuable input.

Did you also tested the Rust API with the max connections config?

@@Vagelis_Prokopiou After increasing the number of Gunicorn workers as well, Rust is still about 5x - 7x faster than FastAPI on my computer. This result seems more reasonable to me.

@@tdd44 Rust is expected to be faster. The issue is that even with the db overhead, the difference is substantial.

Thank you too man.

found out in the #2 video, phenomenal!

Can you provide the link?

@@Vagelis_Prokopiou th-cam.com/video/bXSHGnb4YoU/w-d-xo.html

​@@Vagelis_Prokopiou th-cam.com/video/zWLsa-TEs2s/w-d-xo.htmlsi=tqL6He_SiG7AClQS

such bullshit 😂

@@Vagelis_Prokopioujust search up Phil Jackson on Michael Jordan in conversation with John Salley live talks Los Angeles LiveTalksLA and go to 2:09-3:00 and you’ll hear Phil Jackson admit Michael Jordan was worried about winning scoring titles

Thank you Marcelo.

Hi @YourComputer. Thanx for your feedback. The objective of this benchmark was not fairness. The objective was to compare these 2 off the shelf solutions and help "you" make educated decisions, based on the comparison data, going forward.

@@Vagelis_Prokopiou Educated decisions tend to require fairness (albeit, decisions are often the result of politics).

I believe the point you raise is not valid because no one in the industry hand writes his db related code when he chooses to use Laravel. But anyone is entitled to his opinion and everyone can reach his own conclusions. So your feedback is of course welcome.

@@Vagelis_Prokopiou And yet, you likely wrote your own db related code for rust, which would have been a very trimmed-down version in comparison to Laravel. If we're simply speaking within the context of web APIs (especially if the decision is not up to the developer, which many times is not up to the developer), assuming that everyone would use Laravel for web API development for PHP would be false. Since all you were doing for Laravel was filling an array with 100 instances of `User`s and echoing the array in JSON format, I decided to do the same but without bloat of Laravel, and even ran it through PHP's built-in server. Running the same benchmark I get: Concurrency Level: 1000 Time taken for tests: 2.247 seconds Complete requests: 50000 Failed requests: 0 Total transferred: 452000000 bytes HTML transferred: 443850000 bytes Requests per second: 22248.29 [#/sec] (mean) Time per request: 44.947 [ms] (mean) Time per request: 0.045 [ms] (mean, across all concurrent requests) Transfer rate: 196410.66 [Kbytes/sec] received

@@YourComputer Nice 🙂 My objections are: 1) I did not write any db related code for Rust. The db code is provided off the shelf. 2) The issue is not what I am doing with Laravel. After all, I am doing the same thing for Rust. The issue is what is the performance of Lavarel under the given conditions. 3) Your setup is not Laravel, so it is irrelevant to this specific benchmark. 4) No one runs a PHP application through the PHP's build-in server in production. Therefore, your setup is irrelevant to this specific benchmark. Apart from that, thanx for the benchmark info. It is interesting indeed.

Hi @rallealyt. Thanx for your feedback. Yes, I agree. A lot of things must be considered before commiting to a certain solution.

Hi Luke. Nice observation. I don't have any opinion on this. I have to research it.

Να σαι καλά :-)

Hi man. Whatever testing, should be done before the code reaches production. Production is never disrupted, apart from the deployments.

Glad it was helpful man.

Hi man. You can check the paper by the creators themselves. It is pretty descriptive. www.openbsd.org/papers/bcrypt-paper.pdf

​@@Vagelis_Prokopiouthanks a lot sir, this paper is very helpfull for my kriptonalisis study 👍

You're welcome man

Thank you too for the feedback 🙂

Happy to help Dario 🙂

Thanks for the feedback Farzher. Do you have some other workload in mind that would be more appropriate?

@@Vagelis_Prokopiou yeah, simply make an endpoint that returns Hello World. that way you're only benching the web server framework

The issue is that I don't want to bench web frameworks, but language efficiency for a given workload. It could be something non web related for that matter. I just chose a web related workload because the web is such a fundamental "thing". Therefore I don't think that what you propose can provide us with enough valuable feedback. Maybe calculating something else or rendering something could be a viable alternative.

@@Vagelis_Prokopiou yeah idk depends. if you're doing light work like serving a cached file/string then that's basically equal to Hello World and i just want to see how much overhead the web framework it using. how many concurrent requests will it let me do, with how much latenchy

@@stephenkamenar Well, that's a different comparison. I am doing "languages" comparisons so far, but you want "web frameworks" comparisons. I will make a video describing this difference and asking for feedback. If I get enough feedback I will do the "web frameworks" comparisons too.

Hi man. I haven't to check to tell you the truth. I think I have a repo for it.