this post was submitted on 17 May 2025
45 points (100.0% liked)

It's Foss News

147 readers
27 users here now

RSS News community for It's Foss News.

If you dislike RSS Bot communities please block this one.

It’s FOSS is an award-winning web portal that focuses on Open Source in general and Linux in particular.

It’s FOSS was founded in mid-2012 as a hobby project to help desktop Linux users and spread awareness about Linux and open source. As the popularity of It’s FOSS grew worldwide, it was incorporated as an independent company CHMOD777 Media Tech in mid-2018. The term chmod 777 is inspired from the Linux command chmod with parameter 777 which signifies access (to knowledge) for everyone.

founded 1 month ago
MODERATORS
[email protected]
737Mika
45
"Rust is so good you can get paid $20k to make it as fast as C" (news.itsfoss.com)
submitted 6 days ago by [email protected] to c/itsfossnews
10 comments fedilink hide all child comments
 

"Rust is so good you can get paid $20k to make it as fast as C"

FFmpeg is one of the most widely used open source multimedia frameworks globally, with it powering audio and video processing for applications like VLC, OBS Studio, HandBrake, Jellyfin, etc. and being widely used in live-streaming, including platforms like Twitch.

On the other hand, rav1d is a new Rust-based decoder developed by Prossimo that aims to be a memory-safe reimplementation of the C-based dav1d decoder, which is widely recognized as the fastest software AV1 decoder in use across multiple platforms.

A recent post by FFmpeg has reignited conversation surrounding the trade-offs between performance, cost, and memory safety when building such critical systems.

FFmpeg is Not Impressed

Rust is so good you can get paid $20k to make it as fast as C:https://t.co/HVDokmLk5r

— FFmpeg (@FFmpeg) May 15, 2025

Commenting on Prossimo's recent performance bounty program for rav1d, FFmpeg mentioned that "Rust is so good you can get paid $20k to make it as fast as C". That looks like a direct jab at Rust and the rav1d decoder to me. 😲

The performance bounty has set out to close a performance gap where rav1d is ~5% slower than dav1d. Prossimo is offering a sizable $20,000 as a reward for anyone who manages to close the gap.

They have put out a bunch of rules that restrict pretty much most of the world from contributing, limiting this bounty only to certain regions and excluding a large portion of the global developer community.

Are they Right?

The information technology industry has been steadily moving towards Rust, a programming language known for its memory safety and performance. The U.S. White House even formally endorsed the shift towards memory-safe programming languages in a 2024 report from the Office of the National Cyber Director (ONCD).

However, the push for Rust adoption hasn't been without controversy. One such instance is the effort to integrate it into the Linux kernel, which has faced resistance and internal tension, with a recent occurrence causing friction and resulting in the creation of a Rust kernel policy to prevent future conflicts.

Despite growing adoption, many in the developer community continue to question whether Rust’s benefits truly outweigh those of longstanding languages like C.

From It's FOSS News via this RSS feed

top 10 comments
sorted by: hot top controversial new old
[–] [email protected] 8 points 5 days ago (1 children)

The question is, would auditing dav1d to ensure memory safety cost more or less than $20k?

[–] mutual_ayed 2 points 5 days ago* (last edited 5 days ago) (1 children)

Yup, all security comes with compromise. C is not memory safe, Rust is. Rust has to expend cycles being memory safe when running in multicore or parallelized workflows. C will always be "faster" than Rust as C does not run those checks by default at runtime. If you make C memory safe and bring in the memory handling libraries that do check for membuff and mempointer, C is as slow as Rust.

Regardless it's a moot point and a pedantic one to be honest. Rust, C, Go, Typescript, brainfuck are only as performant as they are written to be. The engineer is being paid 20k because they're making a functional implementation and adding security.

When you hear people arguing about which language is "faster" and they're not racing packets on a fiberoptic line because they work in FinTech, be very credulous of their claims.

Edit: Please see corrections in bold above, thanks to the comment below. Thank you @[email protected]

[–] [email protected] 3 points 5 days ago (1 children)

This statement is not 100% true, rust perform a lot of memory safety checks at compile time, rather than at runtime, tho in certain case such as when working with multicore programs that involve concurrent data modifications, Rust does perform runtime checks to ensure memory safety.

[–] mutual_ayed 2 points 5 days ago

Thanks for the clarification. And yes I was referencing multicore and parallel functions as those tend to be the target for most "optimizations" when moving away from C.

[–] sturger 5 points 5 days ago

These are all the old arguments Assembly threw at C when C started taking over. Now it's C's turn to be put out to pasture and it's rolling out the same complaints.

C had decades to solve its security problems and has not been successful enough. It's time for a new standard.

If you need faster than Rust and can afford the security issues, run C. If you need faster than C and have some top talent, use Assembly. If you need it faster and can afford the ASIC, use Verilog/VHDL. If you need faster and can afford a fab plant, start laying out your transistors with tape and a light table.

[–] [email protected] 3 points 5 days ago (1 children)

Rust is pretty good. But it isn't better than C when you have good C developers.

[–] [email protected] 2 points 5 days ago

Developers always make errors because perfection isn't realistically achievable. Rust eliminates many sources of errors.

[–] [email protected] 2 points 5 days ago

Kieran trolling is news now 🤣

[–] [email protected] 2 points 5 days ago

Seeing the state of shared libraries on Rust and the size of binaries, I dont see the benefit of this. The C version is still gonna be the better option in most cases.

[–] [email protected] 1 points 5 days ago

The thing is, would a company rather hire a young and dumb rust dev for cheap, or a old veteran C dev for bookoo bucks? Rust will "win" in the long run just due to this.