Towards fearless SIMD, 7 years later

Re:Towards fearless SIMD, 7 years later by mungaihaha on Sunday March 30, 2025 @05:07AM (#43522642)

> I'm still not seeing any description of specific challenges you feel are harder in Rust than in C/C++

It's harder to model domains in Rust. C has the same problem and it is a big reason why the industry standard for game dev is C++

[ Parent | on HN ]

Re:Towards fearless SIMD, 7 years later by tuna74 on Sunday March 30, 2025 @11:34AM (#43524965)

Could you expand on this? What exactly doe "model domains" mean?

[ Parent | on HN ]

Re:Towards fearless SIMD, 7 years later by vlovich123 on Sunday March 30, 2025 @12:25PM (#43525328)

Care to give an example? I think GUIs remain the softest domain modeling area, but that’s about how do you do next gen GUI toolkits that are super high performance, safe, and lower overhead vs what Rust has today. But that’s an underserved niche anyway vs c++ toolkits or electron.

The other weak parts might be ecosystem immaturity (eg Unreal vs Bevy) but that’s not a language modelling issue.

[ Parent | on HN ]

Re:Towards fearless SIMD, 7 years later by mungaihaha on Monday March 31, 2025 @03:19AM (#43532120)

Rust makes it hard to represent mutable graphs. If I have a value that can be updated by multiple GUI nodes, I have to architect my way around the borrow checker in Rust. Is it safe? yes. Is is performant? for a skilled Rust dev, yes. But it takes me >25% more time to do compared to C++. 25% of a year long project is 3 months

[ Parent | on HN ]