Khronos Developing SYCL SC For Safety-Critical C++ Heterogeneous Compute

[phoronix]

Phoronix: Khronos Developing SYCL SC For Safety-Critical C++ Heterogeneous Compute

The SYCL single-source C++ based programming model has begun taking off with Intel investing in it heavily as part of their oneAPI / DPC++ compiler stack and a variety of different open-source projects bringing SYCL to the likes of AMD and NVIDIA GPUs, CPU-based OpenMP implementations, SYCL to Vulkan, and more for heterogeneous compute needs. The Khronos Group announced today they have begun working on SYCL SC as a safety-critical variant of this heterogeneous compute programming model...

https://www.phoronix.com/news/Khronos-SYCL-SC

[CTTY]

Safety-Critical in C++? Why not use rust?

[Veto]

Because there is still no safety certified Rust compiler in sight.

Rust is a promising language, but it needs to stabilize and mature before I will trust my life to it.

[PAUL007]

caution triggered rust devs ahead

[coder]

Safety-Critical in C++? Why not use rust?

Too new, I'd guess. They value maturity and stability. Furthermore, they reference several other standards, which are probably also based on C++.

[cl333r]

Because there is still no safety certified Rust compiler in sight.

Rust is a promising language, but it needs to stabilize and mature before I will trust my life to it.

And by the time Rust takes over there will be Carbon* or something and people will start switching to it because in 90% it provides a better trade-off (because security is not everything) while being safer than C++.

* The language is expected to have a 1.0 release in 2024 or 2025, which IRL would happen like 3 years later.

[Veto]

And by the time Rust takes over there will be Carbon* or something and people will start switching to it because in 90% it provides a better trade-off (because security is not everything) while being safer than C++.

While you may wish so, I don't see any reason why that should happen. It will most likely be abandoned as yet another language experiment by Google.

Rust on the other hand has a tremendous traction towards finally becoming a successor to C/C++ as the de facto systems programming language standard. Being included in key Open Source projects like the Linux kernel, Mesa etc. etc. and soon having support in the two most influential compilers will cement its success. A couple of years down the line the embedded and safety software communities will have built safety certified compilers and developed the standards (MISRA and whatnot) for finally adopting a language that isn't hopeless for the task it is meant to solve.

We are just not there yet...

[Veto]

We are just not there yet...

... I stand (somewhat) corrected. It seems we are almost there: https://www.lynx.com/executive-blog/...itical-markets

Of course they are moving this way. Developers have been yearning for a safe language that isn't Ada for years

[cl333r]

While you may wish so, I don't see any reason why that should happen. It will most likely be abandoned as yet another language experiment by Google.

And I think Dart is much better than JavaScript and Golang is much better than Python, the latter is pretty successful and the former not sure.
Rust succeeded fast exactly because there was only the ancient C and the crufted C++, now that (hopefully) in a few years we'll have a modern, safer, cruft-free C++ alternative Rust won't have such an easy walk to the top. But yeah, since Google is a corporation it might simply make the wrong decision on Carbon and pull the plug, just like they decided to drop support for JpegXL in Chrome.

The fact that Carbon went open source is good but doesn't guarantee much because it's still a Google project by mostly Google employees, we just don't have a good alternative to C++ - and this is my problem.

[Veto]

we just don't have a good alternative to C++ - and this is my problem.

Just being curious; why don't you see Rust as a good alternative, and what does Carbon do for you, that Rust does not?