Matthias Kretz

Correct. It's not there and not proposed yet. Note that I just posted a [request for feedback on the Parallelism TS 2](https://wg21.link/p1915). You can submit feedback at https://github.com/mattkretz/std-simd-feedback/ which I...

I just added a link from the README to the paper: https://web-docs.gsi.de/~mkretz/DNMSO.pdf. And yes, I agree that also allowing a type like `std::strided` in place of a `simd` would be...

Oh, and of course, I need to work on an implementation to go with the paper...

> > And yes, I agree that also allowing a type like std::strided in place of a simd would be helpful. > > I meant it the other way round:...

The problem seems to be `inc_x` and `inc_y`. If these were known to be 1, then you could simply write: ```c++ using V = native_simd; for (int i = 0;...

This failure is strange. Why is the input value different than defined in `test_values`, i.e. the initial `0.` is a `nan` and the `denorm_min` values are `inf`? If you think...

> My hello world example, which I use to make initial build possible have only 3 remaining errors: `__make_dependent_t` was a clever hack to make the second template argument a...

https://godbolt.org/z/NeTaNs apparently only Clang rejects this pattern. That doesn't prove Clang is wrong. Not sure where to start in the standard :wink:

Try this: ```diff @@ -405,7 +405,12 @@ __is_neon_abi() ⸱ // }}} // __make_dependent_t {{{ -template using __make_dependent_t = _Up; +template struct __make_dependent +{ + using type = _Up; +}; +template...

> clang does not consider it a constant expression > https://godbolt.org/z/RpWrqw That's an unfortunate restriction in Clang. The better solution would be if Clang were to support constant expressions involving...