Jubilee Young [Thu, 20 Jan 2022 23:48:46 +0000 (15:48 -0800)]
Annotate signed type in int_divrem_guard
The way the macro expands, it may sometimes infer
"this is a uint, but doesn't impl Neg???"
Also, I made the "wrong path for intrinsics" error.
These fixes allow integration into libcore.
Jubilee [Fri, 31 Dec 2021 23:36:40 +0000 (15:36 -0800)]
Merge portable-simd#219 - ./std-float
impl std::simd::StdFloat
This introduces an extension trait to allow use of floating point methods
that need runtime support. It is *excessively* documented because its mere
existence is quite vexing, as the entire thing constitutes a leakage of
implementation details into user observable space. Eventually the entire
thing will ideally be folded into core and restructured to match the rest
of the library, whatever that structure might look like at the time. This
is preferred in lieu of the "lang item" path because any energy the lang
items require (and it will be significant, by Simulacrum's estimation) is
better spent on implementing our libmvec.
Jubilee Young [Tue, 21 Dec 2021 23:00:35 +0000 (15:00 -0800)]
impl std::simd::StdFloat
While consulting with Simulacrum on how to make available the float
functions that currently require runtime support for `Simd<f32, N>` and
`Simd<f64, N>`, we realized breaking coherence with the classic approach
of lang items was, since `{core,std}::simd::Simd` is a `ty::Adt`, likely
to be quite a bit nasty. The project group has a long-term plan for how
to get around this kind of issue and move the associated functions into
libcore, but that will likely take time as well. Since all routes
forward are temporally costly, we probably will skip the lang item
approach entirely and go the "proper" route, but in the interests of
having something this year for people to play around with, this
extension trait was whipped up.
For now, while it involves a lot of fairly internal details most users
shouldn't have to care about, I went ahead and fully documented the
situation for any passerby to read on the trait, as the situation is
quite unusual and puzzling to begin with.
Jubilee [Thu, 30 Dec 2021 09:22:01 +0000 (01:22 -0800)]
Merge portable-simd#210 - ./wrap-shifts
Refactor ops.rs with wrapping shifts
This approaches reducing macro nesting in a slightly different way. Instead of just flattening details, make one macro apply another. This allows specifying all details up-front in the first macro invocation, making it easier to audit and refactor in the future.
This refactor also has some functional changes. Only one is a true behavior change, however:
- The visible one is that SIMD shifts are now wrapping, not panicking on overflow
- `core::simd` now has a lot more instances of `#[must_use]`, which merely lints
- div/rem now perform a SIMD check but remain as before, which should improve performance but be invisible
Jubilee Young [Wed, 22 Dec 2021 02:28:57 +0000 (18:28 -0800)]
Refactor ops.rs with a recursive macro
This approaches reducing macro nesting in a slightly different way.
Instead of just flattening details, make one macro apply another.
This allows specifying all details up-front in the first macro
invocation, making it easier to audit and refactor in the future.
Jubilee Young [Thu, 9 Dec 2021 00:44:21 +0000 (16:44 -0800)]
Wrap bitshifts in ops.rs
For all other operators, we use wrapping logic where applicable.
This is another case it applies. Per rust-lang/rust#91237, we may
wish to specify this as the natural behavior of `simd_{shl,shr}`.
I realized that our `select` implementation predated `Simd` being generic over element type, and we don't really need the `Select` trait at all. The function signature is much simpler now (generic over element type, rather than over the entire vector). This did require changing mask select to be a different function, but I think that's fine considering they're not necessarily vectors.
Jubilee [Fri, 3 Dec 2021 01:41:30 +0000 (17:41 -0800)]
Merge portable-simd#195 - portable-simd:trait-ops
Generic `core::ops` for `Simd<T, _>`
In order to maintain type soundness, we need to be sure we only implement an operation for `Simd<T, _> where T: SimdElement`... and also valid for that operation in general. While we could do this purely parametrically, it is more sound to implement the operators directly for the base scalar type arguments and then use type parameters to extend the operators to the "higher order" operations.
This implements that strategy and cleans up `simd::ops` into a few submodules:
- assign.rs: `core::ops::*Assign`
- deref.rs: `core::ops` impls which "deref" borrowed versions of the arguments
- unary.rs: encloses the logic for unary operators on `Simd`, as unary ops are much simpler
This is possible since everything need not be nested in a single maze of macros anymore. The result simplifies the logic and allows reasoning about what operators are valid based on the expressed trait bounds, and also reduces the size of the trait implementation output in rustdoc, for a huge win of 4 MB off the size of `struct.Simd.html`! This addresses a common user complaint, as the original was over 5.5 MB and capable of crashing browsers!
This also carries a fix for a type-inference-related breakage, by removing the autosplatting (vector + scalar binop) impls, as unfortunately the presence of autosplatting was capable of busting type inference. We will likely need to see results from a Crater run before we can understand how to re-land autosplatting.
Jubilee Young [Wed, 24 Nov 2021 01:55:14 +0000 (17:55 -0800)]
Drop splats for Simd<T, _>
Unfortunately, splatting impls currently break several crates.
Rust needs more time to review possible mitigations, so
drop the impls for the `impl Add<T> for Simd<T, _>` pattern, for now.
Jubilee Young [Wed, 24 Nov 2021 01:36:54 +0000 (17:36 -0800)]
impl unary.rs for Simd<{i,u}{8,16,32,64,size}, _>
In order to assure type soundness, these "base" impls
need to go directly on Simd<T, _> for every scalar type argument.
A bit of cleanup of ops.rs is still warranted.
Resolves my comment in #197, at least for now; #187 is pending but since these are already here, just commented, it seemed to make sense to me to re-enable them anyway.
Jubilee Young [Wed, 24 Nov 2021 00:43:02 +0000 (16:43 -0800)]
impl assign.rs<U> for Simd<T, _>
Instead of implementing {Op}Assign traits for individual scalar type args
to Simd<_, _>, use parametric impls that reassert the bounds of the binary op.
Jubilee Young [Wed, 24 Nov 2021 00:15:19 +0000 (16:15 -0800)]
impl deref.rs<&Self> for Simd<T, _>
Instead of implementing each "deref" pattern for every single scalar,
we can use type parameters for Simd operating on &Self.
We can use a macro, but keep it cleaner and more explicit.
Jubilee Young [Wed, 10 Nov 2021 07:49:16 +0000 (23:49 -0800)]
Dynamically detect AVX512 in CI
We would like to check for errors with AVX512,
but we don't pick our CPU. So, detect available features.
This variance in checks stochastically reveals issues.
Nondeterminism is acceptable as our goal is protecting downstream.
Proloy Mishra [Tue, 9 Nov 2021 01:28:43 +0000 (06:58 +0530)]
add `Simd::from_slice` (#177)
* add `Simd::from_slice`
uses a zeroed initial array and loops so that it can be const.
unfortunately, parameterizing the assert with slice length
needs `#![feature(const_fn_fn_ptr_basics)]` to work.
Jubilee Young [Fri, 22 Oct 2021 01:20:06 +0000 (18:20 -0700)]
Fixup import pathing for core
This changes simd_swizzle! to a decl_macro to give it a path,
so it can be imported using a path and not the crate root.
It also adds various uses that were missed and adjusts paths.
Jubilee Young [Sun, 19 Sep 2021 01:31:49 +0000 (18:31 -0700)]
Restructure crate as core module
Aligns module with rust-lang/library/core, creating an... unusual
architecture that is easier to pull in as a module, as core itself can
have no dependencies (as we haven't built core yet).
Jubilee Young [Wed, 28 Jul 2021 02:59:50 +0000 (19:59 -0700)]
Rename to portable-simd and remove other names
Clean up references to the repo's previous name.
Removes the authors field, which is non-obligatory since RFC 3052.
Better to omit than confound: let git log be our witness.
Jubilee [Mon, 2 Aug 2021 06:18:24 +0000 (23:18 -0700)]
Merge pull request #152 from rust-lang/feature/const_eval_checked
Feature/const eval checked
Adds the const_evaluatable_checked cargo feature for some experimental (but very useful) function implementations. Adds testing all cargo features to CI.
Merge pull request #142 from rust-lang/feature/traits
Combine LanesAtMost32 and SimdArray into a single trait "Vector"
Attempts to fix some unresolved questions in #139 regarding `SimdArray` having a generic parameter.
In particular, this made it not appropriate for replacing `LanesAtMost32`. Additionally, it made it impossible to use in a context where you otherwise don't know the lane count, e.g. `impl Vector`.
An unfortunate side effect of this change is that scatter/gather no longer work in the trait (nor does anything else that references the lane count in a type. This requires the super-unstable `const_evaluatable_checked` feature).
I also threw in the change from `as_slice` to `as_array` as discussed in zulip, and fixes #51.