Minor cleanup of the Vector64/128/256/512 implementations to improve fallbacks #103095
Conversation
dotnet-issue-labeler
bot
added
the
needs-area-label
| Vector256.Divide(left._upper, right._upper) | ||
| ); | ||
| } | ||
| public static Vector512<T> Divide<T>(Vector512<T> left, Vector512<T> right) => left / right; |
There was a problem hiding this comment.
Cases like this were where size regressions were coming from. This ensures that instead we hit the intrinsic path when we're accelerated and otherwise we get the same overall codegen as before.
| @@ -1988,7 +2004,7 @@ public static Vector512<T> Min<T>(Vector512<T> left, Vector512<T> right) | |||
| /// <returns>The product of <paramref name="left" /> and <paramref name="right" />.</returns> | |||
| /// <exception cref="NotSupportedException">The type of <paramref name="left" /> and <paramref name="right" /> (<typeparamref name="T" />) is not supported.</exception> | |||
| [Intrinsic] | |||
| public static Vector512<T> Multiply<T>(T left, Vector512<T> right) => left * right; | |||
| public static Vector512<T> Multiply<T>(T left, Vector512<T> right) => right * left; | |||
There was a problem hiding this comment.
Cases like this aren't strictly needed, but it avoids needing +1 inlining step when we aren't accelerated and defers to the core algorithm instead.
This overload is just convenience so that scalar * vector and vector * scalar both work, since the operation is commutative, so there's no optimization opportunity missing here either.
tannergooding
added
area-System.Runtime.Intrinsics
and removed
needs-area-label
|
Tagging subscribers to this area: @dotnet/area-system-runtime-intrinsics |
|
Size regressions in APIs like For There's a few places where the fallbacks involving The meaningful places that will execute when |
|
CC. @EgorBo |
Continuation of #102301, just cleaning up some of the logic to require less inlining and defer to the same size core API for the fallback implementation where trivially possible.