Skip to content

[Bug] Vector types incorrectly passed as ptr #2427

Description

@zerkawei

The vector types in System.Numerics (float2, float4, int32_4, ...) when passed as a function argument are passed as ptr rather than using the proper <n x type>.

For example :

public static float4 Mult(float4 a, float4 b)
{
	return a * b;
}

becomes :

define <4 x float> @_ZN2bf8TestZone7Program4MultENS_6System8Numerics6float4ES4_(ptr %a, ptr %b) #0

This does not match C behavior regarding these types where :

typedef float vec4 __attribute__((vector_size(16)));

vec4 mult(vec4 lhs, vec4 rhs) {
    return lhs * rhs;
}

becomes :

define dso_local <4 x float> @mult(<4 x float> noundef %0, <4 x float> noundef %1) #0

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions