Skip to content

[BUG] Incorrect gating of FADD, FSUB (accumulate into ZA) #426

Description

@rockdreamer

Describe the bug

In the "ADD, SUB, BFADD, BFSUB, FADD, FSUB (accumulate into ZA)" section, we guard half precision element operations using __ARM_FEATURE_SME_F16F16 only.

However, "Two ZA single-vectors of half-precision elements" and "Four ZA single-vectors of half-precision elements" encodings are available when either (FEAT_SME_F16F16 || FEAT_SME_F8F16) features are available.

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

    bugSomething isn't working

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions