You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Closes: #692
Note that I have no clue how these vector types would be used -- I don't even have a processor that supports them, but their mere presence in LLVM's builtins caused the above-mentioned issue.
This follows along the paths of existing vector types, reusing the code already established for SVE. As this produces a `Float16` rather than a `Half` builtin down the road, support for that is added by mapping to Half. That seems to be OK because, unlike C's `int` size mess, `half` is commonly understood to mean an IEEE 16-bit float.
0 commit comments