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
round_to_int can't be directly implemented in LLVM, since llvm.lround doesn't work on vectors. I'm not sure this function is actually necessary since you can do x.round().to_int().
The text was updated successfully, but these errors were encountered:
The name/semantics of simd_cast are confusing so I'd tend to agree. simd_cast should be safe to call on any vector types of the same lane count, and we should have a new intrinsic that does unsafe int-to-float casts.
to_int
(requires comparisons in Comparison functions #36)rount_to_int
(initially proposed in Rounding and Type Change methods #23)trunc
(removed in Use platform intrinsics, not LLVM, for floor & ceil #47)fract
(removed in Use platform intrinsics, not LLVM, for floor & ceil #47)round
(removed in Use platform intrinsics, not LLVM, for floor & ceil #47)round_to_int
can't be directly implemented in LLVM, sincellvm.lround
doesn't work on vectors. I'm not sure this function is actually necessary since you can dox.round().to_int()
.The text was updated successfully, but these errors were encountered: