binary32 -> binary16: each thread converts one pair of elements
(Realization.Nvidia.SM86.Cast's geometry: elements / 2 threads). A launch
of one thread per element writes a second output's worth past the plane.
1461def cgCastDown =
1462 (lambda unrestricted output : Nat .
1463 (lambda unrestricted input : Nat .
1464 (lambda unrestricted elements : Nat .
1465 (cgUnary cgImageCastDown output input (naturalDivideUnchecked elements 2)))))The compiler supplied declaration spans and resolved links from this source snapshot. This page does not assert that this file belongs to a checked closure.