triton.experimental.gluon.language.nvidia.blackwell.fma2

triton.experimental.gluon.language.nvidia.blackwell.fma2(lhs, rhs, acc, dtype=None, _semantic=None)

Perform a native two-lane packed fused multiply-add.