Half precision in lapack function

[Юрий Сидоров]

I have this configuration:

ArrayFire v3.9.0 (CUDA, 64-bit Linux, build b59a1ae53)
Platform: CUDA Runtime 12.2, Driver: for
[0] NVIDIA GeForce RTX 3050, 7832 MB, CUDA Compute 8.6

I'm trying to use half precision in calculations. Matrix and vector operations: multiplication, addition work, but as soon as I use lapack functions an error occurs:

lu decomposition:

In function af_lu
In file src/api/c/lu.cpp:73
Invalid type for argument 1

for norm:

In function af_norm
In file src/api/c/norm.cpp:144
Invalid type for argument 1

As I see, there is no f16 support for lapack functions?

A similar situation with ArrayFire Python.