Replacing min() and max() for DCTL

[nmbr73]

We replace float max(float,float) with DCTL's float _fmaxf(float, float).

@J-i-P-i why did you suggest to replace it with float max_f(float,float)- did I get something wrong?!?

But my actual question is: Why is it, that float min(float,float) does not need to be substituted? I have overseen an occurence of min()in FractalLand.fuse, but it runs on all plattforms?!?

[J-i-P-i]

The origin is in the Readme from Blackmagic in (Partion): \ ProgramData \ Blackmagic Design \ DaVinci Resolve \ Support \ Developer \ DaVinciCTL \ README.txt

All mathematical functions and their "appearance" are given there.
I have to admit, I have done this conversion very "blindly" so far, but have also seen that e.g. the function clamp () works without modification.

I have now specifically examined the max () in Cuda:

• with float you can use max (float)

• if more dimensions are added, the _fmaxf () is required for Cuda, which was not so clear to me until now. It still has to be tested whether OpenCL is also supported for _fmaxf float2 to float4.

Now, I tested max() in OpenCL and there are no Convertions necessary, ->

So the change in _fmaxf() is only necessary for Cuda, but the explicit distinctions max_f () can be omitted