Hi, our application uses 80 bit FP code because we need the additional precision. Since we are always keen to optimize and speed-up our algorithms I'm wondering if oneAPI offers something for us.
Has anyone experience/is it possible at all to use the GPU for 80 Bit floating point calculations? Or are there are libraries that can use a GPU to simulate 128 Bit floating point calculations?
To us it looks x87 stuff is still available (and hopefully will be in the future) but is not an active development focus, so no innovations to expect.
For more complete information about compiler optimizations, see our Optimization Notice.