__half2float question

[jinz2014]

For the CUDA function: __half2float(x),

Are the two functions below equivalent in a SYCL program ? If 1 is not "equal" to 2, are there simple functions for conversion between half and float ? The

1 sycl::vec<sycl::half, 1>(x).convert<float, sycl::rounding_mode::automatic>()[0]

2 float(x)

[tomflinda]

Good advice for bullet 2. We will do a further investigation and refine it if the result is accepted.