b-sumner
b-sumner
It's a real pleasure to hear from such a well-known and remarkable group! Please expect our response soon.
My current understanding is that we would like to see this work succeed for AMD CPUs and GPUs. On the GPU side, how would you like to communicate?
I think all the tools you need to develop a correctly rounded math library for HIP or OpenMP offload language applications are available. But I expect there may be information...
Sorry for my confusion. I understand better now what is being proposed.
@zimmermann6 I'm afraid that is not likely. There is quite a willingness here to trade accuracy for performance. A correctly rounded library would be good as an alternative though, but...
Thank you for the notice!
From my perspective, the C spec does not distinguish between QNaN and SNaN in its description of fmax() and fmin(), so this PR, if it is accepted, might need a...
Personally, I don't think the footnote is good enough. The spec for fmax/fmin says only NaN, and that means "any kind of NaN". If we allow fmax/fmin to return a...
The concern here is that the patch wants fmax/fmin to be able to return a QNaN if one or more of the arguments is SNaN. This is specifically not permitted...