Pradeep Garigipati
Pradeep Garigipati
@gaika I am assuming you haven't already come across [this](https://github.com/JuliaComputing/ArrayFire.jl). That wrapper is officially maintained by JuliaComputing group. May be you want to check it before re-implementing the entire wrapper...
@pavanky So, this would entail adding an additional function - something like the following `af::deviceMemSize` & `af_device_mem_size` ?
That is correct, this hasn't been added yet.
Thanks for raising the issue based on our slack discussion. This should be possible once we have changes done to our C API that can use existing `af_array` handles as...
Our C++ API is just a wrapper like our rust or python wrapper of ArrayFire's C-API. Entire magic of ArrayFire lies inside our C-API, all wrappers provide respective language's syntatic...
[`af_approx_v2`](https://arrayfire.org/docs/group__signal__func__approx2.htm#ga7448f8eecc9e454d814833b1b19af699) thats for one. The C++ API doesn't yet take advantage of these new versions of API yet.
Sure, I think a line added to `randn` documentation could work as an example. However, I think this kind of operation is not specific to this function - these are...
Is the behavior same on non-Intel devices too ?
@brianmcconnel We cannot reproduce this on our CI NVIDIA cards, so we would need your help running through debugging for collecting necessary stack trace. Can you please run the test...
@brianmcconnel Did you get the chance to try the debugging steps I suggested earlier ?