Ben Corbett
Ben Corbett
Yeah we'd definitely want two macros. The original one is used all over the place (all the GEOSX macros flow through LvArray), and while streams can be annoying they're less...
Move `bufferManipulation::HasMemberFunction_move` into `typeManipulation.hpp`.
> When dealing with various containers generically (e.g. `Array` and `ArrayOfArrays` that may represent connectivity maps between different kinds of mesh objects), one cannot uniformly extract the primary dimension size...
This would be awesome but I think it would require some changes to RAJA https://docs.nvidia.com/cuda/nvrtc/index.html For the future.
So there seem to be a few things that are missing. At least the first issue needs to be solved before we make this a requirement for GEOSX. ### Pre...
> Pretty neat! Only thing is I feel like it could be a completely standalone library (it's leaning on some `LvArray` facilities but not too heavily). I agree. I originally...
@CusiniM thanks! I think the easiest way to reduce the compile times is to 1. Reduce the number of typed test instantiations. Use say `int32[ 1 ]` and `float[ 4...
CUDA has a `fma` as well, just like `cos` and whatnot. I'm not sure it would be beneficial but worth checking out.
Yeah but it *could* be slower: https://stackoverflow.com/questions/34265982/automatically-generate-fma-instructions-in-msvc For things like `AiBi` it is very applicable. But how you'd go about applying it to things like ```c++ dstSymMatrix[ 3 ] =...
And redo the input/output tests.