James Schloss
James Schloss
We have no stack or queue code, but sure
This is still a relatively important issue. We are waiting on finalized implementations for these languages before we can put them up in the AAA, so I feel like this...
Right, the issue is that the other languages are missing tests against fftw (or similar lib). This is why I created a separate section for them
I know this is still a draft, but is it close to being reviewed?
I think it fits best in the [Code style guide](https://www.youtube.com/watch?v=ijktorPAHGs) as a general principle. I also agree that we should create a "disclaimer" chapter in place of the current how...
My understanding here is that we would still like pure functions, but we also more heavily comment these functions to make it more clear what we are doing?
Hey, I kinda left this in the dark as issues with other projects cropped up. So far, I 1. Mocked up a simple NBody simulator in Julia to get an...
I had no intention of touching simulator definitions directly. In fact, I don't really think a switch to KA would speed up much outside of `accelerations` and `find_neighbors`, so those...
The kernels branch looks good and is easily extendable to KA for AMDGPU / CPU support as well (it shouldn't change CUDA performance). I have a KA branch as well...
Great! Thanks for the info / I'll see if I can work on the Enzyme issues. Apparently the bottleneck is proving the correctness of the adjoint. I'll dig into it...