Patrick Stotko

Thanks. Even though the `device_range()` method comes with an overload that accepts an `execution_policy`, it internally needs an `atomic` whose `load()` and `store()` functions use non-async mempy. We probably need...

Sorry for the long delay. It took a larger refactoring to fill the gaps in the stream support, but with #450 this issue should be resolved.

> Add `bool occupied(index n)` function for unordered_set and unordered_map. So that we can: > > ```cuda-c++ > struct AllocateNewBlocks { > AllocateNewBlocks(stdgpu::unorderd_map block_map, > BlockBuffer buffer_buffer, > stdgpu::unorderd_set not_exist_block_indices)...

The underlying issue has been fixed in #450.

While inserting at the the end of the container is straightforward, inserting from an arbitrary position would involve temporary memory allocations. Once could of course first shift the entries located...

Yes, you observed the same issue [here](https://github.com/isl-org/Open3D/discussions/7055) and I was and still am quite confused of the why this happens. However, if I read the error correctly, then it looks...

Hi, maintainer of stdgpu here. The error mentioned in the initial post has been recently fixed in stdgpu, i.e. support for CUDA 13 has been added (see https://github.com/stotko/stdgpu/pull/466). That said,...