Gao, Xiang comments

Results 24 comments of


Gao, Xiang

Device-side launch of thrust::lower_bound is creating wrong results

Related customer issue: https://github.com/pytorch/pytorch/pull/80714

Device-side launch of thrust::lower_bound is creating wrong results

Wait, is the `thrust::lower_bound` here a CDP launch? From `nsys nvprof`, I only see one kernel, which is the one I wrote: ``` Time (%) Total Time (ns) Instances Avg...

[RFC] Async Extensions

I would recommend that we modify the behavior of things like `$(command &)` and `!(command &)` to return an awaitable in sync context. Currently it is returning a NoneType: ```python...

Increase the chemical space coverage of unit tests

Add xyz files in this folder: https://github.com/aiqm/torchani/tree/master/tools/generate-unit-test-expect

integrating with lammps

Hi @chemxyang, this is a good-to-have feature. But currently, we don't have enough man-power to work on it. It is welcome if someone would like to work on it!

Support named tensor

Yes

Clean up old CUDA support for Windows

@peterjc123 Now it is deleted

[MacOS client] Could not create tunnel device: No such file or directory

I had the same problem. It was solved after installing tuntap drivers. @friedrich Thanks!

[WIP] Allow cub::DeviceRadixSort and cub::DeviceSegmentedRadixSort to use iterator as input

@allisonvacanti Looks like in NVIDIA/cub#357, iterators are only supported for offsets, for input and output keys and values, they still need to be pointers?

[WIP] Allow cub::DeviceRadixSort and cub::DeviceSegmentedRadixSort to use iterator as input

1.16 is fine to me