GitLab

* Fixed issue that caused inefficient sorting when a block contained only one atom

* Add the fix to OpenCL and HIP

* Fixed error on GPUs with SIMT width 64

* Interpret sortedBlocks correctly

* Improved sorting of blocks when building neighbor list

* Improved block sorting for OpenCL

* Made sort keys more evenly distributed

* Use large blocks to optimize building the neighbor list

* Large blocks optimization for OpenCL

* Fix test failures

* Select whether to use large blocks based on system size

Without this fix, we see cases in which not all work-items in a thread group end up hitting the same number of barriers, which leads to a hang in OpenCL GPU execution.

* Fixed potential invalid memory access

* Fixed exception

* Fixed range overflow with very large numbers of atoms

* More fixes to overflow with large numbers of atoms

* Fix test failures

Converted the array containing atom block indices for the neighbor list from ushort2 to int.  This removes the hard limit of 2 million atoms.

On Fermi, store forces as 64 bit fixed point and update with atomic operations.  Also various other minor optimizations.