GitLab

* CCMA with a small number of constraints uses a single kernel

* Fixed compilation errors in kernel

* Fixed compilation errors in kernel

* Further optimizations to CCMA with few constraints

* Fixed range overflow with very large numbers of atoms

* More fixes to overflow with large numbers of atoms

* Fix test failures

* Updated to latest values for physical constants

* Updated documentation on physical constants

* Updated Python unit definitions

* Fixed constants in test case

* Added a comment

This ensures that the correct argument size is used when calling clSetKernelArg.

Explicitly cast the second parameter to the type of the first one so the compiler can pick an overload that is supported in device code (fixes error: calling a __host__ function("fmin<float, int, (int)0> ") from a __global__ function("computeBondedForces") is not allowed).

Naming the argument list of a variadic macro is a GNU extension which is not supported by msvc. Since CUDA uses the system preprocessor this will fail to build kernels on Windows (fixes error C2010: '.': unexpected in macro parameter list).

The exclusion mask was 8-bits wide, but future CPU support will require
more bits than this (e.g., AVX-512 will be 16-bit).

* Cleaned up checkpointing code for NoseHooverIntegrator

* Fixed compilation error

* Fixed OpenCL compiler errors

* Fixed incorrect references to BAOAB

* Fixed a method that was incorrectly made public

* Convert Nose-Hoover into LF middle scheme by copying NH kernels

* Rebrand VelocityVerletIntegrator as NoseHooverIntegrator

* Consolidate NH tests

* NoseHooverChainKernel begone

* Make Windows builds happy

* Add missing header for Windows build

* Fix mistake in CommonKernels header

* Add 6th Yoshida-Suzuki and make it the default

* CPU: Refactored code to be generic across vector CPU platforms.

Ewald and non-Ewald interactions now share a common code base, templated on
their interaction type.

The vec4 and vec8 implementations have been replaced by a single generic implementation
class which is templated on SIMD type. Currently works for SIMD4 and SIMD8 types, but
can be extended in future to support other types (e.g., AVX-512).

Modified runtime CPU support to lay groundwork for future SIMD types.

Pulled out some vector utility functions (gather pair, reduce),
and refactored the AVX CPU code to make use of them.

* CPU: Fixed coding standards and incorrect header include.

* CPU: Fixed code review comments from PR #2661

* CPU: Fixed CI build issues.

* CPU: Further CI fixes.

* CPU: Fix for unit test failure on MacOS.

Reverted optimised code to go back to a version which is thought to work
on MacOS. The optimisation will be ...

* Remove unused variable from CpuCustomGBForce::calculateIxn

* Remove unused variable from CpuCalcCustomManyParticleForceKernel::initialize

* Remove unused variable from CudaCalcCustomCVForceKernel::initialize

* Remove unused variable from ReferenceCalcCustomManyParticleForceKernel::initialize

* Remove unused variables from ReferenceObc::computeBornEnergyForces

* Remove unused variable from AmoebaTorsionTorsionForceProxy::serialize