Major update: MBX 1.3 contains numerous improvements, mostly to performance.
Changelog
Added a variety of new configuration options to enable optional optimizations.
Evaluation of MB-pol 2- and 3-body PIPs can now use SIMD instructions if enabled with new configuration options. Depending on your hardware, this produces anywhere between a 1.25x and 2.0x overall speed-up for the entire MB-pol PEF.
Refactored dispersion and Lennard-Jones to be more efficient on large boxes.
Intel Threading Building Blocks library can be used if enabled with new configuration options, significantly improving performance on large boxes.
Intel intrinsic math functions can now be used if enabled with new configuration options, significantly improving performance on large boxes.
Many small improvements to OpenMP parallelism.
Many small performance improvements, including many that are most impactful on large boxes.
Can now specify the 2-body PIP cutoff separately from the real-space cutoff used for dispersion and electrostatics.
