Celeritas is a new Monte Carlo transport code designed to accelerate scientific discovery in high energy physics by improving detector simulation throughput and energy efficiency using GPUs.
As part of this I added an improved grow function for the bbox and realized while doing so that the existing min/max that we've defined is suboptimal for certain architectures (viz., ARM) and that the CUDA floating point min/max are exactly equal to fmin/fmax which is supported across architectures.
As part of this I added an improved
grow
function for the bbox and realized while doing so that the existingmin/max
that we've defined is suboptimal for certain architectures (viz., ARM) and that the CUDA floating pointmin/max
are exactly equal tofmin/fmax
which is supported across architectures.