Predicting the cyclic response of soils is still challenging in many geotechnical fields, and motivates massive research to shed light on lesser-known aspects of the problem. In this area, the continual efforts on the constitutive modelling of cyclic sand behaviour demand new and reliable dataset for model validation -especially for loading conditions involving many loading cycles ('high-cyclic' loading). In this letter, the recent memory-enhanced bounding surface formulation by Liu et al.(2018a) is considered as a suitable platform to reproduce the high-cyclic response of sands, and its transition into either 'ratcheting' or 'shakedown' behaviour. New evidence of its suitability is found against the latest dataset presented in Park & Santamarina (2018), comprising the results of highcyclic oedometer tests at varying initial/loading conditions. Model-simulations prove in satisfactory agreement with most experimental findings, especially regarding the prediction of so-called 'terminal densities'.