In large-eddy simulation (LES), large-scale turbulent structures are explicitly resolved on the numerical grid while the dissipative turbulent eddies, typically smaller than the grid size, must be modeled. Because in the atmospheric boundary layer a large disparity of turbulent scales exists (about 9 orders of magnitude separate the largest and smallest scales), LES is considered as an essential modeling approach to capture the physics and dynamics of boundary layer clouds. A new LES solver developed at Stockholm University is presented here for the first time. The model solves for nonhydrostatic anelastic equations using high-order low-dissipative numerical schemes for the advection of scalars and momentum. A two-moment bulk microphysics scheme is implemented representing five types of hydrometeors including ice crystals and snow. The LES is evaluated based on simulations of two well-documented stratiform cloud events that were previously used for LES intercomparisons. In the first one, a marine drizzling stratocumulus observed during DYCOMS-II, the model is shown to predict bulk cloud microphysical and dynamical properties within the range of the intercomparison model results. In the second case, based on a monolayer Arctic mixedphase cloud observed during ISDAC, we found that when using fast-falling crystals, ice quickly precipitates out of the cloud without significant growth, resulting in very low ice water paths. The simulated clouds are also found to be very sensitive to the prescribed ice crystal number concentration: multiplying the ice concentration by a factor 2.5 results in rapid cloud dissipation in the most extreme case. Overall, these results are found to be consistent with former studies of Arctic mixed-phase clouds as well as in situ measurements. More specifically, when the ice number concentration and parameterized ice habit are constrained by measurements, simulated microphysical properties such as the ice water path and ice crystal size distribution are found to agree well with observations.