The main design objectives of a high pressure compressor are the aerodynamic efficiency and the operating range (e.g. the surge margin). Those quantities are impacted by secondary and leakage flows occurring in the blade passage such as corner separation or stall and tip leakage flows. The turbulence modeling influences strongly the prediction of the overall performances. The aims of the present study were (i) the validation of the combination of the SAS approach with the DRSM turbulence model by comparison to experimental data, especially to laser measurements in the tip of a rotor of a high pressure compressor and (ii) the discussion of the flow prediction improvements with respect to turbulence approaches classically used in CFD and industry: URANS simulations and standard SAS simulation i.e. combined with SST turbulence model. The SAS results are compared to experimental data and to URANS results (SST and DRSM). Only the simulations with IGV wakes predict the velocity fluctuations near tip gap, from the leading edge. Concerning the time-averaged performances, the stagnation pressure losses are slightly overestimated by SAS, especially with DRSM model. This is due to an amplification of the hub corner separation. Moreover, the isentropic efficiency is very sensitive to the SAS approach and to the turbulence model. The spectral analysis shows that the prediction of the amplitude and frequencies of the power spectral density of static pressure is improved using the SAS approach instead of URANS one. The SAS approach leads to PSD similar to ZDES, especially with the DRSM model. Thus, the SAS-DRSM is able to well predict the tip leakage flow with the fine mesh. Nevertheless, this approach amplifies the hub corner separation leading to a strong underestimation of overall performances.