The use of 3-D CFD combustion models based on tabulated chemistry is becoming increasingly popular. Especially the runtime benefit is attractive, as the tabulated chemistry method allows including state-of-the-art chemical reaction schemes in CFD simulations without significant penalties in terms of computational time. In this work, the Tabkin FGM combustion model in AVL FIRE is used to perform LES simulations of a diesel ICE (AVL SCRE). Four load conditions are investigated with three different fuel surrogates. Predicted data are compared with reference ones (measurements or data from calibrated 0-D/1-D model) while discussing differences between them. CPU benefits are quantified. The main conclusion is that such CFD model has high predictive ability while requiring low calibration effort and being relatively fast, hence it is an interesting alternative to RANS-based industrial applications.