The Amazon basin plays a significant role in large-scale circulations and biogeochemical cycles. The ubiquitous Amazonian precipitation bias common in global climate models challenges our ability to understand changes in rainfall and the implications to the global water and carbon cycles. Previous studies suggested a variety of processes contributing to the precipitation bias, such as radiative transfer, terrestrial processes, and background large-scale circulations, to name a few. Here we first aim to identify model components responsible for the bias in the Community Atmosphere Model versions 5 (CAM5) at one-degree horizontal grid spacing. CAM5 exhibits spatially dependent bias of seasonal precipitation across the Amazon basin, and the most persistent is the underestimation over the southwestern Amazon and characterized by erroneous diurnal cycle. It is particularly puzzling during the wet season because the modeled evapotranspiration and easterly moisture flux are both rather overestimated. Our sensitivity simulations and concurrent community efforts point to the cumulus parameterization and numerical handling of topography as the largest contributors to the precipitation bias. Focusing on the cumulus parameterization, test simulations are carried out with several convection trigger functions adopted from recent studies, such as those based on the generation of convective available potential energy (CAPE) by resolved dynamics, modified calculation of CAPE, consideration of convective inhibition, and relaxation of parcel starting level. Together with the evaluation of these triggers using the MERRA reanalysis product, it is found that the functions performing well in the mid-latitude continents or oceanic regions do not necessarily work well over the southwestern Amazon during the wet season. Further analysis is ongoing to understand the dependence of the tested triggers on spatial scales and on the tropospheric moisture, which is found to be important to convection over the Amazon by recent observational studies.