<p>We perform a comparative study of the techniques for spatial processing of partially coherent signals with discrete spatial spectrum, which are received by a large antenna array in a randomly inhomogeneous multimode waveguide. The array gain in conventional terms of the signal-tonoise-plus-interference ratio is used as a criterion of the processing effectiveness. Our main attention has been paid to consideration of heuristically motivated suboptimal array processors, which, unlike the optimal processors, do not require complete information about the coherence functions (matrices) the useful signal and the interference at the array input. The key issue here is to estimate the range of the problem parameters where suboptimal processors turn out to be the most effective with an appropriate choice of their implementation parameters. Numerical calculations are carried out on the basis of a previously developed physical model of the partially coherent multimode signals, which corresponds to their formation at the input of a horizontal acoustic array arranged in a shallow-water acoustic channel (for a channel from typical of the Barents Sea in the summer season). It is shown that in the most complicated scenario where a relatively weak and coherence-degraded multimode signal is received against the background of intense multimode interference under conditions of a significant overlapping of their spatial (modal) spectra, the array gain for the proposed suboptimal techniques can reach values close to the maximum possible ones and significantly exceed the gain level determined by the total number of receiving-array elements.</p>