This paper explores numerically the efficiency of $\lun$ minimization for the recovery of sparse signals from compressed sampling measurements in the noiseless case. Inspired by topological criteria for $\lun$-identifiability, a greedy algorithm computes sparse vectors that are difficult to recover by $\ell_1$-minimization. We evaluate numerically the theoretical analysis without resorting to Monte-Carlo sampling, which tends to avoid worst case scenarios. This allows one to challenge sparse recovery conditions based on polytope projection and on the restricted isometry property.