This paper provides a comparison of the computational performance of D-Wave quantum annealers and IBM quantum computers. The discrete knapsack problem was selected as the benchmark. The study includes a Quadratic Unconstrained Binary Optimization (QUBO) problem formulation, a comparison of two constraint enforcement methods, and the design of a quantum circuit realizing the Quantum Approximate Optimization Algorithm (QAOA). Computations on quantum annealers were performed using D-Wave Leap and the Ocean SDK, while computations on gate-based quantum computers were conducted using the IBM Quantum Platform and Qiskit. Quantum annealing proved to be a more efficient approach for solving the considered problem due to its significantly shorter computation time, lower cost, and greater size of solvable problems. For instances characterized by high knapsack capacity, quantum annealing was found to be faster than the classical dynamic programming algorithm, making it an interesting alternative to classical computation for this subset of knapsack problem instances. Results, including solution value, its difference from the optimum, total computation time, Quantum Processing Unit time, and computation costs, were presented in tables and visualized in charts.