<p>Evolutionary multitask optimization (EMTO) is an emerging topic in evolutionary computation. Compared with single-task optimization, EMTO can solve multiple related or similar tasks simultaneously by transferring knowledge between them. High-performance computing (HPC) provides powerful computational capabilities for solving complex multitask optimization problems, making EMTO possible for real-world applications. Although EMTO has shown great application potential, it still faces challenges in solving accuracy, requiring in-depth research and improvement. Therefore, we propose an EMTO algorithm based on block-level knowledge transfer (BLKT) and beluga whale optimization (BWO), namely BLKT–BWO. It mainly consists of a knowledge transfer module and an independent evolution module. Specifically, in the knowledge transfer module, we introduce a simple and efficient weighted average knowledge transfer rule to reduce computational complexity. In addition, by dividing and clustering individuals, the algorithm achieves knowledge transfer between similar but unaligned dimensions. This approach accelerates the solution process and helps tasks to escape from local optimum. In the independent evolution module, the BWO algorithm is employed as a solver for updating positions to enhance global convergence, owing to its strong search performance in single-objective optimization. Experimental results on benchmarks of CEC2017-MTSO and WCCI2020-MTSO, and a real-world multitask optimization problem all demonstrate that the performance of BLKT–BWO is superior to current state-of-the-art algorithms.</p>