you modified the number of memory and threads, but did not improve performance because of the changes in these parameters, indicating that memory was abundant and the number of threads was not a bottleneck.
the first thing to do is measure . Without measurement optimization, there is no goal. You need to measure the time spent in each phase of the system. The database aspect is usually the most time-consuming part. Connection pooling solves the problem of shortening the time it takes to establish a connection. The number of connections is generally only 2030 by default (you do not provide the corresponding information, it is not clear what your setting is), your 400 concurrency may not be enough connection pooling. Even if sufficient, the query time will not be reduced. The most effective way to optimize database queries is usually appropriate caching.

the test results are the same before and after optimization, indicating that it is not optimized to the bottleneck, indicating that the bottleneck of the system is not the configuration of tomcat.
I do not know what the response time of the system is without so much pressure. If it is still so slow, first optimize the database query and so on