An analysis with Valgrind uncovered that the costliest functions are mutex locking/unlocking, malloc/cost-free and QImage dealing with. So I did some optimizations relating to QImage. In commit 1f976e5b5ee3cfe40921c74b97edb8d84ff39dd3 I reorganized the Smalltalk bitmap to QImage conversion which resulted in testStandardTests jogging time of 142 sec