This paper presents an analysis of performance improvement of two CFD codes, the structured two-dimensional code GHOST and the unstructured two-/three-dimensional code UNCLE, both in-house codes at the University of Kentucky, with the specific target of cache performance optimization. The primary technique of choice is the blocking of data arrays, either within specific subroutines or of the computational grid as a whole. Both methods have produced significant improvements in performance without compromising accuracy. More advanced 64-bit commodity hardware reduces some of the relative gain from blocking due to its improved bandwidth and memory structure, but significant gains in performance are still seen. This study suggests that as cache sizes continue to increase, blocking may prove a general useful tool for accelerating CFD code performance.