M. Jung, S. Kim, H. Han, J. Choi, and S. K. Cha, B2r2: Building an efficient front-end for binary analysis, in The NDSS Workshop on Binary Analysis Research, Internet Society, 2019.
Hash-Consing (idea: de-duplicate subtrees of the AST, identical nodes shares the same underlying memory, and nodes are compared by hashing node contents)
Parallelism
Arbitrary-precision integers: costly on memory and CPU.
Store use-def info in metadata to speed up analysis. (Any IR in SSA form gets this for free).
M. Jung, S. Kim, H. Han, J. Choi, and S. K. Cha, B2r2: Building an efficient front-end for binary analysis, in The NDSS Workshop on Binary Analysis Research, Internet Society, 2019.
https://softsec.kaist.ac.kr/~jschoi/data/ndssbar2019.pdf