The authors compare two program-comprehension systems targeted to support a
utomatic parallelization: PARAMAT and the PAP Recognizer. The authors illum
inate the main differences of each method, discussing the trade-off between
the generality of one and speed of the other.