wilson@uicbert.eecs.uic.edu (09/22/88)
I'm looking for empirical data on the performance of generation-based garbage collectors while running realistic programs. If anybody out there has such data, I would be very interested in seeing it. (I already have Bob Shaw's dissertation, which I am depending on too heavily. I need more data points. I also have Courts' new CACM paper, Ungar and Jackson's OOPSLA paper, and Ungar's older stuff.) I am interested in ANY of the following kinds of data: 1. rates of allocation and of scavenge survival, even if only for the first generation/level 2. frequency and/or locality of inter-generational references stored into the heap (perhaps in terms of number of pages holding them) 3. frequency and/or locality of all writes to older generations If you have any such data and are willing to share it, please send me a note. Even very rough figures would be fine, either for compute-bound programs or interactive sessions. Thanks, Paul Paul R. Wilson Human-Computer Interaction Laboratory U. of Illin. at C. EECS Dept. (M/C 154) wilson%uicbert@uxc.cso.uiuc.edu Box 4348 Chicago,IL 60680