Greetings! One other thought -- it would also be useful presuming 1) is successful to just try `make clean && make` after you reset to 1Gib and before reconfiguring. Then proceed with test 2). I want to know if a system compiled successfully with 2MiB hugepages can run on a system with a different setting.
Take care, [ This email came through directly ] David Scherfgen via Maxima-discuss <maxima-disc...@lists.sourceforge.net> writes: > As an experiment, I have disabled hugepages. The complete build works just > fine now! > That indicates that there's nothing wrong with my system, hardware, compiler > etc. Wonderful! Now we are getting somewhere. Two experiments to try: 1) set your hugepages to 2MiB and configure && make with no options 2) set your hugepages to 1GiB, configure with the --enable-min-pagewidth=21, and then before make do echo #undef HAVE_MADVISE_HUGEPAGE >>h/config.h Then make. I suspect that even though the madvise call succeeds without error when the pagesize mismatches your system pagesize, it is doing something bad under the hood. If both 1) and 2) succeed, then we just need to figure out how to gracefully handle really huge hugepages. Take care, -- Camm Maguire c...@maguirefamily.org ========================================================================== "The earth is but one country, and mankind its citizens." -- Baha'u'llah
