On the amos-help mailing list https://lists.sourceforge.net/lists/listinfo/amos-help I just saw that there is an extra define flag to mummer to allow for larger memory usage, which is necessary for larger datasets. I propose this to be default. make CPPFLAGS="-O3 -DSIXTYFOURBITS" and that will compile it in 64-bit mode which will use more memory, but remove the max input seq length limit. >On Wed, Jan 12, 2011 at 5:30 AM, Nagarjun V <arjun53ster@xxxxxx.xxx> wrote: >> Could anybody please let me know if Minimus2 can be made to handle much >> larger datasets? while using the latest version of Minimus2 i run into a >> limit when the reads are very long.
Does -DSIXTYFOURBITS works on x86? -O3 is definetly a no, because optimization is users business.
+*mummer-3.22-r1 (13 Jan 2011) + + 13 Jan 2011; Justin Lecher <jlec@gentoo.org> +files/3.22-ldflags.patch, + +files/3.22-prll.patch, -mummer-3.22.ebuild, +mummer-3.22-r1.ebuild: + Respecting *FLAGS, #335447; added real 64bit support, #351502 +
I tested whether the binary compiled with -DSIXTYFOURBITS works on 32bit, but it does not. It somehow blocks. It prints some output before doing a search but that is all. Here is the output of both 64bit and 32bit version for some clue. The current 3.22-r1 ebuild correctly forces this CFLAG for only amd64 arch: 32-bit binary on 32bit computer: # reading input file "test.dna" of length 176189 # construct suffix tree for sequence of length 176189 # (maximum reference length is 536870908) # (maximum query length is 4294967295) # process 1761 characters per dot #.................................................................................................... [real results were cut away] 64bit binary on 32bit computer: # reading input file "test.dna" of length 176189 # construct suffix tree for sequence of length 176189 # (maximum reference length is 1073741820) # (maximum query length is 4294967295) # process 1761 characters per dot [somehow locked process]