[developers] -limit, -memlimit, -nsolutions, packing

Bart Cramer bart.cramer at gmail.com
Mon Oct 20 18:10:22 CEST 2008


Dear developers,

I am currently using PET, and it seems I have one particularly
pathological sentence in my test suite, making it >10 GBs large. My
parameters were:

-limit=50000 -packing=15

I heard that, in order for selective unpacking to happen at all,
nsolutions has to be set. Is this true (wiki!)? Anyway, I set it to:

-limit=50000 -packing=15 -nsolutions=1000

To no avail. A somewhat rougher cutoff would then be to limit the memory:

-limit=50000 -packing=15 -nsolutions=1000 -memlimit=512

On the Delph-in Summit, it was said there was a factor 2 either way
involved (which way?), but that didn't seem to be the case here.
Again, it ran beyond 4 GB, and decided to stop it. Setting
-memlimit=10 did help (made it 16M big), -memlimit=30 did not. As a
workaround, I could use (non-deterministic) timeouts:

-timeout=5

This works. However, it does not work in combination with packing=15
or packing=7, which is obviously what we want:

-timeout=5 -packing=7 -nsolutions=1000
-timeout=5 -packing=15 -nsolutions=1000

The version I am using is 64-bit, latest SVN. Am I running into bug(s)
here? Or am I not using the parameters properly? I hope someone can
clarify these issues!

Best,

Bart.

PS: I made some changes to the wiki-page. Maybe qualified persons can
add more information (and remove less interesting things)?



More information about the developers mailing list