[codes-ross-users] codes_mapping_setup() seg fault and RNG issue

[Jenkins, Jonathan P.]

Hi Joe,

Looks like a mismatch in CODES/ROSS revisions. A recent revision of CODES modified the # of RNGs requested by LPs, which required some fixes to ROSS's RNG management code. The ROSS fix was introduced in commit hash 119e16c, "Merge pull request #19 from JohnPJenkins/rm-g_tw_rng_max", which I've used in fully-updated CODES repos just fine. ROSS HEAD should work OK, but I'm holding off on updating my local copy until a few features on their end get finished up.

Thanks,
John
________________________________
From: codes-ross-users-bounces at lists.mcs.anl.gov [codes-ross-users-bounces at lists.mcs.anl.gov] on behalf of Timothy Scott [tscott2 at clemson.edu]
Sent: Wednesday, January 14, 2015 10:32 AM
To: codes-ross-users at lists.mcs.anl.gov
Subject: [codes-ross-users] codes_mapping_setup() seg fault and RNG issue

Hello All,

I am back trying to get a basic model set up in CODES.  I am using a variation of doc/heterogenous (ping with two groups and forwarders).

I also found that the doc/example_heterogeneous fails in the same way.

In codes_mapping_setup_with_seed_offset() called with parameter 0 (from codes_mapping_setup()), tw_define_lps() prints

~/src/ROSS/ross/rand-clcg4.c:289: RNG max streams exceeded: 2 > 1.

I found that if I manually g_tw_rng_max to a number greater than 2 in codes_mapping_setup_with_seed_offset() before the tw_define_lps() call, the code runs fine. I can't find any failure caused by this.

Is this a bug or am I (and the example) invoking something wrong?

Thanks,
Joe
--
Joe Scott
PhD Student
Clemson University
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/codes-ross-users/attachments/20150114/01ea3bea/attachment.html>