Since upgrading to SQL 2012 we've been having deallocks on local #tmp tables. This is occuring accross processes(spids). They appear to on partition objects, Sch-M and Sch-S. They are occuring on the ALTER of the #tmp tables. Below is the deadlock data. We've been running this code/procs for 7+ years with no problems. We've worked around the issue but would still like to know if a there is a behavior change in tempdb/sql2012. I can reproduce in SSMS running the procs in multple query windows. Similarly architected procs do not cause deadlocks. I can give more info if I haven't bored you yet.
Specs
This is a cluster
16CPUs
24GB ram
XEON 2660
Microsoft SQL Server 2012 - 11.0.2325.0 (X64) Enterprise Edition (64-bit) on Windows NT 6.1 <X64> (Build 7601: Service Pack 1)