Bug in DPSK-Mod/Error-Rate in v3.6.3&v3.6.4 not in v3.6.2

Hello G.s.

I made a very simple flowgraph from stock blocks and after a few
seconds(3-5) it unexpectedly stops simulating.

I have a random source - Throtte - DPSKMod - DPSKDemod - Errorrate -
Scope/Filesink

Both the Scope and the Filesink stop receiving data after a few seconds,
although it should run forever.
Internally it does not report any error and ‘runs’ until I close it.

The Bug occurs in v3.6.3 & 3.6.4 but not in v3.6.2

‘Tcollins’ from Gnuradiochat confirmed 3.6.3 not working &3.6.2 is
working

Here for .grc code: <?xml version='1.0' encoding='ASCII'?><flow_graph> <timestamp>Mon Mar 18 19 - Pastebin.com
Here for pythoncode (generated from .grc): Bug - Pastebin.com

Installed version try 1: GNU Radio v3.6.3-124-gf1ae6c5c ; Ubuntu 32 Bit
12.04 ; GNU C++ version 4.6.3; Boost_104601;
UHD_003.005.001-29-g3cb515f7
Installed version try 2: GNU Radio v3.6.4-86-g77743314 ; Ubuntu 32 Bit
12.04; GNU C++ version 4.6.3; Boost_104800;
UHD_003.005.001-37-g21311276

What I noticed:
With each of the version I get a specific number of Bytes in the
filesink (repeatable):
v3.6.3: 532,480 byte
v3.6.4: 515,580 byte

The Cpu-Load is in the beginning at 20% and once it hangs it drops to 1%