Hello. 

There is an issue in particle_hp model which results in unlimited growth of G4IonTable. 

If we are dealing with an isotope which 
- does not have determined excitation levels for some interaction; 
- does not have angular of energy-angular distributions for outgoing particle; 
- has photon data represented as multiplicities and contimuum or arbitary energy distributions; 
then line #608 in G4ParticleHPInelasticCompFS.cc 
G4DynamicParticle* targ = new G4DynamicParticle( G4IonTable::GetIonTable()->GetIon ( (G4int)theBaseZ , (G4int)theBaseA , totalPhotonEnergy )  , G4ThreeVector(0) ); 
inserts a new entry into G4IonTable via consecutive calls of GetIon(), FindIon() , FindIonInMaster(), CreatIon() and InsertWorker(). 

When G4IonTable tries to find an ion it checks, among other things, the excitation energy. Sampling the energy and multiplicity of final state photons from G4NDL creates a unique "totalPhotonEnergy" which later is used as an excitation. Thus G4IonTable is expanding and worker threads are waiting longer to acquire the mutex for it. 

This behaviour could be reproduced shooting 14MeV neutrons into Mg target. The (n,p) and (n,a) reactions generate photons which are sampled from distributions. Even on a 4-core machine the speed halves in several minutes from starting the execution. 

Proposed change for line 608: 
G4DynamicParticle* targ = new G4DynamicParticle( G4IonTable::GetIonTable()->GetIon ( (G4int)theBaseZ , (G4int)theBaseA , theFinalStatePhotons[it] ? theFinalStatePhotons[it]->GetLevelEnergy() : 0 )  , G4ThreeVector(0) );