[gmx-users] FMM in GMX

[Smart Eagle]

Hi,
     So, if the size of the simulation is big enough, FMM can be faster than PME.  Even though I heard that FMM is very hard to be coded into program, I just want to know the state of implement of FMM to GMX5.x .  
     It seems that FMM probabily can behave better than PME in massively parallelled architecture hardware, for example the mutliple GPU or CPU or MIC nodes clusters.  I quite look forword that.  
  

 cheers