12 Jun
                
                    2015
                
            
            
                12 Jun
                
                '15
                
            
            
            
        
    
                8:49 p.m.
            
        I hadn't noticed you'd done that work and it is interesting but no, I happened to be thinking more along the lines of a loosely coupled cluster of heterogeneous architectures (say ARM, x86 and Power) and perhaps using Byzantine fault tolerant replication so as to lift the API over any n failures, compromised machines or hardware design bugs.