Ex Parte McKenney et al - Page 8

                  Appeal 2007-1600                                                                                         
                  Application 09/753,062                                                                                   

                         With respect to claims 9 and 10, Appellants argue that Kermani lacks                              
                  any teaching of a “data structure” or “bit mask,” let alone such an item                                 
                  indicating which processor is waiting for the lock. We agree that the specific                           
                  terms are not present in Kermani. Fig. 2 of Kermani, however, illustrates a                              
                  plurality of agents (numbered 1-n), each of which sends a memory access                                  
                  request to arbiter 102a, which includes a priority encoder 190 (FF 5). A                                 
                  memory access request by an agent is a digital signal that indicates that said                           
                  agent desires access to the shared synchronous memory (FF 6). As is well                                 
                  known in the art, a priority encoder operates by receiving a plurality of                                
                  inputs (bits), and selecting the input having the highest priority. We find that                         
                  the set of memory access request signals, taken together, may fairly be                                  
                  characterized as a “data structure” having a “bit mask” indicating which                                 
                  processors (agents) are waiting for the lock (i.e., waiting to access shared                             
                  synchronous memory). We therefore affirm the Examiner’s rejection of                                     
                  claim 9, as well as the Examiner’s rejection of claim 10, not separately                                 
                  argued by Appellants.                                                                                    
                         With respect to claims 11, 20, and 30, Appellants argue that there is                             
                  no express or inherent teaching in Kermani of the use of a release flag to                               
                  prevent races between processors between acquisition and release of a lock                               
                  (Br. 9). We agree with Appellants. While the section of Kermani cited by the                             
                  Examiner (see FF 8) does discuss an “open window” generated when super                                   
                  agent A is not accessing the shared synchronous memory, there is no                                      
                  teaching of a “release flag” or any other mechanism in Kermani to prevent                                




                                                            8                                                              

Page:  Previous  1  2  3  4  5  6  7  8  9  10  Next

Last modified: September 9, 2013