"... In PAGE 31: ... 1.2 The reduction relation The reduction relation of the -calculus is the least relation satisfying the rules in Table1 , The rules (T-DONE) and (T-ABORT) deserve some discussion. (T-DONE) models the successful comple- tion of a transaction done .... In PAGE 32: ...Comparison with the -calculus 19 RED T-DONE done T-ABORT abort CONTEXT LIFT (5) Table1 . The reduction rules of the -calculus outer transaction fails, rule (T-ABORT) specifies that the failure manager must be executed after the compensation of every enclosed transaction.... In PAGE 55: ... , where 10. , where Table1 . Possible transitions for an atomic commit protocol from to id AC6 AC6 AC4 AC7 AC6 AC6 id val AC6 AC3 AC6 AC5 AC2 AC3 id AC2 AC2 AC5 AC2 AC2 AC2 AC6 AC2 id AC6 AC2 AC5 AC6 AC6 AC2 AC2 AC6 id AC2 AC6 AC5 AC2 AC2 AC2 AC2 AC2 id AC2 AC2 AC2 AC6 AC2 AC2 AC2 AC2 id AC6 AC6 AC2 AC2 AC2 AC6 AC2 AC6 id (3) Each of the properties imposes constraints on which inter-state transitions are disallowed, or which must exist.... In PAGE 55: ... means there exists a sequence . Table1 summarizes all the transitions, and an explanation follows. By , the two states and are never reached;... In PAGE 56: ...Table 2. An atomic commit protocol d11 d19 d31 d39 d71 d71 d71 d71 d71 d71 d71 d71 d71 d71 d11 d19 (4) To explain, Table1 shows whether some sequence of transitions from a state (row) to a state (column) is possible or not. For instance, a sequence from is possible due to .... In PAGE 57: ...Proof 1 Straightforward from Table1 lookup. 3 General Rendezvous and Atomic Commit As mentioned, traditional synchronous rendezvous from the pi calculus is a special case of atomic commit.... In PAGE 60: ... The process represents the choice, can create an arbitrary number of copies of . The pi calculus is summarized in Table1 : labeled transitions define the possible re- actions of a process, contexts are processes with holes filled by other processes, and represent environments. Simulation is a relation characterizing when two processes have the same behavior.... In PAGE 61: ... Table1 . The asynchronous pi calculus Terms in the asynchronous pi calculus are as follows.... In PAGE 69: ... We have for inductive hypothesis that , . It is straightforward that where By the standard results properties of ( Table1 ) we have the following: sv m col (3.3.... In PAGE 82: ... Let us consider a generic node . We have for inductive hypothesis that (8) By the standard results properties of ( Table1 ) we have the following: (9) For Lemma 6.... In PAGE 91: ...denote an instance of pi-CNP with the strategy of Table1 . A preliminary lemma and correctness of the encoding are presented below.... In PAGE 105: ...0.1 A Classification of Transactional Behaviors Table1 . Transaction types in the WSA and their relation out in unprotected protected semi-protected real negotiable unprotected yes no no no yes protected yes yes no no yes semi-protected yes yes yes yes yes real yes yes yes yes yes negotiable yes yes yes yes yes depends in part on the nature of its enclosed transactions (e.... In PAGE 105: ... Such services may advertise more than one of the defined de- grees in their description and manifest the actual compensational behavior depending on the requirements imposed by the system. In Table1 we present the categorization of transactions in the Web scenario together with the less restrictive relation representing the possible nesting relations between them. The cell represents the possibility of nesting the transaction type within the transaction type .... ..."

"... In PAGE 31: ... 1.2 The reduction relation The reduction relation of the -calculus is the least relation satisfying the rules in Table1 , The rules (T-DONE) and (T-ABORT) deserve some discussion. (T-DONE) models the successful comple- tion of a transaction done .... In PAGE 32: ...Comparison with the -calculus 19 RED T-DONE done T-ABORT abort CONTEXT LIFT (5) Table1 . The reduction rules of the -calculus outer transaction fails, rule (T-ABORT) specifies that the failure manager must be executed after the compensation of every enclosed transaction.... In PAGE 55: ... means there exists a sequence . Table1 summarizes all the transitions, and an explanation follows. By , the two states and are never reached;... In PAGE 56: ...Table 2. An atomic commit protocol d11 d19 d31 d39 d71 d71 d71 d71 d71 d71 d71 d71 d71 d71 d11 d19 (4) To explain, Table1 shows whether some sequence of transitions from a state (row) to a state (column) is possible or not. For instance, a sequence from is possible due to .... In PAGE 57: ...Proof 1 Straightforward from Table1 lookup. 3 General Rendezvous and Atomic Commit As mentioned, traditional synchronous rendezvous from the pi calculus is a special case of atomic commit.... In PAGE 60: ... The process represents the choice, can create an arbitrary number of copies of . The pi calculus is summarized in Table1 : labeled transitions define the possible re- actions of a process, contexts are processes with holes filled by other processes, and represent environments. Simulation is a relation characterizing when two processes have the same behavior.... In PAGE 61: ... Table1 . The asynchronous pi calculus Terms in the asynchronous pi calculus are as follows.... In PAGE 69: ... We have for inductive hypothesis that , . It is straightforward that where By the standard results properties of ( Table1 ) we have the following: sv m col (3.3.... In PAGE 82: ... Let us consider a generic node . We have for inductive hypothesis that (8) By the standard results properties of ( Table1 ) we have the following: (9) For Lemma 6.... In PAGE 90: ... Table1 . The strategy for atomic commit localc localf if localf otherwise if localc localf otherwise commit protocols, have been used to define the high level semantic for correct atomic commit pro- tocols: a transition system is an atomic commit protocol iff it is bisimilar to the following diagram: d11 d19 d31 d39 d71 d71 d71 d71 d71 d71 d71 d71 d71 d71 d11 d19 (1) Also, that general rendezvous mechanism presented in 3, is used to denote an high level atomic commit.... In PAGE 91: ...denote an instance of pi-CNP with the strategy of Table1 . A preliminary lemma and correctness of the encoding are presented below.... In PAGE 105: ...0.1 A Classification of Transactional Behaviors Table1 . Transaction types in the WSA and their relation out in unprotected protected semi-protected real negotiable unprotected yes no no no yes protected yes yes no no yes semi-protected yes yes yes yes yes real yes yes yes yes yes negotiable yes yes yes yes yes depends in part on the nature of its enclosed transactions (e.... In PAGE 105: ... Such services may advertise more than one of the defined de- grees in their description and manifest the actual compensational behavior depending on the requirements imposed by the system. In Table1 we present the categorization of transactions in the Web scenario together with the less restrictive relation representing the possible nesting relations between them. The cell represents the possibility of nesting the transaction type within the transaction type .... ..."

"... In PAGE 11: ... In these results, the per-path stacks are copied in a single cycle. Table8 presents return-prediction accuracies; note that for the no-stack case, this accuracy presents the return-address hit-rate in the BTB. Performance is normal- ized to the conventional, unified-stack case.... ..."

"... In PAGE 11: ... In these results, the per-path stacks are copied in a single cycle. Table8 presents return-prediction accuracies; note that for the no-stack case, this accuracy presents the return-address hit-rate in the BTB. Performance is normal- ized to the conventional, unified-stack case.... ..."

"... In PAGE 11: ... In these results, the per-path stacks are copied in a single cycle. Table8 presents return-prediction accuracies; note that for the no-stack case, this accuracy presents the return-address hit-rate in the BTB. Performance is normal- ized to the conventional, unified-stack case.... ..."

"... In PAGE 11: ... In these results, the per-path stacks are copied in a single cycle. Table8 presents return-prediction accuracies; note that for the no-stack case, this accuracy presents the return-address hit-rate in the BTB. Performance is normal- ized to the conventional, unified-stack case.... ..."

"... In PAGE 9: ...Table 7: Comparison with 2PC protocols. Table7 summarizes the comparison between the 2PC PrN and the protocols introduced in this paper. Although there exist 2PC optimizations, such as the presumed commit (PrC) variation [13, 11], the important performance features, including the number of synchronous log accesses and message round-trips on the critical path would still be the same.... ..."

"... In PAGE 9: ...Table 7: Comparison with 2PC protocols. Table7 summarizes the comparison between the 2PC PrN and the protocols introduced in this paper. Although there exist 2PC optimizations, such as the presumed commit (PrC) variation [13, 11], the important performance features, including the number of synchronous log accesses and message round-trips on the critical path would still be the same.... ..."

"... In PAGE 36: ....3. Direct Mapping Technique (DM) This method consists of encoding each data bit as phase difference between signals on two particular non-adjacent wires as illustrated in table 1. Data Wires I1 a amp; c I2 a amp; d I3 b amp; d Table1 : Direct Mapping for 4 wire phase Encoded Channel Direct mapping masks all type 1 faults as phase between adjacent wires does not carry any information. It also allows the detection and/or the correction of type 2 faults if extra testing circuitry is included in the design.... In PAGE 80: ... From the above discussed results, it is evident that any new test generation algorithm to be developed should focus on testing the intermediate nodes which will be overseen by the circuit models which are modeled with only the input and output signals of each gate. Table1 gives the fault coverage obtained for several benchmarks using testify. Table 1: Fault coverage using symbolic technique Table2: Fault coverage using scan insertion technique The fault coverage and test patterns based on the second approach was generated by cutting the global loops manually and by inserting the scan latch at the feedback paths.... In PAGE 80: ... Table 1 gives the fault coverage obtained for several benchmarks using testify. Table1 : Fault coverage using symbolic technique Table2: Fault coverage using scan insertion technique The fault coverage and test patterns based on the second approach was generated by cutting the global loops manually and by inserting the scan latch at the feedback paths. After inserting the latches, the netlist was fed to the Synopsys Tetramax ATPG tool to generate the test patterns and calculate the fault coverage.... ..."