"... In PAGE 15: ...1 Experiments on Detecting Deadlocks This section is dedicated to experiments with protocols that contain deadlocks. Table4 depicts experimental results with these protocols. For parameterized protocols, we have used the largest con guration that a breadth- rst search (BFS) can solve.... ..."

"... In PAGE 16: ...1 Experiments on Detecting Deadlocks This section is dedicated to experiments with protocols that contain deadlocks. Table4 depicts experimental results with these protocols. For parametrized pro- tocols, we have used the largest con guration that a breadth- rst search (BFS) can solve.... ..."

"... In PAGE 16: ...1 Experiments on Detecting Deadlocks This section is dedicated to experiments with protocols that contain deadlocks. Table4 depicts experimental results with these protocols. For parametrized pro- tocols, wehave used the largest conn0cguration that a breadth-n0crst search n28BFSn29 can solve.... ..."

"... In PAGE 16: ...1 Experiments on Detecting Deadlocks This section is dedicated to experiments with protocols that contain deadlocks. Table4 depicts experimental results with these protocols. For parametrized pro- tocols, wehave used the largest con guration that a breadth- rst search (BFS) can solve.... ..."

"... In PAGE 16: ...1 Experiments on Detecting Deadlocks This section is dedicated to experiments with protocols that contain deadlocks. Table4 depicts experimental results with these protocols. For parametrized pro- tocols, we have used the largest con guration that a breadth- rst search (BFS) can solve.... ..."

"... In PAGE 8: ... This is the elliptic curve discrete logarithm problem. Further on, we compare our protocols with other protocols (see Table1 ). The complexity is measured by the number of dominant computations (e.... ..."

"... In PAGE 11: ... The values presented in the tables are the percentage of cache lines in each con guration for each application that are \optimally quot; maintained using each protocol. For example, in Table5 , 34.9% of the cache lines in mp3d are read-only (and thus should be handled byany protocol but Migratory, 64/9% of the cache lines should be handled using the Migratory protocol, and 0.... ..."

"... In PAGE 18: ...3 Experiments on Detecting Assertion Violations Wehave a small group of models containing errors such as violation of assertions summarize as follows. Model Assertion Lynch apos;s Protocol i = last i +1 Barlett mr =(lmr + 1)% max Mutex in =1 Relay (k 14 1 =(s 1 1 ^:k 12 1 )) ^ (k 12 1 =(dienstv ^ (:s 1 1 _ k 12 1 )))^ (k 14 2 =(s 1 2 ^:k 12 2 )) ^ (k 12 2 =(dienstv ^ (:s 1 2 _ k 12 2 ))) ^ (dienstv =(k 14 1 _ k 14 2 ))) lt;= :(k 14 1 ^ k 14 2 ) GARP false Table7 depicts experimental results with these protocols. The data shows... ..."

"... In PAGE 19: ...Assertion Lynch apos;s Protocol i = last i +1 Barlett mr =(lmr + 1)% max Mutex in =1 Relay (k 14 1 =(s 1 1 ^:k 12 1 )) ^ (k 12 1 =(dienstv ^ (:s 1 1 _ k 12 1 )))^ (k 14 2 =(s 1 2 ^:k 12 2 )) ^ (kj12 2 =(dienstv ^ (:s 1 2 _ k 12 2 ))) ^ (dienstv =(kj14 1 _ k 14 2 ))) lt;= :(k 14 1 ^ k 14 2 ) GARP false Table7 depicts experimental results with these protocols. The results show that directed search strategies in HSF-SPIN o er shorter counterexamples for assertion violations than SPIN.... ..."

"... In PAGE 19: ...Assertion Lynch apos;s Protocol i = last i +1 Barlett mr =n28lmr + 1n29n25 max Mutex in =1 Relay n28k 14 1 =n28s 1 1 ^:k 12 1 n29n29 ^ n28k 12 1 =n28dienstv ^ n28:s 1 1 _ k 12 1 n29n29n29^ n28k 14 2 =n28s 1 2 ^:k 12 2 n29n29 ^ n28kj12 2 =n28dienstv ^ n28:s 1 2 _ k 12 2 n29n29n29 ^ n28dienstv =n28kj14 1 _ k 14 2 n29n29n29 n3c= :n28k 14 1 ^ k 14 2 n29 GARP false Table7 depicts experimental results with these protocols. The results show that directed search strategies in HSF-SPIN on0ber shorter counterexamples for assertion violations than SPIN.... ..."