"... In PAGE 3: ...ituation where an object does not have the minimum properties required for execution, i.e. thread and execution-state, those of its user (caller) are used. Table1 shows the different abstractions possible when an object possesses different execution properties. Case 1 is objects, such as routines or class-objects that have none of the execution properties.... In PAGE 4: ... Currently, C++ only supports the explicit approach, but nothing in its design precludes implicit approaches. The abstractions in Table1 are expressed in C++ using two new type specifiers, uCoroutine and uTask, which are extensions of the class construct, and hence, define new types. In this paper, the types defined by the class construct and the new constructs are called class types, monitor types, coroutine types, coroutine-monitor types and task types, respectively.... ..."

"... In PAGE 5: ... It is also established in Section 4 that there is not much loss in generalityinhaving the gap parameter only at processors, and not at memory locations. 3 Algorithmic Results Table1 summarizes the time and work bounds for qsm algorithms for several basic problems. Most of these results are the consequence of the following four Observations, all of which are from #5B22#5D.... ..."

"... In PAGE 12: ... In practice, however, one must understand the implications of several other issues. Table3 shows how these issues relate to implicit assumptions OPMs make concerning the option being evaluated. ------ INSERT TABEL 3 ABOUT HERE ------ 4.... ..."

"... In PAGE 2: ... For more details refer to [17]. As an example, Table2 shows how NFS namespace operations are mapped to these two primitives. ... In PAGE 4: ...igure 2. Hard links and back pointers. Log record The proposed protocols make use of intention logs to record execution state in persistent storage. The structure of a log record is shown in Table2 . The fields refer to the name to object binding that is to be created or removed, in the case of link and unlink respectively.... ..."

"... In PAGE 2: ... For more details refer to [17]. As an example, Table2 shows how NFS namespace operations are mapped to these two primitives. ... In PAGE 4: ...igure 2. Hard links and back pointers. Log record The proposed protocols make use of intention logs to record execution state in persistent storage. The structure of a log record is shown in Table2 . The fields refer to the name to object binding that is to be created or removed, in the case of link and unlink respectively.... ..."

"... In PAGE 3: ... This paper considers ve such problems | generating a random permutation, multiple compaction, distributive sorting, parallel hashing, and load balancing | and presents fast, work-optimal qrqw pram algorithms for these fundamental problems. These results are summarized in Table1 , and are contrasted with the best known erew pram algorithms for the same problems. All of our algorithms are randomized, and are of the \Las Vegas quot; type;; they always output correct results, and obtain the stated bounds with high probability.... In PAGE 19: ... Examples of cyclic and noncyclic permutations are given in Figure 1. As indicated in Table1 , the best known linear work random permutation algorithm for the erew pram run in O(n ) time, for xed gt;0. This is also the best bound known for the random cyclic permutation problem.... In PAGE 26: ... Our result is for distinct keys. As shown in Table1 , the best known linear work erew pram algorithm for this problem runs in O(n ) time. 6.... In PAGE 29: ...1 Distributive Sorting The sorting from U(0;; 1) problem is to sort n numbers chosen uniformly at random from the range (0;; 1). As indicated in Table1 , the best known linear work erew pram algorithm for this problem runs in O(n ) time, for xed gt;0. erew pram algorithms that run in polylog time are work ine cien tbyatleasta p lg n lg lgn factor.... ..."

"... In PAGE 11: ...1 Distributive Sorting The sorting from U#280; 1#29 problem is to sort n numbers chosen uniformly at random from the range #280; 1#29. As indicated in Table1 , the best known linear work erew pram algorithm for this problem runs in O#28n #0F #29 time, for #0Cxed #0F#3E0. erew pram algorithms that run in polylog time are work ine#0Ecientby at least a p lg n lg lgn factor.... In PAGE 14: ... Our result is for distinct keys. As shown in Table1 , the best known linear work erew pram algorithm for this problem runs in O#28n #0F #29 time. 6.... In PAGE 21: ... Examples of cyclic and noncyclic permutations are given in Figure 1. As indicated in Table1 , the best known linear work random permutation algorithm for the erew pram run in O#28n #0F #29 time, for #0Cxed #0F#3E0. This is also the best bound known for the random cyclic permutation problem.... In PAGE 37: ... This paper considers #0Cve such problems | generating a random permutation, multiple compaction, distributive sorting, parallel hashing, and load balancing | and presents fast, work-optimal qrqw pram algorithms for these fundamental problems. These results are summarized in Table1 , and are contrasted with the best known erew pram algorithms for the same problems. All of our algorithms are randomized, and are of the #5CLas Vegas quot; type; they always output correct results, and obtain the stated bounds with high probability.... ..."