Erik H. Baalbergen. Design and implementation of parallel make. Computing Systems, 1(2):135-158, Spring 1988.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....and writes back the result. When done, it generates the Goal 3 event that Final Goal requires. Preferably, the names of the goals would have descriptive names. All synchronization is done automatically without the use of threads or semaphores, much like in imple mentations of parallel make [1] and the inference engine of a parallel Prolog system (for example, 6] The program is therefore highly portable and efficient, and I believe that both parallelism and synchronization are specified at the right level of abstraction. Performance Figure 1 compares the performance of these three ....

Erik H. Baalbergen. Design and implementation of parallel make. Computing Systems, 1(2):135-158, Spring 1988.

....computations have not been changed by the time of the make request, the results of the optimistic computations are used. Otherwise, the necessary computations are reexecuted. Thus optimistic make requires minor modifications to the kernel and to some of the servers. Parallel make (pmake) [1] speeds up the operation of make by doing compilations in parallel. For each command block to be executed, pmake creates a child process called a virtual processor. pmake doesn t wait for the virtual processor to finish, but continues processing the list of dependencies in which the target ....

E.H. Baalbergen. Design and implementation of parallel make. Computing Systems, 1(2):135--158, Spring 1998.

....the pool processors, a group of CPUs that can be dynamically allocated as needed, used, and then returned to the pool. For example, the make command might need to do six compilations, so six processors could be taken out of the pool for the time necessary to do the compilation and then returned [12]. Third are the specialized servers, such as directory, file, and block servers, data base servers, bank servers, boot servers, and various other servers with specialized functions. Fourth are the wide area network gateways, which are used to link Amoeba systems at different sites in possibly ....

Baalbergen, E.H.: "Design and Implementation of Parallel Make," Computing Systems---The Journal of the USENIX Association , vol. 1, Spring 1988.

....or dynamically changing files. The dynamically allocatable processors together form the so called processor pool . These processors may be allocated for compiling or text formatting purposes, or for distributed or parallel algorithms. Among other applications, we have implemented a parallel make [10] and parallel heuristic search [11] Specialized servers include filing servers such as the Bullet file server, and the directory server. The directory server is used in conjunction with the Bullet server. It s function is to handle naming and protection of Bullet server files and other objects in ....

Baalbergen, E. H., "Design and Implementation of Parallel Make," Computing Systems, Vol. 1, No. 2, pp. 135-158 (Spring 1988).

....such as yacc (Johnson, 1978) that write their output on fixed name files, such as y.tab.c . When multiple yacc s are running in the same directory, they all write to the same file, thus producing gibberish. All of these problems have been dealt with by one means or another, as described in (Baalbergen, 1988). The parallel compilations are directed by a new version of make , called pmake , based on the UNIX one but with additional code to handle parallelism. The makefiles accepted by this program are compatible with the standard one. The performance of pmake depends strongly on the input. When making ....

Baalbergen, E.H. Design and Implementation of Parallel Make, Computing Systems 1., (Spring 1988), pp. 135-158.

....would be better if the dependencies were visible to make. Most concurrent make oids (i. e sequential make oids which spawn several compilations concurrently) rely on explicit user defined annotation of the make rules to indicate potential safe concurrency: these include DYNIX make [Sequ87] pmake [Baal88] and nmake [Fowl90] and it would be better if: compilations were not invoked within a public namespace (they may unwantedly overwrite each other s intermediate and output files) each make oid sufficiently understood the actions and their interdependencies so as to be able to schedule them ....

E. H. Baalbergen, "Design and Implementation of Parallel make", Journal of Computing Systems, vol. 1, no. 2, pp. 135-158 (1988).

....a network of workstations. Finally, Section 8.4 discusses the measured effect of group management when multiple processor sharing applications are run simultaneously in a network. 8.1. Parallel Make Parallel make is a popular processor sharing application that has already been studied extensively [3, 4, 22]. Parallel make is similar to the traditional UNIX make program, except that parallel make runs several compilations simultaneously on a network of machines whereas traditional make invokes compilations sequentially on one machine. Thus, the running time of parallel make can be much shorter than ....

....that of its traditional counterpart. However, other researchers have found that it is difficult for parallel make to achieve very good speedup. Douglis reports speedups ranging from 2.95 to 5.42 for parallel make running on twelve machines [24] Baalbergen reports a speedup of 3. 81 on ten machines [4]. The relatively low speedup numbers are attributed both to some measure of inherent sequentiality in the program and to file system bottlenecks. Still, parallel make is popular among software developers because it does run faster than sequential make. The performance of parallel make in ARCADE is ....

Baalbergen, Erik H., "Design and Implementation of Parallel Make", Computing Systems Vol. 1, Spring 1988, pp. 135-158.

No context found.

E. H. Baalbergen, "Design and Implementation of Parallel Make," Computing Systems, vol. 1, no. 2, pp. 135-158, Spring 1988.

No context found.

E.H. Baalbergen. "Design and Implementation of Parallel Make", Computing Systems, Vol. 1, No. 2, pages 135-158, 1988.

No context found.

Baalbergen, E.H., "Design and Implementation of Parallel Make," Computing Systems, vol. 1, pp. 135-158, Spring 1988.

No context found.

Baalbergen, E.H.: "Design and Implementation of Parallel Make," Computing Systems , vol. 1, pp. 135-158, Spring 1988.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC