October 1974 - Vol. 17 No. 10
Features
Monitors: an operating system structuring concept
This paper develops Brinch-Hansen's concept of a monitor as a method of structuring an operating system. It introduces a form of synchronization, describes a possible method of implementation in terms of semaphores and gives a suitable proof rule. Illustrative examples include a single resource scheduler, a bounded buffer, an alarm clock, a buffer pool, a disk head optimizer, and a version of the problem of readers and writers.
A weighted buddy method for dynamic storage allocation
An extension of the buddy method, called the weighted buddy method, for dynamic storage allocation is presented. The weighted buddy method allows block sizes of 2k and 3·2k, whereas the original buddy method allowed only block sizes of 2k. This extension is achieved at an additional cost of only two bits per block.
Simulation results are presented which compare this method with the buddy method. These results indicate that, for a uniform request distribution, the buddy system has less total memory fragmentation than the weighted buddy algorithm. However, the total fragmentation is smaller for the weighted buddy method when the requests are for exponentially distributed block sizes.
A note on the calculation of average working set size
Finite-length reference string of arbitrary structure are considered, and an exact expression for average working set size in terms of “corrected” interreference interval statistics is derived. An example is discussed; upper and lower bounds are obtained; and the average working set size function is shown to be efficiently obtained for a set of page sizes, in a single pass of the reference string.
This work follows the developments of a paper by Denning and Schwartz, who consider infinite-length reference strings which satisfy certain statistical properties and who derive an expression relating the asymptotic average working set size to the asymptotic missing page rate function under working set replacement.
Programming systems which permit arbitrary linked list structures enable the user to create complicated structures without sufficient protection. Deletions can result in unreachable data elements, and there is no guarantee that additions will be performed properly. To remedy this situation, this paper proposes a Data Structure Description and Manipulation Language which provides for the creation of a restricted class of data structures but ensures the correctness of the program. This is accomplished by an explicit structure declaration facility, a restriction on the permissible operations, and execution-time checks.
A back-end computer for data base management
It is proposed that the data base management function be placed on a dedicated back-end computer which accepts commands (in a relatively high level language such as the CODASYL Data Base Task Group, April 1971 Report) from a host computer, accesses the data base on secondary storage, and returns results. The advantages of such a configuration are discussed. An experimental implementation, called the eXperimental Data Management System, XDMS, is described and certain conclusions about the back-end approach are drawn from this implementation.
On generation of test problems for linear programming codes
Users of linear programming computer codes have realized the necessity of evaluating the capacity, effectiveness, and accuracy of the solutions provided by such codes. Large scale linear programming codes at most installations are assumed to be generating correct solutions without ever having been “bench-marked” by test problems with known solutions. The reason for this failure to adequately test the codes is that rarely are there large problems with known solutions readily available. This paper presents a theoretical justification and an Illustrative implementation of a method for generating linear programming test problems with known solutions. The method permits the generation of test problems that are of arbitrary size and have a wide range of numerical characteristics.
Algorithm 486: Numerical inversion of Laplace transform [D5]
This work forms part of a thesis presented in Grenoble in March 1972. Improvements made to the Dubner and Abate algorithm for numerical inversion of the Laplace transform [1] have led to results which compare favorably with theirs and those of Bellmann [2], and Stehfest [3]. The Dubner method leads to the approximation formula: ƒ(t) = 2eat/T[1/2Re{F(a)} + ∑∞k-1 Re{F(a + ik&pgr;/T)}cos(k&pgr;t/T)], (1) where F(s) is the Laplace transform of ƒ(t) and a is positive and greater than the real parts of the singularities of ƒ(t).
Enumerating full-time programmers
Data from the 1970 Census and the Department of Labor's Area Wage Surveys are used to derive estimates of the number of full-time programmers employed during the years 1969 through 1973. The 1973 figure of 180,000 is considerably less than suggested in earlier reports. It is recommended that educational administrators consider whether the many courses aimed at training programmers are justified on a vocational basis.
