A study of program locality and lifetime functions

In a two-level computer storage hierarchy, miss ratio measurements are often made from a '*cold start", that is, made with the first-level store initially empty. For large capacities the effect on the measured miss ratio of the misses incurred while filling the first-level store can be significant, even for long reference strings. Use of "warm-start" rather than "cold-start" miss ratios cast douht on the widespread helief that the observed "S-shape" of lifetime (reciprocal of miss ratio) versus capacity curve indicates a property of hehavior of programs that maintain a constant number of pages in main storage. On the other hand, if cold-start miss ratios are measured as a function of capacity and measurement length, then they are useful in studying systems in which operation of a program is periodically interrupted by task switches. It is shown how to obtain, under simple assumptions, the cache miss ratio for multiprogramming from cold-start miss ratio values and how to obtain approximate cold-start miss ratios from warm-start miss ratios.

Section: Shape Of the Lifetime Functionmentioning

confidence: 99%

Cold-start vs. warm-start miss ratios

Easton

Fagin

1978

“…A bounded locality interval is the 2-tuple consisting of an activity set membership and its lifetime. Denning [6] has pointed out that these ideas can be reformulated in terms of the LRU distance string. We now present a mechanism for recognizing activity sets and a more formal definition of a bounded locality interval.…”

Section: Bounded Locality Intervalsmentioning

confidence: 99%

“…The vast majority of these programs were for scientific/engineering applications. The sample was selected with the following criteria: (a) the job must consume more than around one second of computational processor time [6], which is arithmetic/logical processor time less the time involved in virtual memory allocation/deallocation and all activities associated with input/output such as blocking, formatting, etc. ; (b) the job must have at least six allocated array segments at some period during its execution.…”

Section: Ti I= T' -Tat' -1)mentioning

confidence: 99%

Characteristics of program localities

Madison

Batson

1976

There are many important correctness properties for parallel programs besides the ones treated here; priority assignments, progress for each process, blocking of some subset of the processes in a program, etc. Many of these properties are dilVicult to define in a uniform way, while others require a language in which there are definite rules for scheduling competing processes. We are working to broaden the range of properties which can be proved with axiomatic methods.The proof techniques we have discussed can be profitably applied at three levels. First, they provide a sound basis for formal proofs of program correctness. Although formal proofs are generally too long to be reasonably done by hand, the axiomatic method would be well suited for an interactive program verifier, in which the programmer provides the resource invariants and some of the pre and post assertions, and the program verifier checks that these satisfy the axioms.A second possibility is informal proofs, like the ones given in this paper. The techniques are easy to use, and are relatively reliable. Although mistakes are possible in any informal proof, the structure of the axioms reduces the probability of error. Once the programmer has defined his resource invariants, the reasoning involved in the proofs is strictly sequential, and thus easy to do. In contrast, many informal proofs involve arguments about the order in which statements can be executed-in these it is dangerously easy to overlook the one case in which the program performs incorrectly.Finally, the language and the axioms give guides for the construction of correct and comprehensible programs. The use of resources isolates the areas in which programs can interfere with each other, and the resource invariant states explicitly what each process can assume about the variables it shares with other processes. The programmer who takes the time to define a resource invariant and check that it is preserved in each critical section is using a valuable tool for producing correct programs. References 1. Brinch Hansen, P. Concurrent programming concepts.

“…The GWS models all one-parameter memory policies whose resident sets satisfy an inclusion property under increasing values of the control parameter (G). The well-known "stack algorithms" [20,6] and "time window working set" [9,14,6] are special cases of the GWS. This model extends the measurement technique to segment referencing, and it unifies previous models as well.…”

mentioning

confidence: 99%

Generalized working sets for segment reference strings

Denning

Slutz

1978

Self Cite

Tne working set concept is extended for programs that reference segments of different sizes. The generalized working set policy (GWS) keeps as its resident set those segments whose retention costs do not exceed their retrieval costs. The GWS is a model for the entire class of demand-fetching memory policies that satisfy a resident set inclusion property. A generalized optimal policy (GOPT) is also defined; at its operating points it minimizes aggregated retention and swapping costs.Special cases of the cost structure allow GWS and GOPT to simulate any known stack algorithm, the working set, and VMIN. Efficient procedures for computing demand curves showing swapping load as a function of memory usage are developed for GWS and GOPT policies. Empirical data from an actual system are included.