The art of computer programming, volume 1 (3rd ed.): fundamental algorithms
The art of computer programming, volume 1 (3rd ed.): fundamental algorithms
A Fixed Optimum Cell-Size for Records of Various Lengths
Journal of the ACM (JACM)
Statistical Properties of the Buddy System
Journal of the ACM (JACM)
ACM Computing Surveys (CSUR)
A comparison of next-fit, first-fit, and best-fit
Communications of the ACM
On the external storage fragmentation produced by first-fit and best-fit allocation strategies
Communications of the ACM
A weighted buddy method for dynamic storage allocation
Communications of the ACM
Minimizing wasted space in partitioned segmentation
Communications of the ACM
A class of dynamic memory allocation algorithms
Communications of the ACM
A note on storage fragmentation and program segmentation
Communications of the ACM
Experience in automatic storage allocation
Communications of the ACM
Dynamic storage allocation techniques.
Dynamic storage allocation techniques.
Efficient implementation of the first-fit strategy for dynamic storage allocation
ACM Transactions on Programming Languages and Systems (TOPLAS)
Thermodynamics and garbage collection
ACM SIGPLAN Notices
Memory occupancy patterns in garbage collection systems
Communications of the ACM
A dynamic storage allocation technique based on memory residence time
Communications of the ACM
An algorithm with constant execution time for dynamic storage allocation
RTCSA '95 Proceedings of the 2nd International Workshop on Real-Time Computing Systems and Applications
| Hi-index | 48.25 |
This paper reports simulation data showing that, in dynamic memory allocation, the average free-to-allocated-block ratio can differ considerably and in both directions from the predictions of the 50 percent rule. A new derivation is given, and it is shown that previous derivations make an assumption that may be violated frequently. On the basis of the simulation data and the derivation, it is hypothesized that the anomalous behavior results from the combined effects of systematic placement and the statistics of the release process. Additional simulations support this hypothesis. Systematic placement, which refers to the natural convention of always allocating storage requests against the same end of the free block selected by the allocation strategy, tends to order blocks within contiguous groups according to their allocation time. The degree of anomalous behavior depends on the extent to which allocated blocks are released in the order of their allocation. For non-Markovian release processes, the extent of the correlation between allocation order and release order varies approximately inversely with the coefficient of variation of the memory residence time distribution. The simulations show that allocation efficiency depends strongly on the residence time distribution; efficiency decreases as the distribution's coefficient of variation increases. Some practical implications are briefly discussed.