Strongly Competitive Algorithms for Caching with Pipelined Prefetching

Strongly competitive algorithms for caching with pipelined prefetching

“…However, by Claim 2.2, we can modify A so that it satisfies the "no harm" rules at any time. (As shown in the proof of Claim 2.2 [11], if A violates a "no harm" rule for the first time at t > i, then we modify the steps of the algorithm at time t or later, therefore A still acts like AGG in the first i steps.) 2…”

“…The case where G is a branch tree is of particular interest in the application of CLPP to pipeline execution of programs on fast processors (see in [11]). For this case, we show that the bound in Theorem 4.1 can be further improved.…”

“…Branch trees model program execution on fast processors (see [11]), in which case, the cache size, k, is the length of the pipe. Since each instruction goes through k − 1 stages in the pipe before it reaches the execution phase, in our discussion of branch trees we assume that d = k − 1.…”

“…The total execution time of σ is 2 A detailed survey of these applications is given in [11]. the time to access the m blocks plus the total stall time, i.e., time(A, σ ) = m + m i=1 d A (r i ).…”

Integrated prefetching and caching in single and parallel disk systems

“…The goal is to design strategies that coordinate prefetching and caching decisions. Both theoretical and experimental studies were presented [5,6,9,10,11,12,13,14,16,17]. It was demonstrated that an integration of prefetching and caching leads to a substantial improvement in systems performance.…”