3

I am running a code that mimics demand paging system here are some of the parameters:

Page size = 4096 bits
Processor = 32 bits
Number of page frames = (variable)

I run my code with a fifo or random page replacement algorithm with number of page frames set to 100. My disk read/ write ends at 63.

Then when I changed the Processor size to 32 bits - this boundary moves to 16.

The latter case, I understand. If the page frame count is more than the possible page table entries, (2^4) there is no need for evictions.

But I don't understand why in the former case, where the page table entries go up to 2^20, the disk read/write stops at 63 with both page replacement algorithms.

What is this mystery 63 number? Any idea where this may be coming from? Thank you.

moejoe
  • 145
  • 2
  • 11
  • 2
    If you counting disk I/O as simulated by code you run, why not assume it's just a bug in that code? If you counting actual OS-level I/O -- what is your OS? – Victor Sorokin Apr 15 '11 at 19:37
  • @Victor I am counting the code level I/O. I understand there is a bug in the code. I am trying to figure out where it may be coming from - logic or syntax. First, I am trying to rule out the logical flaw. – moejoe Apr 15 '11 at 19:49
  • 2
    the only interesting thing I can say about 63 here -- it's 2^6 - 1 :) – Victor Sorokin Apr 15 '11 at 19:51

1 Answers1

0

The reference string had only 63 pages -- now that I have gotten some sleep...

moejoe
  • 145
  • 2
  • 11