Christopher B. Browne's Home Page
cbbrowne@acm.org

4. Memory Fragmentation

There have, in a similar vein, been discussions suggesting that it would be "highly valuable" to create something like the Windows/Mac "RAM Doubler" utilities.

These utilities come with two components:

All this being said, there has been some formal research work into this; the ACM "Operating Systems" SIG had a paper on this topic in 1997, with experimental results under Digital UNIX using several compression algorithms.

The findings were that they could provide overall performance improvements on the order of 10 available by using a variant on Huffman encoding, keeping on disk swap data compressed, with an integration of swap and disk cacheing.

If the recent trend where CPUs are much faster than dynamic RAM, and enormously faster than access to secondary storage continues, it may become worthwhile to use this sort of approach in production systems.

Google
Contact me at cbbrowne@acm.org