Messages in this thread | | | From | Stefan Monnier <monnier+lists/linux/kernel/news/@TEQUILA.SYSTEMSZ.CS.YALE.EDU> | Subject | Re: New Feature Idea: Compress swap file | Date | 20 Feb 1998 11:55:27 -0500 |
| |
Rik van Riel <H.H.vanRiel@fys.ruu.nl> writes: > Sorry Jim and Adam, compressed swap has already been > implemented and tried. The results were: > - - more CPU spent on swapping --> slower system > - - more I/O spent on swapping because the I/O patterns > became more complex than normal swap --> slower system > > Compressed swap would only make sense when: > - - paging space is short > - - you are actually _swapping_ (suspending_) some > process to alleviate the above problem
Indeed, it's been tried and mostly failed. The only way I can see to make it successful is: 1 - compress fast 2 - compress a lot 3 - beware of mapping between incore addresses and compressed addresses.
1 and 2 require a non-generic compression algorithm: the compression code should know what is on the page and compress integers differently from floats or from pointers or strings...
3 requires a lot more flexibility in positioning objects: basically the VM should be allowed to move objects around and update pointers to them.
All in all, unrealizable within Unix, but probably doable in a (non-conservatively) GC'd world (since the GC already knows what is a pointer and what is not and is able to move objects at will). LISP machines, Smalltalk machines, or JavaOS would probably be good systems to try it on.
Stefan
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |