Peter Shenkin wrote:
>
> If you still disagree, I suggest we take the discussion off-line.
>
I'm interested in the discussion, and I guess I'm not the only one...
Jack Scheible said:
> Since, on a 32-bit machine, each PROCESS has 4 gigabytes of "address space,"
> and some of us are running 64-bit machines, we need not worry about
> the fragmentation of the address space.
Some people have to...
1) only a few machines have such large (4Gb) physical memory space
2) our machines are shared by a lot of people and processes: one cannot
assume that he has all the memory for himself alone
3) if you waste memory up to the swap space, it works but slows down
all the system
4) In an industrial environment, executions _must_ be safe: no crash
and constant performances.
Once again, I agree with Peter Shenkin to say that the OS can only
decide how to map (and modify the mapping if needed) the virtual memory
of a process into the physical memory. But the organisation of the
virtual memory is under the responsability of the process.
In my previous example what was represented was the virtual memory
of a process:
let's say that the virtual memory is 32 words long, and that the OS
manages blocks of 8 words of physical memory:
(0)..bb....(1)ccc...aa(2)aaaaaadd(3)ddd.....
It means that the OS takes blocks of 8 words of the virtual
memory to distribute them in the physical space, that's all.
For example:
RAM: DISK(swap):
(0) (1) (2) (3) (4) (1)
(0)..bb....--------(3)ddd.....(1)ccc...aa-------- (2)aaaaaadd
^^^^^^^^
(another process)
block 0 of Virtual Mem -> block 0 of RAM
1 -> 3
2 -> 1 of swap space
3 -> 2 of RAM
block 1 of RAM is attributed to another process. When this latter ends
the OS will be able to move block 2 of VM from DISK to block 1 of RAM.
But it changes noting to the fragmentation of virtual memory...
--
Pierre Hugonnet Seismic Data Processing R&D
COMPAGNIE GENERALE DE GEOPHYSIQUE
mail: 1 rue Leon Migaux phone.....(33) 164 47 45 59
91302 MASSY fax.......(33) 164 47 32 49
FRANCE [log in to unmask]
My opinions are not necessarily those of CGG (nor those of JPII)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|