home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!cs.utexas.edu!qt.cs.utexas.edu!yale.edu!ira.uka.de!smurf.sub.org!nadia!delos!ananke!ak
- From: ak@ananke.stgt.sub.org (Andreas Kaiser)
- Newsgroups: comp.arch
- Subject: 64-bit CPU vs 2 x 32-bit CPUs
- Distribution: world
- Message-ID: <712194543snx@ananke.stgt.sub.org>
- References: <l7133mINN75s@spim.mips.com>
- Date: Sun, 26 Jul 92 23:49:03 GMT
- Organization: ananke
- Lines: 25
-
-
- In article <l7133mINN75s@spim.mips.com> mash@mips.com writes:
-
- > 3) The main reason is to get more address sapce (conveniently).
- > There are not a *huge* number of these things; however, the ones that
- > are there are *extremely* important to the people who use them, as they
- > are things like:
-
- Large applications are one reason, shared libraries (DLLs) and shared
- memory mappings are another. Implementing shared libraries or shared
- memory can enforce a global allocation mechanism, with all DLLs and all
- shared memory areas in the system having system-wide unique virtual
- addresses. Otherwise at least shared libraries could not be shared
- (relocation) with all processes. And this limitation can be exceeded
- quite soon (within some years) in a pure 32-bit virtual address range.
-
- And there are memory mapped files as filesystem implementation
- technique. You clearly need more than 32-bit addressed virtual memory
- to implement memory mapped files. IBM added some kind of segmentation
- to the RS/6000, but 64-bit addressed virtual memory is easier to
- handle.
-
- :::::::::::::::::::: subnet: ak@ananke.stgt.sub.org
- :: Andreas Kaiser :: fidonet: 2:241/7220.9
- ::::::::::::::::::::
-