home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!usc!news!proton!hope.llumc.edu!borodin
- From: borodin@hope.llumc.edu (Sergi Borodin)
- Newsgroups: comp.sys.ibm.pc.hardware
- Subject: Re: Memory Cache questions
- Keywords: Cache
- Message-ID: <1992Sep10.173800.26260@proton.llumc.edu>
- Date: 10 Sep 92 17:38:00 GMT
- References: <1992Sep7.193149.23396@newstand.syr.edu> <1992Sep10.003938.9490@sarah.albany.edu>
- Sender: root@proton.llumc.edu (Operator)
- Organization: LLUMC Radiation Research Lab
- Lines: 26
-
- In article <1992Sep10.003938.9490@sarah.albany.edu> jjo619@thor.albany.edu (John O'neill) writes:
- >
- >I was wondering if you could use regular (non static) ram for a
- >memory cache.
-
- Sure you can.
-
- > I realize that static memory is used because it remembers
- >when the power goes off, but after using the computer the cache should
- >be filled with instuctions. It seems to me that the static memory
- >is only better than regular memory at start up time. The reason for
- >all this is that static ram is much more expensive than regular ram.
- >I do realize that cache ram is fast and so is more expensive but I believe
- >that regular (ie non statice) would be cheaper.
-
- Static memory loses its contents when power goes off. The reason to use
- static memory in cache is the access time, which can be as low as 10 ns
- Access time for dynamic RAM doesn't go below 60 ns. So if you're satisfied
- with 60 ns - stick with it.
- For 50 MHz clock rate you'd need a memory with the access time better
- than 20 ns.
-
- >Is this possible ? Thanks ,
- > John
-
- S.B.
-