home *** CD-ROM | disk | FTP | other *** search
- Path: andrew.cmu.edu!bo24+
- From: Berend Ozceri <bo24+@andrew.cmu.edu>
- Newsgroups: comp.sys.amiga.hardware
- Subject: Re: EDO RAM retrofit to RETINA Z-III
- Date: Mon, 5 Feb 1996 00:34:06 -0500
- Organization: Masters student, Electrical and Computer Engineering, Carnegie Mellon, Pittsburgh, PA
- Message-ID: <sl5NPCG00iWY0183si@andrew.cmu.edu>
- References: <jboros.40.002789D1@indirect.com> <Al5K_Zu00iWY4184wM@andrew.cmu.edu>
- <4f3sc6$t7e@interport.net>
- NNTP-Posting-Host: po9.andrew.cmu.edu
- In-Reply-To: <4f3sc6$t7e@interport.net>
-
- Excerpts from netnews.comp.sys.amiga.hardware: 4-Feb-96 Re: EDO RAM
- retrofit to RET.. by House Mouse@interport.ne
- > >DRAM systems not designed with EDO in mind, cannot take... [snip]
- >
- > Will he get a speed increase if installs 60ns Zips on the card? Can
- > the card take advantage of Static column Zips? Just a suggestion if he really
- > wants to try to speed up the card. :)
-
- No.
-
- Traditional DRAM systems are more often than not synchronous systems.
- They do not base their transactions on handshaking. Rather, they use
- signals that have fixed phase relationships to free running clocks, so
- making one part faster without the other parts knowing won't get any
- performance increases.
-
- In the case of 60 ns vs 80 ns RAM, there has to be support in the Retina
- hardware to perform a clock speed change (in the form of a DRAM speed
- jumper perhaps) because there is no way the RAM can tell the controller
- that the data is now available 20 ns earlier. The controller just knows
- to blindly grab the data 80 ns after it supplies the column address to
- the RAM. That's why putting a 100 ns part would break the system; the
- data would not be available at the time expected by the controller.
-
- As far as the static column issue goes, it is once again a moot issue if
- the controller on the Retina doesn't support it (which I believe it
- doesn't).
-
- Berend Ozceri
- Carnegie Mellon University
-
-