home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.next.hardware
- Path: sparky!uunet!cs.utexas.edu!qt.cs.utexas.edu!yale.edu!cs.yale.edu!news
- From: nathan@laplace.biology.yale.edu (Nathan F. Janette)
- Subject: Re: Upgrade to NextTurbo (memory)
- Message-ID: <1992Aug17.165515.12984@cs.yale.edu>
- Sender: news@cs.yale.edu (Usenet News)
- Nntp-Posting-Host: jacobi.biology.yale.edu
- Reply-To: nathan@laplace.biology.yale.edu
- Organization: Yale University, Department of Computer Science, New Haven, CT
- References: <dillon.0lj8@overload.Berkeley.CA.US>
- Date: Mon, 17 Aug 1992 16:55:15 GMT
- Lines: 51
-
- In article <dillon.0lj8@overload.Berkeley.CA.US> dillon@overload.Berkeley.CA.US
- (Matthew Dillon) writes:
- > >NeXTcube Turbo (68040-33MHz), NeXTstation Turbo
- > >(68040-33MHz), NeXTstation Color Turbo
- > >(68040-33MHz), newer NeXTstations (68040-25MHz), and
- > >newer NeXTstation Color (68040-25MHz):
- > >
- > >These systems have only 4 SIMM slots, but they can take
- > >double-sided SIMMs. SIMMs must be installed in groups of
- > >two in these machines.
- > >
- > >Use 72-pin 4, 8, 16, or 32 MB SIMMs with 100 ns minimum
- > >access time. For maximum performance use 70 ns SIMMs.
-
- > Yes, that's what I read.... but it is extremely obscure.
- > It doesn't say WHY using faster SIMMs results in faster
- > performance, it doesn't say where the break point is...
- > you can't even GET 100ns SIMMs anymore!! They START at
- > 80ns (100ns cost the same or more then 80ns). The NeXT is
- > the only machine I've ever heard of that has a
- > programmable waitstate for it's dynamic ram access.
-
- In case you missed the followup from John Graves at
- NeXT Hardware Engineering:
-
- The memory system has programmable memory timing such
- that the number of processor clocks needed to access a
- given amount of data can be tailored to the speed of the
- memory installed. 70 ns memory is just enough faster than
- 80 ns memory to allow the cpu to access the data with fewer
- clock cycles. This improves memory system performance.
-
- "70 ns" memory is faster than "80 ns" memory in many
- parameters other than just RAS access time. The faster
- CAS access time in particular allows the memory system to
- respond quicker to burst (16 bytes) bus transfers.
-
- I'll add the above to the FAQ.
-
- > It really makes no sense to take an already slow machine
- ^^^^^^^^^^^^^^^^^^^
- IYHO, of course, which is not shared by me, and has no place
- in this fact-based (only) group. Take it to csn.advocacy
- so the folks there can have a good laugh.
-
- --
- Nathan Janette "I'm a NeXTstep man,
- Dept MB&B, Yale Univ I'm a NeXTcube guy"
- New Haven, CT
- nathan@laplace.biology.yale.edu (NeXT)
-
-