[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: SRAM on ZipGSX - how fast for 14MHz?
- Subject: Re: SRAM on ZipGSX - how fast for 14MHz?
- From: mjmahon@aol.com (Michael J. Mahon)
- Date: 28 Jul 2001 08:51:27 GMT
- Newsgroups: comp.sys.apple2
- Organization: AOL http://www.aol.com
- References: <9jsksu$2bd$1@slb6.atl.mindspring.net>
- Xref: archiver1.google.com comp.sys.apple2:11602
Rodney Hester wrote:
>I've found that putting the 45ns wide SRAM chip I have in the data socket
>(and switching the jumpers to 32K cache) does indeed allow me to run
>somewhat reliably at 13.75MHz, but the system will eventually lock up or
>crash. From what I've read, 30ns SRAM is pretty much required in the
>worst-case scenario, which is why I thought 20ns would be perfect.
>
>It appears that brand matters at least as much as speed. If anyone has a
>known, reliable source for reasonably fast SRAM that can be _tested good_ on
>a Zip GS at 14MHz, I'd like to talk to them about purchasing some data RAM.
>I know a gentleman that worked for the company used to provide this service
>(David, I believe it was?), but I can only assume that is no more.
There is a fundamental problem in trying to generalize from anecdotal
evidence about SRAM or DRAM speeds. As another poster noted recently,
the speed rating on memory chips (like other chips) is only a _maximum_
access time. So a 20ns-rated chip may actually be 14ns, or 12ns, or faster.
This is a result of the manufacturer improving the chip design and process
to get better speed, then not having enough slow parts to make the
slower speed ratings meaningful. It is almost impossible to get memory
chips spec'ed for a _minimum_ access time.
As a result, when person X says that Y worked with chip speed Z, it is
not possible to conclude very much about what might work for you.
If you can specify manufacturer and date of manufacture, in addition to
speed, then there is a little more chance of correlation, but not much!
The only way to find out what is really going on is to look at the signals
with a fast scope and see what is working/not working in terms of setup
and hold times. Even then, generalizations are difficult if there is much
variablility in the speed of other chips on the board, particularly the PAL(s)
and the processor.
-michael
Email: mjmahon@aol.com
Home page: http://members.aol.com/MJMahon/