Dave Haynie... A4k 8megs simms
Daniel Pfund
>
> Dave:
>
> I have read in Amiga Format that it is poosible to replace the 4 meg simms in an
> A4000 with 8 megs simms and it will give you 32 megs of fast ram without any problems.
>
> Is this true? Thanks...
Yes, you can use 8Mb SINGLE-SIDED simms, but only 2, which gives 16Mb max
still
>
>
> Also I have a 3640 card rev 3.1 with the u209 xxx-02 is there anyway to fix the bug
> in this pal chip without puting a new one in, ( like a patch or something). Thanks...
>
>
>
> Cory Palmer
> cpa...@nando.net
>
>
>
>
> --
>
>
>
> Somebody Or Other Path to get to you (always begin with a
> Your Address Maybe fully domain'd machine).
>
> Suggest you use only '!'s in your path.
>
--
__
/// Daniel Pfund Email:<Pfu...@uni2a.unige.ch> AX25: HB9VBC@HB9IAP
__/// Amiga-Psion:a perfect match, multitasking power wherever you are!
\\X/ Psion Series3 FAQ at <http://www.bris.ac.uk/~lwmdcg/Psion/FAQ/>
Michael van Elst
>Dave:
> I have read in Amiga Format that it is poosible to replace the 4 meg simms in an
>A4000 with 8 megs simms and it will give you 32 megs of fast ram without any problems.
>Is this true? Thanks...
No, it is not true. You can replace 4 4Meg SIMMs with 2 8Meg SIMMs for
a total of (again) 16MBytes.
Regards,
--
Michael van Elst
Internet: mle...@serpens.rhein.de
"A potential Snark may lurk in every tree."
Paul Chan
>cpa...@parsifal.nando.net writes:
>
>
>>Dave:
>
>> I have read in Amiga Format that it is poosible to replace the 4 meg simms in an
>>A4000 with 8 megs simms and it will give you 32 megs of fast ram without any problems.
>
>>Is this true? Thanks...
>
>No, it is not true. You can replace 4 4Meg SIMMs with 2 8Meg SIMMs for
>a total of (again) 16MBytes.
Sorry to add this, but what about putting in one 16Mb SIMM?
And what speed would you recommend the SIMM to run atif I wanted to later
upgrade to an 060?
Dave Haynie
> I have read in Amiga Format that it is poosible to replace the 4 meg simms in an
>A4000 with 8 megs simms and it will give you 32 megs of fast ram without any problems.
>
>Is this true? Thanks...
Nope. The A4000 can handle four 4MB banks of RAM. This can be in four
4MB SIMMs or two 8MB SIMMs. That's a limitation of the RAMSEY chip.
I have been thinking it's possible to build a hack to extend this to
four banks of 16MB, but I don't currently have an A4000 to experiment
on. This can't be done on an A3000; the RAMSEY chip is used as the
address generator for the DMAC, and any hacking around with RAMSEY
address lines would foul that operation up.
>Also I have a 3640 card rev 3.1 with the u209 xxx-02 is there anyway
>to fix the bug in this pal chip without puting a new one in, ( like a
>patch or something). Thanks...
You can reprogram that chip, but no, you can't fix hardware bugs in
software. On the other hand, the bug in U209 R2 is fairly innocuous;
unless you have a PhonePak or some smiliar realtime device, you won't
notice it. We couldn't even measure its effect on A4091 performance
(which actually surprised me, I though it would have been noticable).
Dave Haynie | ex-Commodore Engineering | for DiskSalv 3 &
Sr. Systems Engineer | Hardwired Media Company | "The Deathbed Vigil"
Scala Inc., US R&D | Ki No Kawa Aikido | in...@iam.com
"Feeling ... Pretty ... Psyched" -R.E.M.
Dave Haynie
Daniel Pfund
> In <48q028$s...@castle.nando.net>, cpa...@parsifal.nando.net writes:
>
>> I have read in Amiga Format that it is poosible to replace the 4 meg simms in an
>>A4000 with 8 megs simms and it will give you 32 megs of fast ram without any problems.
>>
>>Is this true? Thanks...
>
> Nope. The A4000 can handle four 4MB banks of RAM. This can be in four
> 4MB SIMMs or two 8MB SIMMs. That's a limitation of the RAMSEY chip.
>
> I have been thinking it's possible to build a hack to extend this to
> four banks of 16MB, but I don't currently have an A4000 to experiment
> on. This can't be done on an A3000; the RAMSEY chip is used as the
> address generator for the DMAC, and any hacking around with RAMSEY
> address lines would foul that operation up.
Ok, that would be great, but have you thought about something:
the space problem.... I already had troubles finding a SINGLE-SIDED-CHIPS
8Mb simm, now how can anyone find a 16Mb single-sided ???? I don't even
now if such a thing exists!
Regards,
Daniel
Dave Haynie
>mle...@serpens.rhein.de (Michael van Elst) wrote:
>>> I have read in Amiga Format that it is poosible to replace the 4 meg simms in an
>>>A4000 with 8 megs simms and it will give you 32 megs of fast ram without any problems.
>>No, it is not true. You can replace 4 4Meg SIMMs with 2 8Meg SIMMs for
>>a total of (again) 16MBytes.
>Sorry to add this, but what about putting in one 16Mb SIMM?
Nope, that won't work. The RAMSEY chip can handle four 4MB banks of
memory. Since normal 4MB SIMMs have a single 4MB bank, and 8MB SIMMs
actually have two 8MB banks, this works out cleanly. The 16MB SIMMs
have a single 16MB bank, which would be seen as a single 4MB bank by
the RAMSEY chip.
I've been thinking for awhile that it might be possible to hack the
memory system for four 16MB banks. Basically, add a fast PAL around
RAMSEY to create one extra DRAM addess line and spoof the address
inputs, so RAMSEY doesn't know it's really addressing up to
64MB. Unfortunately, this can't work on an A3000 (since RAMSEY on an
A3000 drives addresses out during a disk DMA cycle), and I don't have
an A4000, so I haven't looked into it.
>And what speed would you recommend the SIMM to run atif I wanted to later
>upgrade to an 060?
The A4000 can take advantage of a 60ns part (you need to set the
"skip" bit in RAMSEY, this is only in the Rev 7 part used on the
A4000, not in the A3000's Rev 4 part). PD hacks like SetRAMSEY will do
this for you. It's likely 60ns parts will improve performance on '060
boards too.
Michael van Elst
>>No, it is not true. You can replace 4 4Meg SIMMs with 2 8Meg SIMMs for
>>a total of (again) 16MBytes.
>Sorry to add this, but what about putting in one 16Mb SIMM?
I don't know but I rather doubt this.
>And what speed would you recommend the SIMM to run atif I wanted to later
>upgrade to an 060?
The speed is independent of the CPU. The memory is controlled by the
same Ramsey chip, so 80ns SIMMs are sufficient and faster SIMMs won't
do anything better.
Oliver Kastl
>> Sorry to add this, but what about putting in one 16Mb SIMM? <<
I tried this, and it didn't work.
--
Oliver Kastl, Elaborate Bytes
Home is where the heart is
Juhani Polkko,Naantali -
In a previous article, mle...@serpens.rhein.de (Michael van Elst) says:
>The speed is independent of the CPU. The memory is controlled by the
>same Ramsey chip, so 80ns SIMMs are sufficient and faster SIMMs won't
>do anything better.
>
Except if you use "SpeedRamsey" option from eg. MCP or
SystemPrefs, it gave me a nice 10-15% speed-up (A4EC030)
It works fine with some standard 70ns SIMMs, but not with
all of them.
--
William F. Maddock
>Sr. Systems Engineer | Hardwired Media Company | "The Deathbed Vigil"
>Scala Inc., US R&D | Ki No Kawa Aikido | in...@iam.com
> "Feeling ... Pretty ... Psyched" -R.E.M.
Thanks for posting at least the basics of your proposed hack.
Now then, are there any enterprising and brave souls out there that are
willing to try this out on their A4000's. Don't look at me, I've got a 1200
and hands that have been known to curdle milk just by holding the glass. :(
Anyway, thanks Dave. You may have furthered the development of the 4000 just
by posting a message.
cya,
Bill.
Paul Kolenbrander
> I've been thinking for awhile that it might be possible to hack the
> memory system for four 16MB banks. Basically, add a fast PAL around
> RAMSEY to create one extra DRAM addess line and spoof the address
> inputs, so RAMSEY doesn't know it's really addressing up to
> 64MB. Unfortunately, this can't work on an A3000 (since RAMSEY on an
> A3000 drives addresses out during a disk DMA cycle), and I don't have
> an A4000, so I haven't looked into it.
Aargh!!! Dr. Kittel, Gilles, Cristoph? Any chance of loaning Dave an
A4000? It would be benificial for AT as well if the new AT A4000's
where able to handle 64MB Fast Mem on the motherboard. :-)
Cya,
--__
/_/ |/ E-mail: pa...@serena.iaehv.nl
/aul |\olenbrander WWW : http://www.iaehv.nl/users/paul/
-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-
Michael van Elst
> Except if you use "SpeedRamsey" option from eg. MCP or
> SystemPrefs, it gave me a nice 10-15% speed-up (A4EC030)
> It works fine with some standard 70ns SIMMs, but not with
> all of them.
With 60ns SIMMs this is just at the edge of all tolerances, even
then it most often fails if you use more than 2 SIMMs.
With 70ns SIMMs you must be pretty lucky. That's as reliable as
overclocking your system.
Rob
> the space problem.... I already had troubles finding a SINGLE-SIDED-CHIPS
> 8Mb simm, now how can anyone find a 16Mb single-sided ???? I don't even
> now if such a thing exists!
They exist. I recently ordered a 16 meg SIMM and didn't specify single-sided
but that's what came in the mail.
-Robert Dick (dic...@wckn.dorm.clarkson.edu)-
Geert Uytterhoeven
|> jpo...@mail.freenet.hut.fi (Juhani Polkko,Naantali -) writes:
|>
|> > Except if you use "SpeedRamsey" option from eg. MCP or
|> > SystemPrefs, it gave me a nice 10-15% speed-up (A4EC030)
|>
|> > It works fine with some standard 70ns SIMMs, but not with
|> > all of them.
|>
|> With 60ns SIMMs this is just at the edge of all tolerances, even
|> then it most often fails if you use more than 2 SIMMs.
|> With 70ns SIMMs you must be pretty lucky. That's as reliable as
|> overclocking your system.
I have one 80ns and two 70ns SIMMs, and SpeedRamsey works fine...
Note: my system is up 5 days (=120 hours) a week.
--
Geert Uytterhoeven Geert.Uyt...@cs.kuleuven.ac.be
Wavelets, Linux/m68k on Amiga http://www.cs.kuleuven.ac.be/~geert/
Department of Computer Science -- Katholieke Universiteit Leuven -- Belgium
Alex...
[discussion about putting 16MB SIMM's on an A4000 motherboard... snip]
>Ok, that would be great, but have you thought about something:
>the space problem.... I already had troubles finding a SINGLE-SIDED-CHIPS
>8Mb simm, now how can anyone find a 16Mb single-sided ???? I don't even
>now if such a thing exists!
Yup, they sure do - I have two such beasts in my A4000 (on a
WarpEngine I should add before anyone gets too excited *;o) I think
they should be fairly common, I didn't specify single sided SIMM's
when I bought mine, but that was what I got.
Hope this helps,
Alex...
--
\|/ Alexander Craig
(@ @) Parallel Systems Group Mail: A.C...@dcs.warwick.ac.uk
-oOO-(_)-OOo- University of Warwick http://www.dcs.warwick.ac.uk/~eezer/
Paul Kolenbrander
> > Aargh!!! Dr. Kittel, Gilles, Cristoph? Any chance of loaning Dave an
> > A4000? It would be benificial for AT as well if the new AT A4000's
> > where able to handle 64MB Fast Mem on the motherboard. :-)
>
> Are you sure it`s a good idea? Everybody who needs 64 MB of memory
> will use a Warpengine or Cyberstorm, too.
> Additionally the A4000T uses SCSI-DMA like the A3000, doesn`t it?
Sure it's a good idea. 16MB SIMM's are pretty value for money and
I expect to take them on to say a PowerAmiga. I fear the 4MB ones
that are in my Amiga 4000 now won't be enough for a PPC type of
machine. As I'd want to run NT and stuff as well. Again. check out
pricing. One 16MB SIMM is cheaper than 4 4MB ones and takes less
space. And yes, as far as I know the A4000T does DMA on it's SCSI.
If I remember correctly it borrows from the A4091 technology...
Daniel Pfund
> In article <1995Nov22.091139.1@ugun2a> pfu...@uni2a.unige.ch (Daniel Pfund) writes:
>
> [discussion about putting 16MB SIMM's on an A4000 motherboard... snip]
>
>>Ok, that would be great, but have you thought about something:
>>the space problem.... I already had troubles finding a SINGLE-SIDED-CHIPS
>>8Mb simm, now how can anyone find a 16Mb single-sided ???? I don't even
>>now if such a thing exists!
>
> Yup, they sure do - I have two such beasts in my A4000 (on a
> WarpEngine I should add before anyone gets too excited *;o) I think
> they should be fairly common, I didn't specify single sided SIMM's
> when I bought mine, but that was what I got.
Yup, I now know ;-)
I was thinking that it might not be possible, but several people have mailed
me about them. I originally thought so because I have so much trouble
finding 8Mb single-sided ones (do also exist... but very rare apparently!)
Daniel P.
Clayton Vandiver, Jr
I need at least 8 megs in whatever configuration available; a single 8
meg SIMM would be great!
Thanks for any help!
Clayton
van...@ix.netcom.com
Matthias Meixner
: Anyone got a handle on where I could find some 64 pin SIMMS???
MTec should have them, since they have bought GVP.
--
- Matthias Meixner
-----------------------------------------------------------------------------
EMail: mei...@rbg.informatik.th-darmstadt.de
WWW: http://www.student.informatik.th-darmstadt.de/~meixner/
Uebrigens: Nachts gehen Sonnenuhren nach dem Mond ...
Thomas Tavoly
In article <1995Nov22.1...@scala.scala.com> dave....@scala.com (Dave Haynie) writes:
>
> The A4000 can take advantage of a 60ns part (you need to set the
> "skip" bit in RAMSEY, this is only in the Rev 7 part used on the
> A4000, not in the A3000's Rev 4 part). PD hacks like SetRAMSEY will do
> this for you. It's likely 60ns parts will improve performance on '060
> boards too.
The utility is called SpeedRamsey and can be found on Aminet:
SpeedRamsey12a.lha hard/misc 7K +Sets skip mode in Rev G Ramsey
I have one 60 ns bank and one 70 ns bank, the program works fine and my
system is now 1.03 times the speed of an A3000, compared to 0.90 without
it. Whether a 70 ns SIMM will work depends on how many SIMMs there are
(latency gets bigger with more SIMMs) and how good the manufacturer has
made them, actual speeds may be 60-65 ns or lower for a good brand while
close to or even above 70 ns for a 'bad' brand. I don't remember what
brands I use, but one of them is Siemens I think.
. Thomas Tavoly + A4000 = aTm...@amiga.ow.nl _ .
. WWW - http://www.cistron.nl/~ttavoly _ // .
. Finger - tta...@cistron.nl [.sig v4.0] \X/ .
... >>> Amiga - Back for the future! <<< ...
Dave Haynie
>In article <1995Nov21....@scala.scala.com>, dave....@scala.com (Dave Haynie) writes:
>>> I have read in Amiga Format that it is poosible to replace the 4 meg simms in an
>>>A4000 with 8 megs simms and it will give you 32 megs of fast ram without any problems.
>> Nope. The A4000 can handle four 4MB banks of RAM. This can be in four
>> 4MB SIMMs or two 8MB SIMMs. That's a limitation of the RAMSEY chip.
>> I have been thinking it's possible to build a hack to extend this to
>> four banks of 16MB,
>Ok, that would be great, but have you thought about something:
>the space problem.... I already had troubles finding a SINGLE-SIDED-CHIPS
>8Mb simm, now how can anyone find a 16Mb single-sided ???? I don't even
>now if such a thing exists!
Actually, 16MB single-sided SIMMs are probably more common than 8MB
single-sided SIMMs. You have to understand how 72-pin SIMMs are
organized to realize why this is true. DRAM addresses are naturally
supplied in even-power-of-two sizes (1MB, 4MB, 16MB), simply because
the address that goes into a DRAM is multiplexed. The effective
address for, say, a DRAM with 10 address pins is 20-bits.
In order to get odd-power-of-two SIMMs, there is a second "bank" of
DRAM on the module. So basically, an 8MB SIMM has the same chips on it
as two 4MB SIMMs. That's the reason you can use two of these in the
A4000. Typical 4MB SIMMs use eight 1MB x 4 DRAM chips, typical 8MB
SIMMs use sixteen 1MB x 4 DRAM chips, eight on each side. Old 16MB
SIMMs require 32 of these 1MB x 4 DRAMs, and they're quite
large. Modern versions, however, use 16Mbit-density DRAMs, probably
eight 4MB x 4 parts. So they're the same basic size as 4MB SIMMs.
Technically, you could build 4MB and 8MB SIMMs using 16Mbit-density
parts in the 1MB x 16 configuration (two chips for a 4MB part, four
for an 8MB part), but these don't seem to have caught on yet.
Dave Haynie
>In article <495bf3$f...@serpens.rhein.de>, mle...@serpens.rhein.de (Michael van Elst) writes:
>|> jpo...@mail.freenet.hut.fi (Juhani Polkko,Naantali -) writes:
>|>
>|> > Except if you use "SpeedRamsey" option from eg. MCP or
>|> > SystemPrefs, it gave me a nice 10-15% speed-up (A4EC030)
>|>
>|> > It works fine with some standard 70ns SIMMs, but not with
>|> > all of them.
>|> With 60ns SIMMs this is just at the edge of all tolerances, even
>|> then it most often fails if you use more than 2 SIMMs.
>|> With 70ns SIMMs you must be pretty lucky. That's as reliable as
>|> overclocking your system.
>I have one 80ns and two 70ns SIMMs, and SpeedRamsey works fine...
Here's what's happening in your system. The RAMSEY chip is normally
running a five clock cycle. That's roughly three clocks (120ns) for
access, two clocks (80ns) for a precharge. It's a bit more complicated
than that, but the design will reliably deliver at least 100ns for RAS
access to the DRAM, probably a bit more.
When you set the "skip" bit in RAMSEY, you cut off one of the 40ns
clock cycles in the access phase. So the RAMSEY chip will reliably
deliver at least 60ns for RAS access to the DRAM, probably a bit
more.
A 60ns part will always work with a RAS access of exactly 60ns (as
long as its other specs are met; RAS access is usually one of the
gating factors). A 70ns or 80ns part COULD also always work with a RAS
access of 60ns, but it might not; it's only guaranteed for 70ns or
80ns, respectively. Certainly some 70ns or 80ns parts could have
passed a 60ns test, but the manufacturer didn't need any more 60ns
parts from that run. Equally certain is that some 70ns or 80ns parts
are rejects from a 60ns qualification test. They may work at 62ns, or
they may not work until 70ns or 80ns.
The tricky part is a little thing we like to call simultaneous worst
case. Basically, if everything in a system goes wrong at once, a
system designed for simultaneous worst case still works. If everything
goes wrong in an A4000 at once, a 70ns or 80ns part (failing at 69ns
or 79ns, respectively, since I'm taking worst case) will
fail. However, the only way to get a real simultaneous worst case
condition is to set your room temperature for 40C and fill your Amiga
with its limit on heat-generating cards (about 10W per Zorro
board).
The important part to realize from this discussion is that a 70ns or
80ns part doesn't have to hit an actual worst case situation to fail,
while a 60ns part will work in worst case. A hot summer day could
cause your machine to flake out, or maybe just drop one bit in a
billion. You don't want that.