Disable motherboard RAM
Georges Heinesch
I have an A3000 with:
- 2 MB chip ram
- 16 MB motherboard ZIP static column ram
- 32 MB CyberStormk MKII EDO ram
Is it possible to disable the 16 MB motherboard ram without being
forced to pull all the ZIPs out of their socket?
--
Cu Georges Heinesch, Luxembourg
geo...@ibm.net - geo...@geocities.com
http://www.geocities.com/yosemite/2480
PGP 2.6.3i public key available on request and on public servers
... have a plan and make it work ...
Alexander Frank
Hi Georges !
>I have an A3000 with:
>- 2 MB chip ram
>- 16 MB motherboard ZIP static column ram
>- 32 MB CyberStormk MKII EDO ram
>Is it possible to disable the 16 MB motherboard ram without being
>forced to pull all the ZIPs out of their socket?
Why would you do that ?
The MKII RAM should e at the highest priority in the mem list, so the
onboard RAM will only be used if the requested memory by a program is
too large for the free area in MKII RAM, i.e. it's full, so there
should be no problem - performance wise !
If you want to change the priorities of the RAM areas there's a
program on AmiNet that will allow you to do that, don't remember the
name right now - something like SetRamPri or such...
If some of your ZIPs should be faulty it would be the best idea to
pull them out.
I don't know of any jumper or such to disable onboard FAST RAM !
Live long and prosper,
Alexander Frank // Amiga 500, Amiga 1000
Germany // Amiga 2000, Amiga 3000, Amiga 4000
\\ // Commodore CDTV, Amiga CD 32
alf...@infotip-rv.de \X/ = Member of Team AMIGA =
Mike Redrobe
Hi, Georges Heinesch ,
GH> I have an A3000 with:
GH> - 2 MB chip ram
GH> - 16 MB motherboard ZIP static column ram
GH> - 32 MB CyberStormk MKII EDO ram
GH> Is it possible to disable the 16 MB motherboard ram without being
GH> forced to pull all the ZIPs out of their socket?
a) make sure the motherboard ram a lower mempory priority (it should already,
otherwise get MemPri from aminet to do it)
b) AllocateABS all the motherboard fastram in the startup-sequence, but that
wastes all the ram ;-0
dunno if there's a program to do the latter, but its only a few lines of code to
write for yourself....
Mike
--
---------------------------------------------------------------------------
Mike Redrobe - mailto:mi...@redrobe.demon.co.uk MikeRR on #Amiga
http://www.redrobe.demon.co.uk
---------------------------------------------------------------------------
Ian Gledhill
Georges Heinesch wrote in message <460.7242...@ibm.net>...
>Is it possible to disable the 16 MB motherboard ram without being
>forced to pull all the ZIPs out of their socket?
Why on earth do you want to do that? :-)
The CS RAM should have a higher priority than the RAM on the MB, so it
won`t be slowing down the system until it actually runs out of the CS RAM.
AFAIK.. :-)
If it's not a higher priority, I'm sure there's a proggy on the Aminet to
change it.
In answer to your *question*, though.. ( :-) ) I don`t know. It should
be possible...
-=-
Ian Gledhill
Ia...@amiganet.org
A4000 Tower 060/50 PPC604e/200 CV64 32MB RAM 4xCD-ROM 4.3Gig UW SCSI etc
Joshua B. Helm
Georges Heinesch wrote:
> I have an A3000 with:
>
> - 2 MB chip ram
> - 16 MB motherboard ZIP static column ram
> - 32 MB CyberStormk MKII EDO ram
>
> Is it possible to disable the 16 MB motherboard ram without being
> forced to pull all the ZIPs out of their socket?
>
You could jsut pull out the FIRST ZIP in the FIRST bank. The question
is, why would you want to disable the motherboard RAM? If its because
you want the CS RAM to be used instead, why not just increase the
priority on the CS RAM (which should already be done), or decrease the
priority on the motherboard RAM?
> --
> Cu Georges Heinesch, Luxembourg
> geo...@ibm.net - geo...@geocities.com
> http://www.geocities.com/yosemite/2480
> PGP 2.6.3i public key available on request and on public servers
>
> ... have a plan and make it work ...
--
---
- Joshua B. Helm <jb...@esc.ie.lehigh.edu>
Lehigh Univ, Iacocca Inst, ESC Project Manager
Only AMiGA Makes It Possible - Back For The Future
[As luck would have it, these views are all mine]
Hans Guijt
Ian Gledhill (im...@nospam.aber.ac.uk) wrote:
>>Is it possible to disable the 16 MB motherboard ram without being
>>forced to pull all the ZIPs out of their socket?
You could simply allocate it.
>Why on earth do you want to do that? :-)
Perhaps he has some program that doesn't like this RAM...
Hans
Georges Heinesch
In a message of 31-Oct-97 11:33:34, Ian Gledhill wrote:
> Georges Heinesch wrote in message <460.7242...@ibm.net>...
>>Is it possible to disable the 16 MB motherboard ram without being
>>forced to pull all the ZIPs out of their socket?
> Why on earth do you want to do that? :-)
> The CS RAM should have a higher priority than the RAM on the MB, so it
> won`t be slowing down the system until it actually runs out of the CS RAM.
> AFAIK.. :-)
Yes, that's right!
> If it's not a higher priority, I'm sure there's a proggy on the Aminet to
> change it.
> In answer to your *question*, though.. ( :-) ) I don`t know. It should
> be possible...
The reason is, that there are apparently faulty CyberStorm MKII boards
around, where the RAM interface between the motherboard RAM and the
CyberStorm RAM is faulty. It seems that whenever the CS RAM is full
and the motherboard RAM has to be used, that erroneous data transfer
occurs between the CPU and the motherboard RAM.
BTW, I'd like to know if anybody heard anything about this problem!
Yesterday evening, I discovered for expample that the DIavolo SCSI
transfer buffer set to CHIP created 50% faulty files during the
backup. This was never the case before I installed the CS MKII. I'm
sure about that.
--
Cu Georges Heinesch, Luxembourg
geo...@ibm.net - geo...@geocities.com
http://www.geocities.com/yosemite/2480
PGP 2.6.3i public key available on request and on public servers
... if you can't beat them, join them !
Georges Heinesch
In a message of 31-Oct-97 21:29:10, Joshua B. Helm wrote:
> Georges Heinesch wrote:
>> I have an A3000 with:
>>
>> - 2 MB chip ram
>> - 16 MB motherboard ZIP static column ram
>> - 32 MB CyberStormk MKII EDO ram
>>
>> Is it possible to disable the 16 MB motherboard ram without being
>> forced to pull all the ZIPs out of their socket?
>>
> You could jsut pull out the FIRST ZIP in the FIRST bank. The question
> is, why would you want to disable the motherboard RAM?
see my answer to Ian Gledhill.
> If its because
> you want the CS RAM to be used instead, why not just increase the
> priority on the CS RAM (which should already be done), or decrease the
> priority on the motherboard RAM?
Is there no possiblilty to disable it with a program?
--
Cu Georges Heinesch, Luxembourg
geo...@ibm.net - geo...@geocities.com
http://www.geocities.com/yosemite/2480
PGP 2.6.3i public key available on request and on public servers
... N1, TO/GA, 80 KTS, V1, ROTATE, Postive climb, Gear up ...
Georges Heinesch
In a message of 01-Nov-97 12:53:55, Patrick Kursawe wrote:
> Georges Heinesch wrote:
>>The reason is, that there are apparently faulty CyberStorm MKII boards
>>around, where the RAM interface between the motherboard RAM and the
>>CyberStorm RAM is faulty. It seems that whenever the CS RAM is full
>>and the motherboard RAM has to be used, that erroneous data transfer
>>occurs between the CPU and the motherboard RAM.
> Yes, there are faulty boards around. One the p5 homepage there was (or is
> still?) a page called "Errata" which describes the problem.
No, that's a different problem. This errata refers to CyberSTorms,
which don't see their own FastRAM (the SIMMs on the CyberStorm are
not recognized). My problem is different. Apparently, there *are*
boards who have the same problem as I have (transfer problems with
random data errors!)
>>BTW, I'd like to know if anybody heard anything about this problem!
> I asked p5 support if a cooler was necessary for my card (040), the manual
> said yes, but there was none. The guy there told me that indeed it was not
> needed and, by the way, told my that my board was faulty and that I should
> let them fix it. Didn't do it because I have no fast-mem on the motherboard
> and most likely won't use SCSI with this board.
Hmmm ...
>>Yesterday evening, I discovered for expample that the DIavolo SCSI
>>transfer buffer set to CHIP created 50% faulty files during the
>>backup. This was never the case before I installed the CS MKII. I'm
>>sure about that.
> Quite likely that this is the board's fault. There is some DMA trouble with
> the faulty CS MkII.
Ok, if my problem is more or less the same like the one mentioned on
the p5 errata page, I'd be very surprised, since the SNs listed there
don't match with my SN.
Well, I'll check it out!
--
Cu Georges Heinesch, Luxembourg
geo...@ibm.net - geo...@geocities.com
http://www.geocities.com/yosemite/2480
PGP 2.6.3i public key available on request and on public servers
... have a plan and make it work ...
Georges Heinesch
In a message of 31-Oct-97 07:50:05, Alexander Frank wrote:
>>I have an A3000 with:
>>- 2 MB chip ram
>>- 16 MB motherboard ZIP static column ram
>>- 32 MB CyberStormk MKII EDO ram
>>Is it possible to disable the 16 MB motherboard ram without being
>>forced to pull all the ZIPs out of their socket?
> Why would you do that ?
Described in a different posting.
> The MKII RAM should e at the highest priority in the mem list, so
> the onboard RAM will only be used if the requested memory by a
> program is too large for the free area in MKII RAM, i.e. it's full,
> so there should be no problem - performance wise ! If you want to
> change the priorities of the RAM areas there's a program on AmiNet
> that will allow you to do that, don't remember the name right now -
> something like SetRamPri or such...
The priority is not the problem.
> If some of your ZIPs should be faulty it would be the best idea to
> pull them out.
All RAM tests (with different test programs) reveal that the ZIPs are
fine. On the other hand, I believe that the problem is related to some
kind of weird RAM interface problem of the CyberSTorm and the
motherboard. I heart this already a couple of times, but nobody could
really tell me how I could poddibly check this out.
My problem is not the one mentioned on the p5 "Errata" page.
Another reason why I believe that tehre is a problem with the CS is,
that when I use ChipRAM as SCSI buffer, I get 50% faulty transfers.
Half of the files are corrupt! This also shows that in fact it has
nothing to do with the FastRAM.
> I don't know of any jumper or such to disable onboard FAST RAM !
Yes, this would be another possibility ...
--
Cu Georges Heinesch, Luxembourg
geo...@ibm.net - geo...@geocities.com
http://www.geocities.com/yosemite/2480
PGP 2.6.3i public key available on request and on public servers
... what goes up, must come down ...
Patrick Kursawe
Hi,
Georges Heinesch wrote:
>The reason is, that there are apparently faulty CyberStorm MKII boards
>around, where the RAM interface between the motherboard RAM and the
>CyberStorm RAM is faulty. It seems that whenever the CS RAM is full
>and the motherboard RAM has to be used, that erroneous data transfer
>occurs between the CPU and the motherboard RAM.
Yes, there are faulty boards around. One the p5 homepage there was (or is
still?) a page called "Errata" which describes the problem.
>BTW, I'd like to know if anybody heard anything about this problem!
I asked p5 support if a cooler was necessary for my card (040), the manual
said yes, but there was none. The guy there told me that indeed it was not
needed and, by the way, told my that my board was faulty and that I should let
them fix it. Didn't do it because I have no fast-mem on the motherboard and
most likely won't use SCSI with this board.
>Yesterday evening, I discovered for expample that the DIavolo SCSI
>transfer buffer set to CHIP created 50% faulty files during the
>backup. This was never the case before I installed the CS MKII. I'm
>sure about that.
Quite likely that this is the board's fault. There is some DMA trouble with
the faulty CS MkII.
Bye, Patrick
--
Patrick Kursawe Patrick...@ruhr-uni-bochum.de
Hohenzollernstr. 69 http://www.anachem.ruhr-uni-bochum.de/patrick
45128 Essen Keep smiling! :-)
Amiga makes it... well, not possible, but definitely more fun.
Georges Heinesch
In a message of 31-Oct-97 18:42:45, Hans Guijt wrote:
> Ian Gledhill (im...@nospam.aber.ac.uk) wrote:
>>>Is it possible to disable the 16 MB motherboard ram without being
>>>forced to pull all the ZIPs out of their socket?
> You could simply allocate it.
>>Why on earth do you want to do that? :-)
> Perhaps he has some program that doesn't like this RAM...
Well, it seems more that it is the my hardware (CyberStorm MKII) which
apparently doesn't like the ZIPs. To check that out, I'd like to
deselect them (preferentially with a program, to avoid pulling them
out one more time out of their socket).
--
Cu Georges Heinesch, Luxembourg
geo...@ibm.net - geo...@geocities.com
http://www.geocities.com/yosemite/2480
PGP 2.6.3i public key available on request and on public servers
... if you can't beat them, join them !
<-AD->
> >>>Is it possible to disable the 16 MB motherboard ram without being
> >>>forced to pull all the ZIPs out of their socket?
> apparently doesn't like the ZIPs. To check that out, I'd like to
> deselect them (preferentially with a program, to avoid pulling them
> out one more time out of their socket).
>
> --
> Cu Georges Heinesch, Luxembourg
> geo...@ibm.net - geo...@geocities.com
> http://www.geocities.com/yosemite/2480
> PGP 2.6.3i public key available on request and on public servers
>
>
Degrader (Chris Hames) allows you to block out selected address
ranges.
Good luck.
--
<-AD-> <morse 'at' ahab.demon.co.uk>
John Sheehy
Georges Heinesch <geo...@ibm.net> writes:
>Another reason why I believe that tehre is a problem with the CS is,
>that when I use ChipRAM as SCSI buffer, I get 50% faulty transfers.
>Half of the files are corrupt! This also shows that in fact it has
>nothing to do with the FastRAM.
Why don't you just change the last hex character in your HD masks to
'c'. That should eliminate the data alignment problems. Did for me,
with my GVP board.
--
<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
John P Sheehy <jsh...@ix.netcom.com>
><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><
Charles Cunningham
On 01 Nov 1997 15:38:40 Georges Heinesch wrote about "Re: Disable motherboard
RAM":
>
> In a message of 31-Oct-97 18:42:45, Hans Guijt wrote:
>
> > Ian Gledhill (im...@nospam.aber.ac.uk) wrote:
>
> >>>Is it possible to disable the 16 MB motherboard ram without being
> >>>forced to pull all the ZIPs out of their socket?
>
> > You could simply allocate it.
>
> >>Why on earth do you want to do that? :-)
>
> > Perhaps he has some program that doesn't like this RAM...
>
> Well, it seems more that it is the my hardware (CyberStorm MKII) which
> apparently doesn't like the ZIPs. To check that out, I'd like to
> deselect them (preferentially with a program, to avoid pulling them
> out one more time out of their socket).
>
> --
> Cu Georges Heinesch, Luxembourg
> geo...@ibm.net - geo...@geocities.com
> http://www.geocities.com/yosemite/2480
> PGP 2.6.3i public key available on request and on public servers
>
Many accelerators do not like motherboard ram...the WarpEngine being
one of them. I was advised by the folks at MacroSystems to pull the
memory out of my A3000T...all 16MB of it. They told me it might work
and it might not! They said some machines with it remaining in would
still work, but many if not most would not! However, they did tell me
that if it was left in and in fact did work that it would slow down the
accelerator. The best thing would be to just take it out!
>
Georges Heinesch
The problem doesn't only occur wiuth motherboard FastRAM (SC ZIPs).
There are also heaps of transfer errors when using ChipRAM.
E.g. Diavolo SCSI buffer set to ChipRAM corrupts 50% of all files.
This was not the case before I had the CyberStorm MKII! imho, it's no
solution to get rid of the motherborad ZIPs, if the ChipRAM transfer
is also erroneous. This one is required and solsered in anyway (at
least 1 MB).
--
Cu Georges Heinesch, Luxembourg
geo...@ibm.net - geo...@geocities.com
http://www.geocities.com/yosemite/2480
PGP 2.6.3i public key available on request and on public servers
... Amiga - Why use Windows, since there is a door ...
Georges Heinesch
In a message of 31-Oct-97 18:42:45, Hans Guijt wrote:
> Ian Gledhill (im...@nospam.aber.ac.uk) wrote:
>>>Is it possible to disable the 16 MB motherboard ram without being
>>>forced to pull all the ZIPs out of their socket?
> You could simply allocate it.
>>Why on earth do you want to do that? :-)
> Perhaps he has some program that doesn't like this RAM...
I which it would be the program. Did someone experience problems with
Diavolo 3.x and SCSI transfer buffer selected to ChipRAM (A3000 and
CyberStorm of course)?
--
Cu Georges Heinesch, Luxembourg
geo...@ibm.net - geo...@geocities.com
http://www.geocities.com/yosemite/2480
PGP 2.6.3i public key available on request and on public servers
... Reims ACC: "Say heading" - Pilot: "Heading!" ...
Georges Heinesch
In a message of 31-Oct-97 21:29:10, Joshua B. Helm wrote:
> Georges Heinesch wrote:
>> I have an A3000 with:
>>
>> - 2 MB chip ram
>> - 16 MB motherboard ZIP static column ram
>> - 32 MB CyberStormk MKII EDO ram
>>
>> Is it possible to disable the 16 MB motherboard ram without being
>> forced to pull all the ZIPs out of their socket?
>>
> You could jsut pull out the FIRST ZIP in the FIRST bank. The question
> is, why would you want to disable the motherboard RAM? If its because
> you want the CS RAM to be used instead, why not just increase the
> priority on the CS RAM (which should already be done), or decrease the
> priority on the motherboard RAM?
Simply because I thing theres a problem with the CyberStorm ram
interface. It doesn't handle the mothboard RAM correctly. That's why.
I'll check it out with p5 today.
Diavolo creates tons of corruot files here when backing up with the
SCSI buffer set to CHIP or DMA. That's imho an indication that their
RAM interface is faulty"!
--
Cu Georges Heinesch, Luxembourg
geo...@ibm.net - geo...@geocities.com
http://www.geocities.com/yosemite/2480
PGP 2.6.3i public key available on request and on public servers
... have a plan and make it work ...
Georges Heinesch
In a message of 04-Nov-97 03:06:23, John Sheehy wrote:
> Georges Heinesch <geo...@ibm.net> writes:
>>Diavolo creates tons of corruot files here when backing up with the
>>SCSI buffer set to CHIP or DMA. That's imho an indication that their
>>RAM interface is faulty"!
> Did you try changing the HD dma masks to end in "c", like I suggested?
Yep, I tried it right now.
I changed it from "0x7ffffffe" to "0x7ffffffc". Didn't help ;((
Since we're talking about masks, could you please tell me what this
mask is for exactly?
Why did you think that chandig the terminating "e" into an "c" would
help?
Many thanks!
--
Cu Georges Heinesch, Luxembourg
geo...@ibm.net - geo...@geocities.com
http://www.geocities.com/yosemite/2480
PGP 2.6.3i public key available on request and on public servers
... Only happy at FL 370, Mach 0.74
Mike Redrobe
Hi, Georges Heinesch ,
GH> Diavolo creates tons of corruot files here when backing up with the
GH> SCSI buffer set to CHIP or DMA. That's imho an indication that their
GH> RAM interface is faulty"!
does it happen with programs apart from diavolo ?
are you backing up to a tape drive,DAT drive or partition?
questions,questions ;)
Georges Heinesch
In a message of 04-Nov-97 13:50:45, Mike Redrobe wrote:
GH>> Diavolo creates tons of corruot files here when backing up with the
GH>> SCSI buffer set to CHIP or DMA. That's imho an indication that their
GH>> RAM interface is faulty"!
> does it happen with programs apart from diavolo ?
There's afaik no other program which has the CHIP/DMA/ANY selections.
> are you backing up to a tape drive,DAT drive or partition?
To DAT streamer.
> questions,questions ;)
Hopefully a couple of answers ;))
Do you have a CyberStorm, Diavolo and a DAT streamer?
TIA
--
Cu Georges Heinesch, Luxembourg
geo...@ibm.net - geo...@geocities.com
http://www.geocities.com/yosemite/2480
PGP 2.6.3i public key available on request and on public servers
... niammi niammi in my Tammi ...
John Sheehy
Georges Heinesch <geo...@ibm.net> writes:
>In a message of 04-Nov-97 03:06:23, John Sheehy wrote:
>
>> Georges Heinesch <geo...@ibm.net> writes:
>
>>>Diavolo creates tons of corruot files here when backing up with the
>>>SCSI buffer set to CHIP or DMA. That's imho an indication that their
>>>RAM interface is faulty"!
>
>> Did you try changing the HD dma masks to end in "c", like I suggested?
>
>Yep, I tried it right now.
>
>I changed it from "0x7ffffffe" to "0x7ffffffc". Didn't help ;((
Oh, well, too bad. It works in many cases. You're the first person
I've given this advice to, for whom it didn't work. You did reboot
after the changes, didn't you?
>Since we're talking about masks, could you please tell me what this
>mask is for exactly?
It says what memory addresses DMA can occur to.
>Why did you think that chandig the terminating "e" into an "c" would
>help?
It does in many cases of DMA corruption. On many DMA controllers, like
GVP's, certain 2-byte sequences of data being read by programs that are
reading the data piecemeal and processing it, like many ilbm parsers,
can get dropped if the buffer is not longword aligned. Using 'c' forces
longword alignment.
Georges Heinesch
In a message of 04-Nov-97 16:08:02, John Sheehy wrote:
> Georges Heinesch <geo...@ibm.net> writes:
>>In a message of 04-Nov-97 03:06:23, John Sheehy wrote:
>>
>>> Georges Heinesch <geo...@ibm.net> writes:
>>
>>>>Diavolo creates tons of corruot files here when backing up with the
>>>>SCSI buffer set to CHIP or DMA. That's imho an indication that their
>>>>RAM interface is faulty"!
>>
>>> Did you try changing the HD dma masks to end in "c", like I suggested?
>>
>>Yep, I tried it right now.
>>
>>I changed it from "0x7ffffffe" to "0x7ffffffc". Didn't help ;((
> Oh, well, too bad. It works in many cases. You're the first person
> I've given this advice to, for whom it didn't work. You did reboot
> after the changes, didn't you?
Yes, I rebooted!
But why did you advice other people to do the same. Did they have the
same problem as I have with the CyberStorm?
>>Since we're talking about masks, could you please tell me what this
>>mask is for exactly?
> It says what memory addresses DMA can occur to.
Where is the DMA buffer beginning if I set "0x7ffffffe" and where when
I set the mask to "0x7ffffffc"?
What is the range of the DMA address band?
>>Why did you think that chandig the terminating "e" into an "c" would
>>help?
> It does in many cases of DMA corruption. On many DMA controllers, like
> GVP's, certain 2-byte sequences of data being read by programs that are
> reading the data piecemeal and processing it, like many ilbm parsers,
> can get dropped if the buffer is not longword aligned. Using 'c' forces
> longword alignment.
Hmmm ... would you mind providing me details why the "c" at the end
forces longword alignment (perhaps via e-mail)?
--
Cu Georges Heinesch, Luxembourg
geo...@ibm.net - geo...@geocities.com
http://www.geocities.com/yosemite/2480
PGP 2.6.3i public key available on request and on public servers
... her name was Lola, ...