welland@cbmvax.UUCP (Bob Welland) (11/11/87)
References:
I was dissapointed to see that no one who was at the COMDEX
show reported on the Commodore 68020 board for the A2000. It
was shown in the booth by Dave Haynie (he and I where the
designers). It has some nice features !
CPU, speed 68020, 14.2 Speed in Mhz
FPU speed 14.2, 20, 25 Speed in Mhz
(68881) Faster speeds are optional
MMU Yes Wonder what this is for ?
(68851)
RAM option 0, 1M, 2M Numbers in bytes
RAM width 32 bits
RAM cycle 350ns 70ns of the
time time is due to the
68851 MMU chip.
RAM expansion Yes
on-board?
RAM is Yes
auto-config?
Boot ROM's Yes 64K bytes
A few notes:
The RAM is auto-config so it can be DMA'ed to etc ...
The on boot ROM takes over the system and so it can play
nice games before handing things over to AmigaDos (or not).
The 68881 has an optional crystal that can provide clock
rates greater then 14 Mhz (for those floating point junk-
ies). The RAM speed has 2 wait states (one is from the MMU
and one is from the DRAM refresh arbitration).
We where showing the board running Sculpt 3-D.
Price and delivery have not been established as of yet.
Robert Welland
All of the above opinions above are mine ... they do not
represent Commodore in any way.ejkst@cisunx.UUCP (Eric J. Kennedy) (11/14/87)
In article <2720@cbmvax.UUCP>, welland@cbmvax.UUCP (Bob Welland) writes: > > I was dissapointed to see that no one who was at the COMDEX > show reported on the Commodore 68020 board for the A2000. It > was shown in the booth by Dave Haynie (he and I where the > designers). It has some nice features ! > [...] > RAM option 0, 1M, 2M Numbers in bytes > > RAM width 32 bits > [...] > RAM expansion Yes > on-board? > > RAM is Yes > auto-config? > Can this 32 bit memory be used along with 16 bit memory expansions? If so, how does it work? I.e. what determines which type of memory is used if both types are available to the system. Also, I seem to remember reading that CSA's Turbo-Amiga has a 12 Meg limit of 32 bit memory. How does that work? Just what are the limits with a 32 bit system? Thanks, Eric Kennedy
grr@cbmvax.UUCP (George Robbins) (11/16/87)
In article <5338@cisunx.UUCP> ejkst@cisunx.UUCP (Eric J. Kennedy) writes: > In article <2720@cbmvax.UUCP>, welland@cbmvax.UUCP (Bob Welland) writes: > > > > I was dissapointed to see that no one who was at the COMDEX > > show reported on the Commodore 68020 board for the A2000. It > > was shown in the booth by Dave Haynie (he and I where the > > designers). It has some nice features ! > > > > RAM option 0, 1M, 2M Numbers in bytes > > RAM width 32 bits > > RAM expansion Yes > > on-board? > > RAM is Yes > > auto-config? > > > Can this 32 bit memory be used along with 16 bit memory expansions? > If so, how does it work? I.e. what determines which type of memory > is used if both types are available to the system. The memory will auto-config along with any memory on the expansion bus. It gets configured first, thus will most likely be at the 200000-3FFFFF spot. AmigaDOS doesn't have any real preference for this memory, but one could probably do some tuning by editing the memory list before starting ram-disks and other low-priority items. > Also, I seem to remember reading that CSA's Turbo-Amiga has a 12 Meg > limit of 32 bit memory. How does that work? Just what are the limits > with a 32 bit system? If the memory is placed outside the normal 16MB space, all you want. However, on the current 68020 board we chose to place the memory within the confines of the normal expansion bus architecture, mostly so that DMA devices can transfer data directly into the memory, without having to do a programmed memory-to-memory transfer. I'm not sure of the 12 MB limit on the CSA board, I'd guess it's some kind of electrical or loading limit, since they only get a little memory on each of their static RAM cards. -- George Robbins - now working for, uucp: {ihnp4|rutgers|allegra}!cbmvax!grr but no way officially representing arpa: out to lunch... Commodore, Engineering Department fone: 215-431-9255 (only by moonlite)
daveh@cbmvax.UUCP (Dave Haynie) (11/17/87)
in article <5338@cisunx.UUCP>, ejkst@cisunx.UUCP (Eric J. Kennedy) says: > > Can this 32 bit memory be used along with 16 bit memory expansions? Yes, absolutely, no problem. > If so, how does it work? I.e. what determines which type of memory > is used if both types are available to the system. This memory looks just like any other FAST memory. Given that it is, by the definition of the way we make the Amiga OS see the memory as autoconfig memory, the first memory card configured, it is also, given the way AllocMem() works, the first memory to get used when you allocate FAST RAM. Providing you run SlowMemLast to get $C00000 out of the way. > Also, I seem to remember reading that CSA's Turbo-Amiga has a 12 Meg > limit of 32 bit memory. How does that work? Just what are the limits > with a 32 bit system? The 68020 is limited to 4 gigabytes, but real-life systems don't give you all that much. Our card doesn't offer 32 bit expansion, just the 2 megs on the card. I think CSA sticks their memory outside of normal autoconfiguration space, perhaps up above the 68000 24 bit addressing limit. CSA provides an add-on extension connector on their board, but no on-board RAM. If they're limited to 12 meg of expansion RAM, that's a limitation of their specific implementation, not of the Amiga system as a whole. > Thanks, > > Eric Kennedy -- Dave Haynie Commodore-Amiga Usenet: {ihnp4|caip|rutgers}!cbmvax!daveh "The B2000 Guy" PLINK : D-DAVE H BIX : hazy "Computers are what happen when you give up sleeping" - Iggy the Cat
rusty@vertigo.UUCP (M.W.HADDOCK) (11/17/87)
In article <2801@cbmvax.UUCP> daveh@cbmvax.UUCP (Dave Haynie) writes: >in article <5338@cisunx.UUCP>, ejkst@cisunx.UUCP (Eric J. Kennedy) says: >> >> Also, I seem to remember reading that CSA's Turbo-Amiga has a 12 Meg >> limit of 32 bit memory. How does that work? Just what are the limits >> with a 32 bit system? > >The 68020 is limited to 4 gigabytes, but real-life systems don't give you all >that much. Our card doesn't offer 32 bit expansion, just the 2 megs on the >card. I think CSA sticks their memory outside of normal autoconfiguration >space, perhaps up above the 68000 24 bit addressing limit. CSA provides an >add-on extension connector on their board, but no on-board RAM. If they're >limited to 12 meg of expansion RAM, that's a limitation of their specific >implementation, not of the Amiga system as a whole. > I've had a Turbo-Amiga ever since CSA's first expansion boxes came out and I've never heard of a 12-Meg limit. Oh, maybe you mean their Turbo-Tower which has more slots. Yes, I just checked and it has 7 Zorro slots so alloc' 1(one) for the '020 card and the remaining 6 for their 2-Meg static RAM cards. At least the my memory cards, Zorro I form-factor, 32-bits wide data & address, attach to the CPU card via very short ribbon cables along the sides of the card. The memory starts at address 0x7f000000, well above the 68000's address and autoconfig range. Since I haven't seen the Turbo-Tower, which is different from the original Turbo-Amiga, (where do they git these names? :-) I don't know how they may be working things now. If you want to know more I'll be glad to answer any questions. -Rusty- DISCLAIMER: I am not connected with CSA 'cept as a satisfied customer. My problem is that I doubt I could go back to a lowly, 7.16-MHz 68000 Amiga. Speed withdrawl symptoms, don't ya know... :-) -- Rusty Haddock {{uunet!likewise!}cbosgd,rutgers!moss}!vertigo!rusty AT&T-IS Consumer Products Laboratories - Human Factors Laboratory Holmdel, New Joysey 07733 (201) 834-1023 -- Being schizophrenic is better than living alone. --
hah@mipon3.intel.com (Hans Hansen) (11/18/87)
In article <5338@cisunx.UUCP> ejkst@cisunx.UUCP (Eric J. Kennedy) writes:
$Can this 32 bit memory be used along with 16 bit memory expansions?
$If so, how does it work? I.e. what determines which type of memory
$is used if both types are available to the system.
The 68020 and 68030 have two input pins that are used to perform dynamic
bus sizing. What this means is that the hardware used to decode the
addresses of RAM, ROM, and misc., I/O controllers should/must also
configure these two input pins. As the 68020 addresses various segments
of the Amiga memory map the size bits are changed to reflect the width
of the memory of controller. With this arrangement the 68020 can even
use 8 bit wide RAM and ROMs. The programming model does not change as
the 68020 will adjust its reads and writes depending on the target
address width.
$
$Also, I seem to remember reading that CSA's Turbo-Amiga has a 12 Meg
$limit of 32 bit memory. How does that work? Just what are the limits
$with a 32 bit system?
This is only a limit of their RAM boards. The 68020 can address a full
32 bit address space in each of four modes, (however this feature has
not been used to my knowlege, as 4 GigaBytes of linear 'real' memory
has proven to be enough for now.)
What the 68020 boards do is to treat the Amiga as a 16 MegaByte 16 bit
wide slot within the total address space. The 32 BitWide 68020 RAM
expansion is made outside of the Amiga 16 MegaByte slot. So for the
most part while addressing the Amiga slot the dynamic bus sizing is set
to 16 bits, however there are a few I/O ports that most be decoded to
8 bits, i.e., the 8520s are 8 bit devices.
For more info get a copy of the "MC68020 32-Bit Microprocessor User's
Manual" from Motorola.
$
$Thanks,
$
$Eric Kennedy
Your welcome
Hans hah@inteloa.intel.comdaveh@cbmvax.UUCP (11/19/87)
in article <1235@omepd>, hah@mipon3.intel.com (Hans Hansen) says: > > What the 68020 boards do is to treat the Amiga as a 16 MegaByte 16 bit > wide slot within the total address space. The 32 BitWide 68020 RAM > expansion is made outside of the Amiga 16 MegaByte slot. So for the > most part while addressing the Amiga slot the dynamic bus sizing is set > to 16 bits, however there are a few I/O ports that most be decoded to > 8 bits, i.e., the 8520s are 8 bit devices. You could size the bus for 8520s if you were starting out from scratch, but in reality this isn't done. The main reason is that the 8520s are set up for 16 bit reads and writes. One is connected to 68000 data lines 0-7, the other to 68000 data lines 8-15. The main reason is that the 68000 doesn't do dynamic bus sizing; it only sizes relative to instruction width, not device port size. In fact, it would be impossible on the Amiga to have the 68020 dynamically size the 8520s to 8 bits. The reason is the way the 68020 port devices go up the data bus. A 32 bit device is connected to bits 0-31. A 16 bit device connects to bits 16-31. And an 8 bit device connects to bits 24-31. The entire Amiga system looks like a 16 bit port to the 68020. CSA RAM goes outside the normal 16 bit address space of the 68000. The Commodore-Amiga 68020 card has fast 32 bit memory that's put in the normal 68000 address space (it looks just like a normal 2 meg expansion card to the system). This has the disadvantage of not permitting more than the standard amount of RAM under normal conditions. But it also allows that memory to be normal FAST RAM (so it gets allocated first), and it allows 16 bit DMA devices to work with it. > Hans hah@inteloa.intel.com -- Dave Haynie Commodore-Amiga Usenet: {ihnp4|caip|rutgers}!cbmvax!daveh "The B2000 Guy" PLINK : D-DAVE H BIX : hazy "I can't relax, 'cause I'm a Boinger!"
shs@ji.Berkeley.EDU (Steve Schoettler) (11/20/87)
In article <177@vertigo.UUCP> rusty@vertigo.UUCP (91341-M.W.HADDOCK) writes: > The memory starts at address 0x7f000000, > > -Rusty- > Oops, I was afraid of that. Does anyone know if CSA has a jumper or switch on the board to change the base address of expansion memory for the 68020? It turns out that the best Prolog implementations (to date) are on tagged architectures. The current method for implementing Prolog on 32-bit machines is to use the upper 3 or 4 bits for tag and leave the extra 28 bits for data. The advantage is you get a tremendous speedup and use much less memory than if you put the tags in another memory word. The disadvantage is that pointers and integers can only be 28 bits long. Thus, if a 68020 board has memory above 0x0fffffff, prolog won't be able to access it. To the fellow porting CProlog to the Amiga: Looks like it may not work on a CSA board. That's too bad because Prolog can use all the speedup it can get. To others who know: Can the CSA board be jumpered to use another address space? Do other Amiga 68020 cards (like Finally, CBM, ASDG) put memory way up there? I don't intend to start a flame on whether or not one should use extra address bits for tags. The implementors of Cprolog and similar systems made some design decisions and tradeoffs. I just want to know if they can be made to work on an Amiga. -Steve Schoettler Disclaimer: I am not associated with any company currently marketing prolog or 68020 products.
daveh@cbmvax.UUCP (11/23/87)
in article <21872@ucbvax.BERKELEY.EDU>, shs@ji.Berkeley.EDU (Steve Schoettler) says: > > In article <177@vertigo.UUCP> rusty@vertigo.UUCP (91341-M.W.HADDOCK) writes: >> The memory starts at address 0x7f000000, >> >> -Rusty- > It turns out that the best Prolog implementations (to date) are on tagged > architectures. ... > Thus, if a 68020 board has memory above 0x0fffffff, prolog won't be able > to access it. It's a little more involved that that, Boso! Anything that uses upper bits as tagged won't work on any 68020 system that does fully decoded addressing. So if my memory board is at, say, 0x01000000, it's not going to be at 0x11000000 or 0x21000000, etc. if I'm fully decoding the 68020 addresses. > To others who know: > Can the CSA board be jumpered to use another address space? > Do other Amiga 68020 cards (like Finally, CBM, ASDG) put memory way > up there? The Amiga 68020 card doesn't put memory "up there", but like I explained, it's not the location that matters. The Amiga board DOES have an MMU. This would allow you to map "up there" memory "down here", and (what's more important for tagged systems), MAY let you program memory wrapping, so your tags don't affect anything (I haven't actually tried this, but I suspect it would work). What you really want is the TI LISP chip on a coprocessor card. > -Steve Schoettler -- Dave Haynie Commodore-Amiga Usenet: {ihnp4|caip|rutgers}!cbmvax!daveh "The B2000 Guy" PLINK : D-DAVE H BIX : hazy "I can't relax, 'cause I'm a Boinger!"
shs@ji.Berkeley.EDU (Steve Schoettler) (11/24/87)
In article <2844@cbmvax.UUCP> daveh@cbmvax.UUCP (Dave Haynie) writes: >in article <21872@ucbvax.BERKELEY.EDU>, shs@ji.Berkeley.EDU (Steve Schoettler) says: > >> Thus, if a 68020 board has memory above 0x0fffffff, prolog won't be able >> to access it. > >It's a little more involved that that, Boso! Anything that uses upper bits >as tagged won't work on any 68020 system that does fully decoded addressing. >So if my memory board is at, say, 0x01000000, it's not going to be at >0x11000000 or 0x21000000, etc. if I'm fully decoding the 68020 addresses. It's exactly that involved, but you have to know something about the implementation. The scenerio goes: you fetch a 32 bit value. The upper 4 bits tell you whether it's an integer, a pointer, a list, etc. If it's a pointer, you MASK OFF the 4 bits (and.l #0ffffffff, d0) and access that address. Yes, its slower, but that's what you have to do to work in cases like you mentioned. >The Amiga 68020 card doesn't put memory "up there", but like I explained, >it's not the location that matters. The Amiga board DOES have an MMU. >This would allow you to map "up there" memory "down here", and (what's >more important for tagged systems), MAY let you program memory wrapping, so >your tags don't affect anything (I haven't actually tried this, but I >suspect it would work). The addition of that hardware might make it possible to eliminate the extra MASK step. > >What you really want is the TI LISP chip on a coprocessor card. > What I really want is the Berkeley PLM (Prolog Logic Machine) chip on a coprocessor card. By the way, why do you guys keep calling it a COprocessor card? Isn't it more like an instead-of-the-main-processor card? :~) While we're on the subject (we weren't, but it seemed like a good place to bring it up), I'm really hoping ASDG's SDP can run other tasks on its 68000 if it's not doing I/O. It would be really fun to try to convert it to a "remote" Videoscape spooler or something like that. >-- >Dave Haynie Commodore-Amiga Usenet: {ihnp4|caip|rutgers}!cbmvax!daveh > "The B2000 Guy" PLINK : D-DAVE H BIX : hazy > "I can't relax, 'cause I'm a Boinger!" Steve Schoettler
daveh@cbmvax.UUCP (Dave Haynie) (11/25/87)
in article <21922@ucbvax.BERKELEY.EDU>, shs@ji.Berkeley.EDU (Steve Schoettler) says: >>It's a little more involved that that, Bosco! Anything that uses upper bits >>as tagged won't work on any 68020 system that does fully decoded addressing. >>So if my memory board is at, say, 0x01000000, it's not going to be at >>0x11000000 or 0x21000000, etc. if I'm fully decoding the 68020 addresses. > > It's exactly that involved, but you have to know something about the > implementation. The scenerio goes: > you fetch a 32 bit value. The upper 4 bits tell you whether it's > an integer, a pointer, a list, etc. > If it's a pointer, you MASK OFF the 4 bits (and.l #0ffffffff, d0) and > access that address. Right, in the context of an interpreter. I was thinking in terms of best speed, not best implementability, on a typical 68020 system. So I guess, if you do it that way, that any extra memory could just be ignored, instead of causing problems. However, if you've got the extra mask step, is using the tagged architecture really that much faster than calling each element two words instead of one, and storing the tags in the second word. Certainly this would eat memory at twice the rate, but the mask operation can't be much faster than the second fetch, can it? >>What you really want is the TI LISP chip on a coprocessor card. >> > What I really want is the Berkeley PLM (Prolog Logic Machine) chip > on a coprocessor card. OK. > By the way, why do you guys keep calling it a COprocessor card? Isn't it more > like an instead-of-the-main-processor card? :~) Well, the current implementation of our 68020 card is in fact an "instead of the main processor" card. There's no reason, however, why you could put the TI, PLM, or a 68020 in that slot and implement it as a real, "works in concert with the 68000" coprocessor type card, the "Coprocessor" slot facilitates this type of switching. >>Dave Haynie Commodore-Amiga Usenet: {ihnp4|caip|rutgers}!cbmvax!daveh > > Steve Schoettler -- Dave Haynie Commodore-Amiga Usenet: {ihnp4|caip|rutgers}!cbmvax!daveh "The B2000 Guy" PLINK : D-DAVE H BIX : hazy "I can't relax, 'cause I'm a Boinger!"
randy@bcsaic.UUCP (Randy Groves) (11/26/87)
In article <2844@cbmvax.UUCP> daveh@cbmvax.UUCP (Dave Haynie) writes: > >What you really want is the TI LISP chip on a coprocessor card. > >> -Steve Schoettler > >-- >Dave Haynie Commodore-Amiga Usenet: {ihnp4|caip|rutgers}!cbmvax!daveh > "The B2000 Guy" PLINK : D-DAVE H BIX : hazy > "I can't relax, 'cause I'm a Boinger!" At the risk of starting religious wars - what you REALLY want is a Symbolics Ivory chip on a coprocessor card. Seriously, it is also to a consideration. Which company gets around to it first will be the question. Symbolics is looking at the 386 family pretty heavily for a host for a coprocessor card. If that comes to fruition in the near future, it is definitely something to look at - I'd love to have the complete Symbolics Genera environment available on an Amiga!! -- -randy groves - Boeing Advanced Technology Center UUCP: ..!uw-beaver!uw-june!bcsaic!randy USNail: Boeing Computer Services CSNET: randy@boeing.com PO Box 24346 M/S 7L-68 VOICE: (206)865-3424 Seattle, WA 98124 -- -randy groves - Boeing Advanced Technology Center UUCP: ..!uw-beaver!uw-june!bcsaic!randy USNail: Boeing Computer Services CSNET: randy@boeing.com PO Box 24346 M/S 7L-68 VOICE: (206)865-3424 Seattle, WA 98124
cc1@CS.UCLA.EDU (11/30/87)
In article <21922@ucbvax.BERKELEY.EDU> shs@ji.Berkeley.EDU.UUCP (Steve Schoettler) writes: >While we're on the subject (we weren't, but it seemed like a good place to >bring it up), I'm really hoping ASDG's SDP can run other tasks on its 68000 >if it's not doing I/O. It would be really fun to try to convert it to a >"remote" Videoscape spooler or something like that. Gee, a while ago I asked if we could just plug in a 68010 and put unix on the beast. Realize that it has a mmu and a disk drive connected to it. Michael Gersten