knudsen@ihwpt.UUCP (mike knudsen) (02/19/86)
First, I'd like to know how many chips in the ST are soldered in instead of socketed, and which these are. Please answer this for the early STs, the current models, and the future shipments. I'd like to decide which vintage model would be best for doing hardware upgrades (piggy back RAM, 68010, etc). More serious: How can the ST be bus-expanded? The *other* brand brings its entire bus to the side connector, but the ST has only a read-only 128K game cart socket (what, games? on an Atari??). How can there be an expansion chassis into which we can plug boards with lotsa RAM, 32000 co-procs, or whatever else needs some direct bus access? Will we be stuck with a box hung off the floppy or hard disk serial ports? Which can't be accessed directly? Tecmar is already advertising a true expansion box for the you-know-what. Finally: Is it possible to write directly to the graphics bitmap in RAM (from an assembler or C program)? Where speed is of the utmost, we'll need to be able to do our own custom bit-blits, quickdraws, etc. to compete with the custom graphics chips in the high-priced machine. I heard that on one of the machines (forget which) you were somehow prevented from direct access to the graphics area (this may just be a BASIC restriction, as there is no MMU in either machine to stop machine code routines from doing it). Hoping to get an ST soon, mike k
lbl@druhi.UUCP (LocklearLB) (02/19/86)
> First, I'd like to know how many chips in the ST are > soldered in instead of socketed, and which these are. > Please answer this for the early STs, the current models, > and the future shipments. I'd like to decide which > vintage model would be best for doing hardware upgrades > (piggy back RAM, 68010, etc). In the revision B board, all of the chips are soldered except for the custom chips (Memory Manager, Glue, Video). This probably will not change in the future, except that the custom chips might also get soldered in also. I have done the ram upgrade on my board with no problems. I have also talked with several people who have done the upgrade on newer machines with no problem. Get the latest vintage machine you can. > > More serious: How can the ST be bus-expanded? > The *other* brand brings its entire bus to the side connector, > but the ST has only a read-only 128K game cart socket > (what, games? on an Atari??). How can there be an > expansion chassis into which we can plug boards with > lotsa RAM, 32000 co-procs, or whatever else needs some > direct bus access? Will we be stuck with a box hung > off the floppy or hard disk serial ports? > Which can't be accessed directly? > Tecmar is already advertising a true expansion box > for the you-know-what. The floppy disk and hard disk ports are parallel ports, NOT serial ports. The hard disk port is really a general purpose DMA port for the Atari. This port, with some massaging, can be made to look like a SASI or SCSI port. This, of course, takes a little extra hardware on a PC board. The Atari hard disk is a disk connected to the extra hardware which is then connected to the Atari. The hard disk thinks it is talking SASI (or SCSI, I can't remember at the moment). Presumably, one could build an expansion cabinet using this method of communicating. It isn't as good as a full bus expander, but it can be very useful. > > Finally: Is it possible to write directly to the graphics > bitmap in RAM (from an assembler or C program)? > Where speed is of the utmost, we'll need to be able > to do our own custom bit-blits, quickdraws, etc. > to compete with the custom graphics chips in the > high-priced machine. > I heard that on one of the machines (forget which) > you were somehow prevented from direct access to the > graphics area (this may just be a BASIC restriction, > as there is no MMU in either machine to stop machine > code routines from doing it). > There IS a primitive MMU in the Atari that will prevent access to certain areas of memory if you are not executing in 'supervisor' mode. And yes, it is possible to write directly to the bitmap RAM. You won't need to write your own bit-blits, quickdraws, etc. because they are already provided in the line-A interface in the ROM of the Atari. This code is what the low-level routines in GEM and VDI end up calling. If you wish to bypass GEM to get ultimate speed, all you need to do is use these routines. > Hoping to get an ST soon, mike k I hope you do too. Order a development kit also, it will be worth the time it takes to delve through all of the crud to get the juicy parts. Atari and Digital Research are introducing a new version of the kit within a month that is supposed to have many bug fixes and a total (and much needed!) reorganization of the documentation. You should probably wait until that comes out to buy one. Good Luck. Barry Locklear AT&T Information Systems Labs Denver, CO (303) 538-4954 ihnp4!druhi!lbl
bammi@cwruecmp.UUCP (Jwahar R. Bammi) (02/19/86)
> First, I'd like to know how many chips in the ST are > soldered in instead of socketed, and which these are. > Please answer this for the early STs, the current models, > and the future shipments. I'd like to decide which > vintage model would be best for doing hardware upgrades > (piggy back RAM, 68010, etc). > In all the ST's that I have seen, Only the Glue, MMU, Roms and Dma chip are socketed. The rest are soldered. Having participated in 4 piggy back memory upgrades, I don't see how socketed memory chips would help. It would help ofcourse in replacing chips, but then the socket may be put in at that time. > More serious: How can the ST be bus-expanded? > The *other* brand brings its entire bus to the side connector, > but the ST has only a read-only 128K game cart socket > (what, games? on an Atari??). How can there be an > expansion chassis into which we can plug boards with > lotsa RAM, 32000 co-procs, or whatever else needs some > direct bus access? Will we be stuck with a box hung > off the floppy or hard disk serial ports? > Which can't be accessed directly? > Tecmar is already advertising a true expansion box > for the you-know-what. Well it depends on how you look at it, if you really need a bus coming out to the edge. If you do a co-processor box with its own backplane etc, then all you need the ST for is to do I/o for you. Further, a hard disk could be used directly without the need for going through the ST. So now you would need the St for graphics,Screen/Kbd i/o, floppy and for access to the various ports. If you do the interface right, the 10 Mhz DMA port is plenty fast. I know of two such boxes, the TT (apparently it will be announced at the Hanover Fair next month) and the $199 Ibm Pc compatible 8088 box. Most people who critisize the St for not having expansion slots fail to point out how many of the ports that normally sit on expansion cards on things like the Ibm Pc are built in. Further, a lot of people seem to have the misconception that a particular hardware subsystem has to sit on processors/backplane bus for efficiency. If the functinal partitioning is done correctly (ie. you dont need to get at every signal of the processor/backplane bus, but just need to pass data fast between the two subsystems, and the minimum amount of control or state info that you can get away with) then i claim that the Dma approach is superior (No electrical loads, simpler control, clean interface beween subsystems, no bus - are some of the reasons that i come to mind). I am not totally convinced that a backplane is required. But then I may be totally wrong. Can anyone convince me otherwise? (BTW for the price that Tecmar is charging for the expansion box i can have 3 1/2 St's sitting on my desk!! ) > > Finally: Is it possible to write directly to the graphics > bitmap in RAM (from an assembler or C program)? > Where speed is of the utmost, we'll need to be able > to do our own custom bit-blits, quickdraws, etc. > to compete with the custom graphics chips in the > high-priced machine. > I heard that on one of the machines (forget which) > you were somehow prevented from direct access to the > graphics area (this may just be a BASIC restriction, > as there is no MMU in either machine to stop machine > code routines from doing it). > yes you can write directly to screen memory. In fact you can designate any 32K chunk of memory on a 256 word boundary to be the screen memory (gets switched at the next Vblank ). In the V blanking period the screen memory/ pallette can be manupulated to get interesting effects. It is also possible to dynamically link in your own handlers for the blanking interrupts. > Hoping to get an ST soon, mike k -- Jwahar R. Bammi Usenet: .....!decvax!cwruecmp!bammi CSnet: bammi@case Arpa: bammi%case@csnet-relay CompuServe: 71515,155