Conference hydra::amiga_v1

Re: .0

I have heard of the project, too.  I believe they call themselves
the Excalibur project.  I believe they also have other public domain
hardware projects like ram boards and such in the works.

However, I haven't heard of anything being completed.

    I don't subscribe to GEnie, but I have seen an EXCALIBER.ARC file
    on Plink.  The file descriptor refers to a GEnie do-it-yourself
    Amiga SCSI board & driver.
    
    If I remember, I can download it this weekend.  I think its just
    a bunch of ASCII files.
    
    

    That would be great!  I'll check mail occasionally over the weekend.
    Regards,
    John
    
    p.s. If I can get or generate the necessary pcb artwork, is anyone
    interested in "chipping in" for a small production run of boards?
    I think I might also modify the design so it connects to the Starboard
    expansion port.
    

    ...since yesterday I put together all of the Fish disk listings
    that have been posted and search them for scsi, which turned up
    the following:

AmScsi		Preliminary documentation for a hardware project to
		build a SCSI controller board.  The design does not
		support DMA or AUTOCONFIG'ing.
		Author:  Rich Frantz    Fish disk 66

ScottDevice	A mountable MicroForge SCSI driver.  Supports one SCSI buss
		device with two hard disk units attached to it.  Version 
		33.1, includes source.
		Author:  Scott Turner   Fish disk 84
 
 
    note that the first reference says preliminary; I'd appreciate still
    an upload of the latest possible Genie net/plink project documnetation.
    John

    I've stumbled onto some incomplete message threads on Plink recently.
    Someone was querying Pery Kivolowitz (ASDG) about the status of
    a new product they are developing called SDP.
    
    What caught my eye was one of Perry's remarks regarding hard disk
    throughput... Perry claimed that 300K Bytes/sec was possible now,
    and they were working towards 400K/sec.
    
    I assume Perry is working on a new HD adaptor/driver package.  I
    don't know if it's SCSI or a custom ASDG job.  I do know that 400K/sec
    is pretty damn quick by any standards, since the PC AT runs at about
    150K/sec.
    
    I plan on leaving Perry a note asking for more details.  I'll post
    any info I can glean here.

I heard about that from Perry at the last BCS Amiga general meeting.  He gave
some good general info on the new controller.  It will have its own 68000, ram,
dma, memory management unit, floating point chip socket, slicer, dicer, nuclear
reactor, and (I guess) SCSI controller chip. I'm not guessing about SCSI, it
will do that, only don't know implementation details.  About the speeds, I don't
think 300-400k bytes per second is that fast, considering that SCSI is spec'd at
1.5Mbytes per second asynchronous mode, 4Mbytes per second synchronous.  The
bottleneck for a system with a good controller should be the storage device,
which today probably won't support those speeds.  Anyway, he mentioned something
like $400 for the beast, which is a lot for the function it performs, in my
mind.  All I want is a simple SCSI controller; after that, I would add DMA.
After that I'd want an SCSI device that has eight serial/parallel ports
implemented with Commodore's serila /par. chips so that I could relocate
(physically) the ser., par., and floppy interfaces and throw everyone (and their
buffers) out of chip ram. 
    

    (re .6)
    
    I'm curious as to how to control an external serial port, seeing's how
    I have a T-card with a serial port hanging off, and a blown internal
    serial port.  I have part of the source for a serial.device but it
    dont assemble. Does anyone have source for a working serial.device??
    
    Thanks
    Robb

    don't really know right now; I was thinking that you'd just point
    the serial device driver at the scsi driver, which would have the
    smarts to pass reads and writes to an address across the scsi buss.
    I see it as mapping the serial chip's physical address into scsi
    space.  It would be as if the serial chip's control block and data
    block were written or read and then transferred across the scsi
    buss as scsi control and data blocks.
    

Re: .5, .6:

The ASDG SDP controller is SCSI controller that has its own 68000 and
half meg of ram for its own exclusive use.  It achieves high speeds by
knowing the structure of Amiga DOS disks and prereading as much as possible
into its half-meg internal buffer.  Thus, it should perform really well
even with slow drives.

To meet changing file system formats as AmigaDOS evolves, the SDP reads
a configuration file that teaches it about the AmigaDOS disk structure.

If ASDG manages to pull this off, they will have the fastest controller
for any PC.  However, this is a radically different type of controller
than anyone has built for SCSI.  I am taking a wait and see attitude.

They claim that the board is in layout, that demos will be available
in September, and low volume shipments will occur in October.  The software
that sits in the board will be initially scaled down, and will evolve
to become more intelligent during the first six months or so of the
product's life.

Data point for evaluating all of this:  Last fall during the Commodore
fair held in Marlboro, Perry told me that this disk controller would
be out sometime in the spring.  So, they are already late.

    I downloaded and examined a file on Plink called 'EXCALIBUR.ARC'
    which is the first installment of Genie's homebrew SCSI adaptor/drive.
    
    There is virtually no information in this file, just a few digitized
    pictures of an old IMI 5 1/4" full high drive.  The author (Scott
    Turner) warns us not to open the drive, since debris will enter
    and cause the heads to crash.  Real useful stuff.  No schematics,
    parts list, driver source code, etc.
    
    Either Genie hasn't gotten very far with this project, or I missed
    90% of the useful files.  I doubt if the latter is true, since I searched
    the Plink database pretty thoroughly for threads on 'Genie', Excalibur',
    'SCSI' etcetera.
    
    Any Genie users in the house that can get an update?
                                                              
    If anyone is interested in seeing some med-res digipics of a four-year
    old drive, look in JAKE::USER2:[ACCIARDI.AMIGA]
                       

    re.: .10  Thanks very much for the try.  
    	I looked at the AmSCSI on fish disk 66, and it is a first cut
    at a real SCSI adapter.  Not as thorough as the memory upgrade on
    another fish disk, but a start anyway.  The only driver on the disk
    is a basic peek and poke routine used for hardware shakedown that
    takes  seconds to read a sector.  The hardware, however, is very
    much like the C Ltd. adapter that costs ~ $275.00.  They both use
    the NCR 8530 (second source AMD) and the minimal glue required to
    interface to the buss, i.e., address decoding, buss protocol, etc.
     The difference is the C Ltd. autoconfigure circutry, which is
    necessitated more by marketing considerations than anything else.
     It appears to consume about as much real estate as the rest of
    the circuit sans the 8530.  I imagine an AddMem and Alloc would
    accomplish thhe same task.  I might build one after I finish some
    of my other projects, like painting the house, craeting world peace,
    etc.  If anyone is interested please contact me.
    John

    re.: .11
    	I ststed that the SCSI chip was "8530"... I think its really
    a 5830; I'll check tonight.
    Sorry for any inconveience.
    John
    

Re: .11

>     The difference is the C Ltd. autoconfigure circutry, which is
>    necessitated more by marketing considerations than anything else.

I think autoconfigure in a drive also has other uses.  I believe that
you will only be able to boot directly from a hard drive if the hard
drive is autoconfig.   At the last BCS meeting, the guy from ASDG claimed
that only the Commodore drive, the Byte-by-byte drive (which uses the
Commodore controller), and the upcoming ASDG drive would be able to
boot the system.  That implies that autoconfig is necessary but not
sufficient in order to boot directly from the hard drive.

I am not sure that booting directly from the hard drive is that important
since a floppy can start the boot pass control to the hard drive by
giving a small number of commands.

    re.: .13
    	Yes Randy, that autoconfig is probably required for a hard disk
    boot is a good bet; I also agree that it alone is insufficient.
     But for my purposes right now I don't care ( I leave the system
    on always).  The time required to Kickstart and boot the Workbench
    is nothing compared to copying the c directory, etc. to a ramdisk,
    and any scsi controller will allow one to do this off the hd.
    John
    

    Randy, you're right about the boot times from floppy... the only
    time the DF0: floppy is accessed is to perform the first two commands
    in the startup-sequrnce; Mount and ASSIGN sys: DH0:
    
    From that point, all configuring is performed from the HD at warp
    speed.
    
    All I really care is that non-autoconfig drives will benefit from
    the 1.21 harddisk drivers, which I have been assured (from Supra)
    will be the case. 

    re: .15
    
    Well, semi-warp.  If the startup-sequence on df0: is more then one
    record, then the amiga will read the next record from df0:.
    Unfortunately this is much slower then reading from dh0:.
    
    I guess it is just easier to keep the startup-sequence less then
    2 records.
    
    Another problem is when a LoadWB is done, the computer asks for
    the disk it booted with until you feel like smashing it with a sledge
    hammer.  Oh well...
    
    Mike Leibow

    
    How about:
    
    mount dh0:
    loadwb
    run dh0:s/background-stuff
    endcli >nil:
    
    There's no reason to wait for a lot of the startup-sequence to
    get done on a multitasking machine.
    
    No harddrive for me yet.  I'm still waiting for that $400.00, 100
    Meg drive that's faster than RAM. :-)
    
    regards,
    
    steve mcafee 

    Has anyone called Microbotics to ask about the SCSI board for the
    Satrboard II ?  I've heard "realsoon now", etc., but no details.
     Like price.
    John

    
    Oops in .17 I meant
       dh0:c/run dh0:c/execute dh0:s/background-stuff
    not run...
    

    Wow, I really blew the part number!  If anyone cares, its 5380.
    This is one of a large family of SCSI chips that NCR makes and AMD
    second sources.  There is a 53C80 that is a CMOS version of the
    NMOS 5380, with more pins (extra grounds) and a few bug fixes. 
    Note that there is a simple workaround for every bug in the 5380
    that the CMOS part corrects, and the driver software would (should)
    know which one's installed in order to take advantage of it.  I
    don't believe that implementing the workarounds on the C part will
    hurt anything.  In any event, the CMOS part uses less power.  
    By the way, does anyone know where to purchase these?
    John.