Remote job submission from PDP-11

[jim stephens]

On 10/7/2020 10:32 AM, Ethan Dicks via cctalk wrote:
> On Wed, Oct 7, 2020 at 12:07 PM Glen Slick via cctalk
> <cctalk at classiccmp.org> wrote:
>> On Wed, Oct 7, 2020 at 8:37 AM Ethan Dicks wrote:
>>> On Wed, Oct 7, 2020 at 5:07 AM Peter Allan wrote:
>>>> RJE/HASP
>>>> 2780/3780 Protocol Emulator
>>> I used to do this with specialty hardware...
>>>
>>> http://www.bitsavers.org/pdf/softwareResults/Software_Results_Comboard_Brochure_1983.pdf
>>>
>>> There were also intelligent sync serial engines for this from other
>>> vendors, including DEC (KMC11, and probably a later one but the device
>>> name escapes me).
>> I was curious about this DEC M8704 DMS11-DA that sold cheap a few days
>> ago. It has eight SMC COM5025 "Multi-Protocol Universal Synchronous
>> Receiver/Transmitter USYNR/T" chips:
>>
>> https://www.ebay.com/itm/373243388363
>>
>> Apparently it can't do anything on its own. It needs to be connected
>> to a UNIBUS through a companion KMC11 processor board, which might not
>> be too common if someone wanted to put together a working
>> configuration.
> I have very dim memories of that arrangement - it was definitely
> ordinary that you had a dedicated I/O processor to run multiple lines.
>
> Part of the salable benefit of the COMBOARD was having an 8MHz 68000
> manage the actual protocol - it moved block file data to/from the DEC
> host via DMA.  The M68K wrapped it up as needed, handled
> re-transmission, etc.
>
> Using a KMC11 is similar, IIRC, certainly at the bottom levels.
>
> -ethan
>
>
> -ethan
>
The protocol HASP used was unique because they had a novel BSC protocol 
handshake to reduce delays on the link due to handshake latency.

The normal protocol had an ack requirement that involved dead time on 
the link till the ACK had returned.  The HASP workstation protocol sent 
an ack on the front end  of another record if the prior record came and 
streamed data with NAKs as needed stopping and causing handshake 
delays.  A good link was much better used than a usual BSC link.

Also if you were printing an reading job submission data it got an 
almost full duplex use of the link speed going as the acks for prior 
records were prefaced on the front of the next data, rather than a 
standalone turn of the link just for ack.

Hasp had a term for their data protocol that's not come back to me.

The link between Hercules and whatever simulator's BSC would need to be 
able to handle that bit of trickery.

Thanks
Jim