 ABAQ

 Written by Perihelion, Ltd.


 Hardware Specification
 ----------------------

 The base machine outline specification is as follows:
 -----------------------------------------------------

 T800-20 Transputer 10MIPS, 1.5 Mflop
 Three 20Mhz links, buffered
 4Mbyte DRAM
 1 Mbyte dual-port video RAM
 Colour blitter
 True DMA SCSI port for 40M (minimum) hard disc
 Three internal expansion slots
 68000 Mega ST as I/O processor (plug in card connects fourth 20Mhz link)

 Screen Resolution and Use
 -------------------------

 The table below lists the screen resolutions and their probable typical
 use. All the following are at 60Hz with portrait orientation.

 Mode Resolution  Width          Description
 ---- ----------  -----          -----------

  0   1280 x 960  4 bits/pixel   4 bits/colour or monochrome 
                                 (Desk Top Publishing, engineering
                                 drawings)

  1   1024 x 768  8 bits/pixel   8 bits/colour
                                 (CAD, colour pictures, graphs)

  2    640 x 480  8 bits/pixel   8 bits/colour 2 screens
                                 (Animation)

  3    512 x 480  32 bits/pixel  24 bits colour, 1 overlay bit, 7 tag
                                 bits (True colour, smooth shading,
                                 3D modelling)

 The Blitter
 -----------

 The Perihelion blitter is based on work done by Dr Phil Willis of the 
 University of Bath. It provides meaningful operations with colour and 
 colour look-up tables (CLUTs) and implements very fast 2-D raster
 graphics operations, such as fast font drawing. It also provides a
 32-bit wide pipeline (with four tests on each of eight pixels
 concurrently), and is synchronised with blanking. Using the blitter,
 square area fill takes 128 megapixels per second, arbitrary two colour
 character drawing takes up to 64 megapixels per second, and full 2-D
 block copy takes 16 megapixels per second.

 Expansion Capability
 --------------------

 The Perihelion design provides for three expansion cards within the box.
 These can be memory cards, providing a maximum of 64Mbytes using 4M
 parts, or various versions of alternative graphics cards. The full
 transputer bus is brought out so any type of peripheral may be connected.

 The expansion sockets also bring out the transputer links and control
 signals. This means that cards containing extra transputers can be added,
 and the size of the cards allows for four transputers with up to 1Mbyte
 of RAM each on a single card. One workstation can therefore contain 13 
 processors.  Other link connections can be made outside the box to
 parallel processor farms of multiple processors. The link connections can
 also be made to fast peripherals such as a laser printer or disc server.

 The Transputer

 The T414 is a 32-bit processor that consists of a RISC style CPU,
 2K of fast on-chip RAM, an external memory interface and four serial
 links which may run at 5, 10 or 20 Mbits/second. The T800 is similar
 except that it also contains a floating point processor and 4K of RAM.

 The programmer's model consists of a three register evaluation stack, a
 workspace pointer and an instruction pointer.  A small number of
 instructions exist for loading and storing values on the stack and for
 altering the flow of control, the remainder operate on operands on the
 stack.

 The processor has microcoded support for processes at 2 priority levels.

 High priority processes may preempt low priority processes after any
 instruction and run until they give up the processor. High priority
 processes are essentially equivalent to interrupt routines on
 conventional processors.  Low priority processes are round-robin
 scheduled on a timesliced basis.  Timeslicing only occurs on particular
 instructions which are defined so that the minimum of state need be
 saved; process switching is therefore very fast.

 The transputer achieves inter-process communication through channels,
 which are single words of memory. Two processes that wish to communicate
 rendezvous at a channel and exchange data by copying from one buffer to
 another. As this is implemented by the microcode, the cost of copying
 lies only in the memory accesses for the data and not in instruction
 fetches. Communication is strictly one-to-one and channels may not be
 shared by more than one sender or receiver.  The inter-processor links
 are designed to behave exactly like channels, and are used with the same
 instructions.

 Parallel Programming
 --------------------

 The unique aspect of the Atari/Perihelion design is that is provides
 multiple processors within a single workstation.  The use of multiple
 processors means that is is possible to write application programs which
 make use of the possible parallelism inherent in such systems.

 Application programs can run under Helios using three programming
 philosophies.  The first of these is the traditional programming model.
 A program can be taken from another environment, such as Unix or a PC,
 and with little or no change converted to run under Helios. C and the
 Unix C library is provided, and such programs will run as a single
 process in the machine.

 Other programs, again probably from Unix, will run in several sections
 all of which may be run in different processes and connected by pipes.

 Helios encourages the use of many small programs which work together to
 create a final product. A common example is a pre-processor, a compiler
 front end, a compiler back end, an assembler and a linker. These can all
 be run together with intermediate connections made by pipes.

 Under other operating systems the different processes are timesliced on
 the one single processor.  Under Helios these different processes can be
 allocated to different processors, so that the individual parts actually
 run at the same time. 

 This type of "per-process" parallelism is easily understood, and many
 applications are already in this form. Examples include a word processor
 with background spooling and spelling checking or background jobs such
 as message systems or archiving. If an application is being altered then
 the use of extra processes should be kept in mind.

 The final way in which parallelism may be exploited is by the use of
 parallel algorithms. These tend to be hard to find for programmers used
 to the sequential nature of normal computers, but a look at the real
 world shows, of course, everything running in parallel.

 Applications using parallel algorithms will normally be written from
 scratch with such ideas in mind.  The benefit is that such programs will
 run much faster when the user provides more power in the form of more
 processors. Many examples of parallel algorithms exist, such as ray
 tracing, spreadsheet calculations, even computer chess!

 Helios Overview
 ---------------

 Helios is a true distributed operating system; there are no central
 services upon which the whole system relies.  This results in increased
 system reliability since the failure of any processor, or the
 partitioning of the network, will not cause unrelated parts of the
 system to fail (although they may continue at a somewhat reduced
 capacity).  The distributed nature of Helios is transparent both to the
 user at his terminal and to programs running within it which need never
 be aware of the exact location of any services.  This feature
 differentiates it from a network operating system where the distributed
 nature is more explicit.

 Helios is intended to be an open system architecture in which parts may
 be added, removed, modified or replaced transparently to suit specific
 purposes.  In many ways Helios is simply a set of conventions, or codes
 of practice, for the behaviour of programs.  It may be thought of as a
 "software backplane" providing an infrastructure for processes to locate
 and communicate with each other. 

 Finally, the emphasis throughout the development of Helios has been on
 finding practical solutions to the problems of distributed computing.

 For this reason many of its features are not new but have been derived
 from existing research systems.  The two most important influences have
 been the Cambridge Distributed Operating System and another system
 called Amoeba.

 fin

