Section 11: HARDWARE

On which platforms does IRAF run?

 IRAF currently runs on the following platforms:
        Platform                Descrition
        AIX4                    IBM/RS6000 AIX 4.x</td>
        DUNX                    Digital Unix 4.0 (OSF, Compaq Tru64)
        HPUX                    HP-UX 10.20
        IRIX                    SGI IRIX 6.x
        PCIX                    FreeBSD 3.3
        PCIX                    Slackware Linux 4.0
        PCIX                    RedHat Linux 5.x
        PCIX                    RedHat Linux 6.x
        PCIX                    SuSE Linux 6.x
        PCIX                    Solaris 7 for Intel
        SSOL                    SunOS 4.x
        SSOL                    Solaris 5.5, 5.6, 5.7
        VMS7                    VAX/VMS 7.1
Plans for supporting LinuxPPC on a Macintosh are also in the works.

What's the best machine to buy if I plan to run IRAF?

	Because of the different budgets people have available, the deals
they can get from vendors for multiple purchases, local system management
support, the number of expected users, and the extent of non-IRAF use the
machine might get, it is impossible for the IRAF Project to recommend a
specific platform to users considering a purchase.  If there is some doubt
about whether IRAF is supported for a particular machine users should contact
the IRAF group with any questions, we would also be happy to answer any
questions about specific devices or configurations.

What are the recommended hardware requirements of a workstation?

	Hardware requirements depend largely on the number of expected users,
the types of reductions to be done, and what additional software will be
used.  A typical single-user workstation requires a minimum of 32Mb RAM
(2-3 times that in swap space), enough disk for the OS, X windows, the IRAF
system and sufficient room for data.  Some type of tape drive is also des-
ireable.  Servers or multiple-user systems will typically have more RAM and
disk, graphics accelerator cards are not required but may, in some cases,
speed up window performance.

What type of mag tape devices are supported and not supported?

	The dev$tapecap file can be used to configure most tape devices for
use with IRAF.  Current entries include Exabyte, DAT, 1/2" reel, QIC cart-
ridge tapes, Mac DC2000 cartridge tape, and Mac FDHD floppy disks.  There
are currently no entries for other types of floppy drives or optical devices,
but it's possible entries could be written for these.
	The distributed tapecap file is missing entries for DAT devices using
the native Sun ST driver, contact the IRAF group for information on how this
may be installed.

Should I buy a 24-bit frame buffer for my Sun?

	IRAF doesn't currently take advantage of 24-bit frame buffers, though
several experiments with 24-bit display have been done.  While IRAF may not
use such a card, graphics accelerators on some such cards may speed up window
performance.  Users considering such a card should make purchase decisions
based on factors other than expected IRAF use.

Can I use my Windows 9x PC to display IRAF images?

	This is possible provided you have an X server running on the PC
(several are available for PCs).  If displaying from a remote machine,
typical usage would be to run the display server (SAOimage or XImtool) on
the IRAF host system, setting the '-display' command line flag or 'DISPLAY'
environment variable to display the window on the PC.  Since the server is
running on the same node as IRAF there is no need to set the IRAF 'node'
environment variable (or it should be set to the name of the IRAF host).
Care should be taken to reset the number of colors for the display to use
only 8-bits (i.e. 256 colors) in order for the display servers to work
properly.

Do you have benchmarks for IRAF on various machines?

	A complete set of benchmarks for IRAF has not been done since 1987,
the results of those are obsolete now anyway.  While some machines are gen-
erally faster than others, benchmark results often depend on the task being
tested, disk access times, available memory and a variety of other factors.
Proper benchmarks should fairly use both I/O-bound and CPU-bound tasks in
testing, as well as multiple image formats and data types, in order to draw
any objective conclusions about relative performance.  Head-to-head comparisons
between machines may not be of much interest to certain users if their usage
is much different that that being tested.
	The 1987 benchmark results are available in the iraf archive as
iraf.noao.edu:iraf/docs/bench.ps.Z.