ISTR that Hasp used various "interleave" factors to write logical
blocks 1,2,3 et cetera as physical blocks 1,3,5.. or 1,4,7.. so that
in the time that it took to traverse the intervening trackspace the
Problem Program (such an appropriate term) had the opportunity to
generate another block of output lines. If the interleave factor was
severe enough then on, say, a five block track, logical blocks
1,2,3,4,5 might be written as 1,5,4,3,2 (giving a three-block
latency between each logical block) so that it would indeed appear
that they were being "written backwards". Of course these techniques
become less important as the multiprogramming level rises and many
p/programs "compete" for the spool disk write head.
IA(lso)STR that early PC hard disks (perhaps I should say "Fixed
disks" on IBM Main) used interleave extensively for the very same
performance-directed reason.
Perhaps today "interleave" is taken for granted, or perhaps is
irrelevant when whole "tracks" are cached in ram buffers that are
effectively part of the disk device.
It's interesting that we appear to be approaching a time when "disks"
in just about every computing device on the planet will become
solid-state devices, while shades of the original 90MB Storage Tek
device (late 1970s?) hover indistinctly in the twilight.
Curiously, interleave was/is? used in ram storage devices too, again
to mitigate the performance impacts of the unavoidable latencies in
the materials that real devices are made of. Even the preferential
ordering of instructions (usually by compilers) to avoid pipeline
stalls is a form of interleave.
Today, while waiting on the 'phone for a customer to check a library
for something, I filled and switched the electric kettle on,
interleaving two tasks, using the latencies in each to do something
useful in the other. Women, generally, are better trained at this
multitasking, IMO :-) .
Graeme.
At 08:54 PM 18/10/2005, you wrote:
>It was fun to wrote the code to allow the DASD records to be read in
reverse order
>(IOW: If there were 5 records on the current track, read them as 5, 4, 3,
2, 1
>before going to the prior track).
HASP accessed its SPOOL data this way until users began to complain that
they had no program that would back up/restore SPOOl volumes
to tape. Then the
HASP team made this record alternation an option. The thought was that
accessing every other record in sequence would provide a little boost in
performance. The same technique was (and maybe still is) used in
CMS files. HASP's
<snips>
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
