Re: Announcement: New PL/I kit for Alpha

Tom Linden wrote:
On Wed, 07 Sep 2005 23:02:44 -0400, John W. Kennedy <jwkenne@xxxxxxxxxxxxx> wrote:

glen herrmannsfeldt wrote:

Tom Linden wrote:

A new kit is now available for download from our website,
which includes a complete rebuild of the GEM interface.


This fixes a cute bug. When compiling with /SHOW=STATISTICS
the compile speed is computed in lines per minute. Time is measured
in in units of a hundredth of a millisecond, which turned out to be
zero for a small program on an ES47. When this code was originally written
in 1978 for the VAX I guess it was never thought that computers would get
that fast! 

  The Hercules group working on S/370 emulators has found places
where OS designers didn't expect computers to be that fast.
One is in VM/370 where a tight loops is expected to take some number
of clock ticks.  When zero ticks occur during the loop a zerodivide
occurs.


On actual S/370 hardware, by definition, every instruction takes at least one clock tick. If the instructions go too fast, either the clock has to have a higher resolution (or at least append incrementing garbage bits below the last true ticking bit) or the instruction processor has to slow down. 


The issue is not the resolution of the clock but the manner and resolution
that was chosen in the program reporting the statistics.

I say again, the S/370 Principles of Operation /defines/ the clock as being faster than the instruction processor, by whatever means necessary -- i.e., unless the clock is reset, it must never store the same result twice. It is impossible, on 370 hardware, for "zero ticks [to] occur during the loop".
--
John W. Kennedy
"Never try to take over the international economy based on a radical feminist agenda if you're not sure your leader isn't a transvestite."
-- David Misch: "She-Spies", "While You Were Out"
.

Relevant Pages

  • Re: Announcement: New PL/I kit for Alpha
    ... where OS designers didn't expect computers to be that fast. ... One is in VM/370 where a tight loops is expected to take some number of clock ticks. ... If the instructions go too fast, either the clock has to have a higher resolution or the instruction processor has to slow down. ...
    (comp.lang.pl1)
  • Re: Announcement: New PL/I kit for Alpha
    ... where OS designers didn't expect computers to be that fast. ... One is in VM/370 where a tight loops is expected to take some number of clock ticks. ... If the instructions go too fast, either the clock has to have a higher resolution or the instruction processor has to slow down. ...
    (comp.lang.pl1)
  • Re: Variations on XTAL clock AND time synchronization
    ... >computers will have one GPS receiver and all the other stand alone computers ... but I do not know how) the probable variation that a computer clock may ... they're driven by separate quartz crystals. ... amount of clock error and keep a record of the amount of drift (in ...
    (comp.arch.embedded)
  • Re: Variations on XTAL clock AND time synchronization
    ... >computers will have one GPS receiver and all the other stand alone computers ... but I do not know how) the probable variation that a computer clock may ... they're driven by separate quartz crystals. ... amount of clock error and keep a record of the amount of drift (in ...
    (sci.electronics.misc)
  • Re: Variations on XTAL clock AND time synchronization
    ... >computers will have one GPS receiver and all the other stand alone computers ... but I do not know how) the probable variation that a computer clock may ... they're driven by separate quartz crystals. ... amount of clock error and keep a record of the amount of drift (in ...
    (sci.electronics.components)
Loading