3020 Floating Point Coprocessor Failure

[Tony Facca]

Subject: 3020 Floating Point Coprocessor Failure

>Has anyone experienced a failure of the floating point coprocessor on an
>IRIS 3000-series?  Ours suddenly stopped working, and now all programs compiled
>with the -Zf option do not work.  It has been getting a little warm in the
>computer room because the air conditioner wasn't working correctly -- could
>this have something to do with it?  Any info would be appreciated.
>
Well as a matter of fact, even as we "speak", I have the service guys sitting
in the computer room swapping out our floating point board.  It's the first
one to go in a couple of years, and we just rolled the machine down the hall
to a new room (which is about 20 degrees cooler).  But, it's a simple board
swap (at least it was when SGI had the service contract), now it's still a
simple board swap, it just takes a lot longer for the service guys to believe
me.. :-)

Try this program (f77 -Zg fpu.f -o fpu).  If the answers are wrong, get a new
board.

--------------------------------- fpu.f --------------------------------------
	program fpu

c	*** test the floating point processor board 
c	*** compile program:  f77 -Zg fpu.f -o fpu

c	*** these functions invoke the floating point accellerator.  If the
c	*** board is bad, the results will be incorrect, but the integers
c	*** will be correct.

	do 100 i = 1, 20

	   x = float (i)
	   cosx = cos(x)
	   sinx = sin(x)

	   print *,'integer: ',i,'  x= ',x,' cosx= ',cosx,' sinx= ',sinx 

  100   continue

	stop '  ** end test **'
	end
-------------------------------------------------------------------------------

--
..ahead, warp factor...two 
						+                   *	    +
							*        
								+      * 
- - - - -------======<<<<<{{{{{{[[[[[[ TONY FACCA  fsfacca at avelon.lerc.nasa.gov
						 +                    +
          					      *       *           *