Looking at z/POP, there's a note that if the AFP-register-control bit (bit
45 of CR0) is zero, you'll get a data exception - but the z/POP says that
you should always get a 2 in byte 2 of the FPC on data exceptions.

Could be something CMS-y - have you tried it on z/OS?

Best regards,
Ray
(ex-SAG)

--
M. Ray Mullins - Software Developer
CIMS Lab, Inc.
Roseville, CA, USA


> -----Original Message-----
> From: Ferguson, Neale [mailto:[EMAIL PROTECTED]
> Sent: Tuesday 27 January 2004 15:00
> To: [EMAIL PROTECTED]
> Subject: BFP
>
>
> There's something fundamental that I don't understand about
> Binary Floating Point. I wrote a simple program:
>
> TEST  START
>         USING TEST,R12
>         STM     R14,R12,12(R13)
>         LR    R15,R12
>         LHI   R1,X'FC0'
>       LD    1,A
>       TCDB  1,0(1)
>       SLR   15,15
>       BR    14
> A     DC    XL8'4008000000000000' (3 in BFP)
>       END
>
> The code tests that 'A' is a BFP number (not infinite, not
> "not a number" etc.).
>
> Running this on CMS I get a data exception. Running the equivalent
> bit of code on Linux (using _asm_ (...)), it works without a problem.
>
> Are there some setup things I need to do in order to use BFP? The
> FPC register is all zeroes.
>
> Neale
>

Reply via email to