[llvm-dev] [FPEnv] Rust/Go/Swift/Flang/other llvm IRBuilder needs?

John Reagan via llvm-dev llvm-dev at lists.llvm.org
Thu Jan 17 11:23:53 PST 2019


We are certainly interested.  I've been following along all the
discussions about constrained math, FENV_ACCESS, etc. and it all seems
quite reasonable to me.

We are needing to handle the IEEE settings we have on our Alpha and
Itanium targets.  We have various C and Fortran command line settings like:

CC

  /IEEE_MODE

        /IEEE_MODE=option
        /IEEE_MODE=DENORM_RESULTS (D)

     Selects the IEEE floating-point mode to be used.

     Options:

      FAST              During program execution, only finite values (no
                        infinities, NaNs, or denorms) are created.
                        Exceptional conditions, such as floating point
                        overflow and divide by zero, are fatal.

      UNDERFLOW_TO_ZERO Generate infinities and NaNs.  Flush
                        denormalized results and underflow to zero
                        without exceptions.

      DENORM_RESULTS    Same as UNDERFLOW_TO_ZERO, except that denorms
                        are generated.  This is the default for I64
                        systems.

      INEXACT           Same as DENORM_RESULTS, except that inexact
                        values are trapped.  This is the slowest mode.


CC

  /ROUNDING_MODE

        /ROUNDING_MODE=option
        /ROUNDING_MODE=NEAREST (D)

     For /FLOAT=IEEE_MODE, the /ROUNDING_MODE qualifier lets you select
     one of the following IEEE rounding modes:

      o  NEAREST (default)

      o  DYNAMIC

      o  MINUS_INFINITY

      o  CHOPPED


     For /FLOAT=G_FLOAT or /FLOAT=D_FLOAT, rounding defaults to
     /ROUNDING_MODE=NEAREST, with no other choice of rounding mode.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20190117/11c1aa0a/attachment.sig>


More information about the llvm-dev mailing list