[libre-riscv-dev] [isa-dev] Re: FP transcendentals (trigonometry, root/exp/log) proposal
hendrik at topoi.pooq.com
Thu Aug 8 19:47:54 BST 2019
On Thu, Aug 08, 2019 at 06:51:42AM -0700, Andrew Waterman wrote:
> On Thu, Aug 8, 2019 at 6:27 AM lkcl <luke.leighton at gmail.com> wrote:
> > On Thursday, August 8, 2019 at 8:56:44 PM UTC+8, lkcl wrote:
> > > What I do not know is whether there was something unique about the 370
> > mainframe and applications being written for it, or, if now in 2019, this
> > is sufficiently well understood such that all FP applications writers have
> > properly taken *better* accuracy (not worse accuracy: *better* accuracy)
> > into consideration in the design of their programs.
> > I *think* this is what Andrew might have been trying to get across.
> That story might have roots in IBM's alternate base-16 floating-point
> format, but in any case, that wasn't the point I was trying to make. I
> stand by my first message to this thread.
Amdahl used the same base-16 floating point. Amdahl just gave the more
accurate result on division.
More information about the libre-riscv-dev