This is the mail archive of the
libc-alpha@sources.redhat.com
mailing list for the glibc project.
Re: [PATCH] PPC64 Use hardware sqrt.
- From: Paul Mackerras <paulus at samba dot org>
- To: Tom Gall <tom_gall at vnet dot ibm dot com>
- Cc: Kumar Gala <kumar dot gala at freescale dot com>,Steve Munroe <sjmunroe%ibmus dot vnet at RCHGATE dot RCHLAND dot IBM dot COM>,dgm69%ibmus dot vnet at RCHGATE dot RCHLAND dot IBM dot COM,janis187%ibmus dot vnet at RCHGATE dot RCHLAND dot IBM dot COM,libc-alpha at sources dot redhat dot com, Kumar Gala <kumar dot gala at motorola dot com>
- Date: Sat, 1 May 2004 16:46:36 +1000
- Subject: Re: [PATCH] PPC64 Use hardware sqrt.
- References: <OF06703AF6.4BE351BE-ON86256E86.0056D522-86256E86.0057EE3A@us.ibm.com><EF12675B-9B16-11D8-B348-000393DBC2E8@freescale.com><4093109E.6080402@vnet.ibm.com>
Tom Gall writes:
> Kumar Gala wrote:
>
> > Understand, but future PPC64 from Freescale (former Motorola SPS) are
> > unlikely to implement it.
>
> If they do then likely they will also leave the other optional floating
> point operations unimplemented and in that case the correct way to build
> glibc is with the --nf option which is already the case for PPC32
> processors without floating point.
Hmmm, the basic floating-point instructions aren't optional in the
PowerPC architecture AFAIK, but fsqrt, fsqrts, fres, frsqrte, fsel
(and maybe 1 or 2 others) are. All the 32-bit "desktop" CPUs from
Motorola (601, 604, 750, 74xx) had floating-point but didn't implement
fsqrt or fsqrts.
Since Kumar works for Freescale, if he says Freescale is likely to
implement 64-bit PowerPC cpus with the basic floating-point
instructions but without fsqrt[s], I'd be inclined to believe him.
(I just hope they don't do 64-bit book E... :)
Maybe we will end up needing another bit in the CPU features word in
the aux vector.
Regards,
Paul.