This is the mail archive of the
mailing list for the GCC project.
Re: [autovect] [patch] vectorize reduction without specialized target support
- From: Richard Henderson <rth at redhat dot com>
- To: Dorit Naishlos <DORIT at il dot ibm dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Wed, 15 Jun 2005 13:22:28 -0700
- Subject: Re: [autovect] [patch] vectorize reduction without specialized target support
- References: <OF103D3DC4.C6903379-ONC2257021.003D9471-C2257021.006E6758@il.ibm.com>
On Wed, Jun 15, 2005 at 11:05:54PM +0300, Dorit Naishlos wrote:
> This scheme requires adding new tree-codes for the vector-shifts. I
> introduced the simplest vector-shift possible - the shift amount is in
> bytes, and is a constant (immediate value). This has to potential to be
> easily expanded to the most efficient code, and be applicable to as many
> targets as possible.
One change I'd prefer is that the shift count remain in bits, as with
any other shift that gcc supports. I think that's less confusing.
(Despite the fact that both SSE and Altivec actually implement that
shift by encoding the number of bytes into the instruction.)