This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH][ARM] Remove movdi_vfp_cortexa8
- From: Wilco Dijkstra <Wilco dot Dijkstra at arm dot com>
- To: Ramana Radhakrishnan <ramana dot gcc at googlemail dot com>
- Cc: Kyrill Tkachov <kyrylo dot tkachov at foss dot arm dot com>, GCC Patches <gcc-patches at gcc dot gnu dot org>, nd <nd at arm dot com>
- Date: Tue, 17 Jan 2017 12:10:52 +0000
- Subject: Re: [PATCH][ARM] Remove movdi_vfp_cortexa8
- Authentication-results: sourceware.org; auth=none
- Authentication-results: spf=none (sender IP is ) smtp.mailfrom=Wilco dot Dijkstra at arm dot com;
- Nodisclaimer: True
- References: <VI1PR0802MB262138D6011F3C17FCD7B139838D0@VI1PR0802MB2621.eurprd08.prod.outlook.com> <VI1PR0802MB2621E445CDAA4211DAFC9E6283820@VI1PR0802MB2621.eurprd08.prod.outlook.com> <AM5PR0802MB2610989BCB172720719E3F69839A0@AM5PR0802MB2610.eurprd08.prod.outlook.com> <585176AB.4010006@foss.arm.com> <AM5PR0802MB2610BA9D4EEAF7BB81A4FA26839A0@AM5PR0802MB2610.eurprd08.prod.outlook.com> <AM5PR0802MB2610A534E2658C678530F80D839A0@AM5PR0802MB2610.eurprd08.prod.outlook.com>,<CAJA7tRb=r1ydEF_DHTJ-1QYpMHxDdSgwGEMf=S9LyOr1oQ9S5g@mail.gmail.com>,<AM5PR0802MB2610BCF3D540291CF4DA656F83910@AM5PR0802MB2610.eurprd08.prod.outlook.com>
- Spamdiagnosticmetadata: NSPM
- Spamdiagnosticoutput: 1:99
Wilco Dijkstra wrote:
> Ramana Radhakrishnan wrote:
>> On Wed, Dec 14, 2016 at 5:43 PM, Wilco Dijkstra <Wilco.Dijkstra@arm.com> wrote:
>
> > > Yes, the reason to split the pattern was to introduce the '!' to discourage Neon->int moves on Cortex-A8
> (https://patches.linaro.org/patch/541/). I am not removing the optimization for Cortex-A8, however
> I haven't been able to find an example where it makes a difference, even on high register pressure code.
> >
> > even on crafty with -mtune=cortex-a8 ?
>
> Indeed the '!' makes no difference on crafty either with -mcpu=cortex-a8 -mfpu=neon.
> Even if it did make a difference in the past, it is either totally ignored by the register
> allocator or has no useful effect on deciding whether to use a Neon or integer register.
Any comments?
Wilco