Re: [AArch64] Fix integer vabs intrinsics

2014-05-07 Thread Richard Earnshaw
On 07/05/14 11:32, Richard Biener wrote: > On Wed, May 7, 2014 at 12:30 PM, Richard Earnshaw wrote: >> On 05/05/14 09:04, Richard Biener wrote: >>> On Fri, May 2, 2014 at 12:39 PM, Richard Earnshaw wrote: On 02/05/14 11:28, James Greenhalgh wrote: > On Fri, May 02, 2014 at 10:29:06AM +01

Re: [AArch64] Fix integer vabs intrinsics

2014-05-07 Thread Richard Biener
On Wed, May 7, 2014 at 12:30 PM, Richard Earnshaw wrote: > On 05/05/14 09:04, Richard Biener wrote: >> On Fri, May 2, 2014 at 12:39 PM, Richard Earnshaw wrote: >>> On 02/05/14 11:28, James Greenhalgh wrote: On Fri, May 02, 2014 at 10:29:06AM +0100, pins...@gmail.com wrote: > > >>

Re: [AArch64] Fix integer vabs intrinsics

2014-05-07 Thread Richard Earnshaw
On 05/05/14 09:04, Richard Biener wrote: > On Fri, May 2, 2014 at 12:39 PM, Richard Earnshaw wrote: >> On 02/05/14 11:28, James Greenhalgh wrote: >>> On Fri, May 02, 2014 at 10:29:06AM +0100, pins...@gmail.com wrote: > On May 2, 2014, at 2:21 AM, James Greenhalgh > wrote: >

Re: [AArch64] Fix integer vabs intrinsics

2014-05-05 Thread Richard Biener
On Fri, May 2, 2014 at 12:39 PM, Richard Earnshaw wrote: > On 02/05/14 11:28, James Greenhalgh wrote: >> On Fri, May 02, 2014 at 10:29:06AM +0100, pins...@gmail.com wrote: >>> >>> On May 2, 2014, at 2:21 AM, James Greenhalgh wrote: > On Fri, May 02, 2014 at 10:00:15AM +0100, A

Re: [AArch64] Fix integer vabs intrinsics

2014-05-02 Thread Richard Earnshaw
On 02/05/14 11:28, James Greenhalgh wrote: > On Fri, May 02, 2014 at 10:29:06AM +0100, pins...@gmail.com wrote: >> >> >>> On May 2, 2014, at 2:21 AM, James Greenhalgh >>> wrote: >>> On Fri, May 02, 2014 at 10:00:15AM +0100, Andrew Pinski wrote: On Fri, May 2, 2014 at 1:48 AM, James Gree

Re: [AArch64] Fix integer vabs intrinsics

2014-05-02 Thread James Greenhalgh
On Fri, May 02, 2014 at 10:29:06AM +0100, pins...@gmail.com wrote: > > > > On May 2, 2014, at 2:21 AM, James Greenhalgh > > wrote: > > > >> On Fri, May 02, 2014 at 10:00:15AM +0100, Andrew Pinski wrote: > >> On Fri, May 2, 2014 at 1:48 AM, James Greenhalgh > >> wrote: > >>> > >>> Hi, > >>>

Re: [AArch64] Fix integer vabs intrinsics

2014-05-02 Thread pinskia
> On May 2, 2014, at 2:21 AM, James Greenhalgh wrote: > >> On Fri, May 02, 2014 at 10:00:15AM +0100, Andrew Pinski wrote: >> On Fri, May 2, 2014 at 1:48 AM, James Greenhalgh >> wrote: >>> >>> Hi, >>> >>> Unlike the mid-end's concept of an ABS_EXPR, which treats overflow as >>> undefined/impo

Re: [AArch64] Fix integer vabs intrinsics

2014-05-02 Thread James Greenhalgh
On Fri, May 02, 2014 at 10:00:15AM +0100, Andrew Pinski wrote: > On Fri, May 2, 2014 at 1:48 AM, James Greenhalgh > wrote: > > > > Hi, > > > > Unlike the mid-end's concept of an ABS_EXPR, which treats overflow as > > undefined/impossible, the neon intrinsics vabs intrinsics should behave as > > th

Re: [AArch64] Fix integer vabs intrinsics

2014-05-02 Thread Andrew Pinski
On Fri, May 2, 2014 at 1:48 AM, James Greenhalgh wrote: > > Hi, > > Unlike the mid-end's concept of an ABS_EXPR, which treats overflow as > undefined/impossible, the neon intrinsics vabs intrinsics should behave as > the hardware. That is to say, the pseudo-code sequence: Only for signed integer

[AArch64] Fix integer vabs intrinsics

2014-05-02 Thread James Greenhalgh
Hi, Unlike the mid-end's concept of an ABS_EXPR, which treats overflow as undefined/impossible, the neon intrinsics vabs intrinsics should behave as the hardware. That is to say, the pseudo-code sequence: a = vabs_s8 (vdup_n_s8 (-128)); assert (a >= 0); does not hold. As in hardware abs