Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-07-15 Thread Richard Biener
On Tue, 12 Jul 2016, Richard Biener wrote: > On Tue, 12 Jul 2016, Uros Bizjak wrote: > > > On Tue, Jul 12, 2016 at 10:58 AM, Richard Biener wrote: > > > On Sun, 10 Jul 2016, Uros Bizjak wrote: > > > > > >> On Wed, Jul 6, 2016 at 3:18 PM, Richard Biener wrote: > > >> > > >> >> > 2016-07-04 Rich

Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-07-13 Thread Richard Biener
On Wed, 13 Jul 2016, Uros Bizjak wrote: > On Sun, Jul 10, 2016 at 10:12 AM, Uros Bizjak wrote: > > On Wed, Jul 6, 2016 at 3:18 PM, Richard Biener wrote: > > > >>> > 2016-07-04 Richard Biener > >>> > > >>> > PR rtl-optimization/68961 > >>> > * fwprop.c (propagate_rtx): Allow SUBREGs of

Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-07-12 Thread Uros Bizjak
On Sun, Jul 10, 2016 at 10:12 AM, Uros Bizjak wrote: > On Wed, Jul 6, 2016 at 3:18 PM, Richard Biener wrote: > >>> > 2016-07-04 Richard Biener >>> > >>> > PR rtl-optimization/68961 >>> > * fwprop.c (propagate_rtx): Allow SUBREGs of VEC_CONCAT and CONCAT >>> > to simplify to a non-c

Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-07-12 Thread Richard Biener
On Tue, 12 Jul 2016, Uros Bizjak wrote: > On Tue, Jul 12, 2016 at 10:58 AM, Richard Biener wrote: > > On Sun, 10 Jul 2016, Uros Bizjak wrote: > > > >> On Wed, Jul 6, 2016 at 3:18 PM, Richard Biener wrote: > >> > >> >> > 2016-07-04 Richard Biener > >> >> > > >> >> > PR rtl-optimization/689

Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-07-12 Thread Uros Bizjak
On Tue, Jul 12, 2016 at 10:58 AM, Richard Biener wrote: > On Sun, 10 Jul 2016, Uros Bizjak wrote: > >> On Wed, Jul 6, 2016 at 3:18 PM, Richard Biener wrote: >> >> >> > 2016-07-04 Richard Biener >> >> > >> >> > PR rtl-optimization/68961 >> >> > * fwprop.c (propagate_rtx): Allow SUBREGs

Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-07-12 Thread Richard Biener
On Sun, 10 Jul 2016, Uros Bizjak wrote: > On Wed, Jul 6, 2016 at 3:18 PM, Richard Biener wrote: > > >> > 2016-07-04 Richard Biener > >> > > >> > PR rtl-optimization/68961 > >> > * fwprop.c (propagate_rtx): Allow SUBREGs of VEC_CONCAT and CONCAT > >> > to simplify to a non-constant

Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-07-10 Thread Uros Bizjak
On Wed, Jul 6, 2016 at 3:18 PM, Richard Biener wrote: >> > 2016-07-04 Richard Biener >> > >> > PR rtl-optimization/68961 >> > * fwprop.c (propagate_rtx): Allow SUBREGs of VEC_CONCAT and CONCAT >> > to simplify to a non-constant. >> > >> > * gcc.target/i386/pr68961.c: New testca

Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-07-06 Thread Richard Biener
On Tue, 5 Jul 2016, Richard Sandiford wrote: > Richard Biener writes: > > On Sun, 3 Jul 2016, Richard Sandiford wrote: > > > >> Richard Biener writes: > >> > On Wed, 15 Jun 2016, Richard Sandiford wrote: > >> > > >> >> Richard Biener writes: > >> >> > With the proposed cost change for vector co

Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-07-05 Thread Richard Sandiford
Richard Biener writes: > On Sun, 3 Jul 2016, Richard Sandiford wrote: > >> Richard Biener writes: >> > On Wed, 15 Jun 2016, Richard Sandiford wrote: >> > >> >> Richard Biener writes: >> >> > With the proposed cost change for vector construction we will end up >> >> > vectorizing the testcase in

Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-07-04 Thread Richard Biener
On Sun, 3 Jul 2016, Richard Sandiford wrote: > Richard Biener writes: > > On Wed, 15 Jun 2016, Richard Sandiford wrote: > > > >> Richard Biener writes: > >> > With the proposed cost change for vector construction we will end up > >> > vectorizing the testcase in PR68961 again (on x86_64 and like

Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-07-03 Thread Richard Sandiford
Richard Biener writes: > On Wed, 15 Jun 2016, Richard Sandiford wrote: > >> Richard Biener writes: >> > With the proposed cost change for vector construction we will end up >> > vectorizing the testcase in PR68961 again (on x86_64 and likely >> > on ppc64le as well after that target gets adjustme

Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-06-29 Thread Richard Biener
On Mon, 27 Jun 2016, Richard Biener wrote: > On Wed, 15 Jun 2016, Richard Sandiford wrote: > > > Richard Biener writes: > > > With the proposed cost change for vector construction we will end up > > > vectorizing the testcase in PR68961 again (on x86_64 and likely > > > on ppc64le as well after

Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-06-27 Thread Richard Biener
On Wed, 15 Jun 2016, Richard Sandiford wrote: > Richard Biener writes: > > With the proposed cost change for vector construction we will end up > > vectorizing the testcase in PR68961 again (on x86_64 and likely > > on ppc64le as well after that target gets adjustments). Currently > > we can't o

RFC: 2->2 combine patch (was: Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961))

2016-06-16 Thread Segher Boessenkool
On Fri, Jun 10, 2016 at 11:20:22AM +0200, Richard Biener wrote: > With the proposed cost change for vector construction we will end up > vectorizing the testcase in PR68961 again (on x86_64 and likely > on ppc64le as well after that target gets adjustments). Currently > we can't optimize that away

Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-06-15 Thread Richard Sandiford
Richard Biener writes: > With the proposed cost change for vector construction we will end up > vectorizing the testcase in PR68961 again (on x86_64 and likely > on ppc64le as well after that target gets adjustments). Currently > we can't optimize that away again noticing the direct overlap of >

Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-06-13 Thread Richard Biener
On Mon, 13 Jun 2016, Richard Biener wrote: > On Fri, 10 Jun 2016, Richard Biener wrote: > > > > > With the proposed cost change for vector construction we will end up > > vectorizing the testcase in PR68961 again (on x86_64 and likely > > on ppc64le as well after that target gets adjustments).

Re: [PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-06-13 Thread Richard Biener
On Fri, 10 Jun 2016, Richard Biener wrote: > > With the proposed cost change for vector construction we will end up > vectorizing the testcase in PR68961 again (on x86_64 and likely > on ppc64le as well after that target gets adjustments). Currently > we can't optimize that away again noticing t

[PATCH] Allow fwprop to undo vectorization harm (PR68961)

2016-06-10 Thread Richard Biener
With the proposed cost change for vector construction we will end up vectorizing the testcase in PR68961 again (on x86_64 and likely on ppc64le as well after that target gets adjustments). Currently we can't optimize that away again noticing the direct overlap of argument and return registers. T