Re: [PATCH][i386] Fix vec_construct cost, remove unused ix86_vec_cost arg

2018-10-18 Thread Richard Biener
On Thu, 18 Oct 2018, Jan Hubicka wrote: > > > > So like the following which removes the use of ix86_vec_cost > > for SSE loads and stores since we have per-mode costs already. > > I've applied the relevant factor to the individual cost tables > > (noting that for X86_TUNE_SSE_SPLIT_REGS we only

Re: [PATCH][i386] Fix vec_construct cost, remove unused ix86_vec_cost arg

2018-10-18 Thread Jan Hubicka
> > So like the following which removes the use of ix86_vec_cost > for SSE loads and stores since we have per-mode costs already. > I've applied the relevant factor to the individual cost tables > (noting that for X86_TUNE_SSE_SPLIT_REGS we only apply the > multiplication for size == 128, not

Re: [PATCH][i386] Fix vec_construct cost, remove unused ix86_vec_cost arg

2018-10-18 Thread Jan Hubicka
> > The following fixes vec_construct cost calculation to properly consider > that the inserts will happen to SSE regs thus forgo the multiplication > done in ix86_vec_cost which is passed the wrong mode. This gets rid of > the only call passing false to ix86_vec_cost (so consider the patch >

Re: [PATCH][i386] Fix vec_construct cost, remove unused ix86_vec_cost arg

2018-10-15 Thread Richard Biener
On Thu, 11 Oct 2018, Richard Biener wrote: > > The following fixes vec_construct cost calculation to properly consider > that the inserts will happen to SSE regs thus forgo the multiplication > done in ix86_vec_cost which is passed the wrong mode. This gets rid of > the only call passing false

Re: [PATCH][i386] Fix vec_construct cost, remove unused ix86_vec_cost arg

2018-10-15 Thread Richard Biener
On Thu, 11 Oct 2018, Richard Biener wrote: > On Thu, 11 Oct 2018, Richard Biener wrote: > > > > > The following fixes vec_construct cost calculation to properly consider > > that the inserts will happen to SSE regs thus forgo the multiplication > > done in ix86_vec_cost which is passed the

Re: [PATCH][i386] Fix vec_construct cost, remove unused ix86_vec_cost arg

2018-10-11 Thread Richard Biener
On Thu, 11 Oct 2018, Richard Biener wrote: > > The following fixes vec_construct cost calculation to properly consider > that the inserts will happen to SSE regs thus forgo the multiplication > done in ix86_vec_cost which is passed the wrong mode. This gets rid of > the only call passing false

[PATCH][i386] Fix vec_construct cost, remove unused ix86_vec_cost arg

2018-10-11 Thread Richard Biener
The following fixes vec_construct cost calculation to properly consider that the inserts will happen to SSE regs thus forgo the multiplication done in ix86_vec_cost which is passed the wrong mode. This gets rid of the only call passing false to ix86_vec_cost (so consider the patch amended to