Re: [PATCH] [AARCH64] Improve vector generation cost model

2019-12-13 Thread Kyrill Tkachov
Hi Andrew, On 3/15/19 1:18 AM, apin...@marvell.com wrote: From: Andrew Pinski Hi,   On OcteonTX2, ld1r and ld1 (with a single lane) are split into two different micro-ops unlike most other targets. This adds three extra costs to the cost table: ld1_dup: used for "ld1r {v0.4s}, [x0]"

Re: [PATCH] [AARCH64] Improve vector generation cost model

2019-12-07 Thread Andrew Pinski
On Thu, May 2, 2019 at 9:10 AM Andrew Pinski wrote: > > On Thu, Mar 14, 2019 at 6:19 PM wrote: > > > > From: Andrew Pinski > > > > Hi, > > On OcteonTX2, ld1r and ld1 (with a single lane) are split > > into two different micro-ops unlike most other targets. > > This adds three extra costs to

Re: [PATCH] [AARCH64] Improve vector generation cost model

2019-05-02 Thread Andrew Pinski
On Thu, Mar 14, 2019 at 6:19 PM wrote: > > From: Andrew Pinski > > Hi, > On OcteonTX2, ld1r and ld1 (with a single lane) are split > into two different micro-ops unlike most other targets. > This adds three extra costs to the cost table: > ld1_dup: used for "ld1r {v0.4s}, [x0]" > merge_dup: