Re: [PATCH] Optimize 128-bit vector insertion into zero 512-bit vector (PR target/85480)

Hello Jakub! On 23 апр 20:31, Jakub Jelinek wrote: > Hi! > > As mentioned in the PR, vmov{aps,apd,dqa{,32,64}} 128-bit instructions > zero the rest of 512-bit register, so we can optimize insertion into zero > vectors using those instructions. > > Bootstrapped/regtested on x86_64-linux and i686-l

[PATCH] Optimize 128-bit vector insertion into zero 512-bit vector (PR target/85480)

Hi! As mentioned in the PR, vmov{aps,apd,dqa{,32,64}} 128-bit instructions zero the rest of 512-bit register, so we can optimize insertion into zero vectors using those instructions. Bootstrapped/regtested on x86_64-linux and i686-linux, ok for stage1? 2018-04-23 Jakub Jelinek PR tar