Re: [PATCH] vect: Account for unused IFN_LOAD_LANES results

2021-01-14 Thread Richard Biener
On Thu, 14 Jan 2021, Richard Sandiford wrote: > At the moment, if we use only one vector of an LD4 result, > we'll treat the LD4 as having the cost of a single load. > But all 4 loads and any associated permutes take place > regardless of which results are actually used. > > This patch therefore

[PATCH] vect: Account for unused IFN_LOAD_LANES results

2021-01-14 Thread Richard Sandiford via Gcc-patches
At the moment, if we use only one vector of an LD4 result, we'll treat the LD4 as having the cost of a single load. But all 4 loads and any associated permutes take place regardless of which results are actually used. This patch therefore counts the cost of unused LOAD_LANES results against the fi