Re: [PATCH] gimple-isel: Fall back to using vcond_mask [PR98560]

2021-01-07 Thread Richard Biener
On Thu, 7 Jan 2021, Richard Sandiford wrote: > Richard Biener writes: > > On Wed, 6 Jan 2021, Richard Sandiford wrote: > > > >> PR98560 is about a case in which the vectoriser initially generates: > >> > >> mask_1 = a < 0; > >> mask_2 = mask_1 & ...; > >> res = VEC_COND_EXPR ; > >> > >> T

Re: [PATCH] gimple-isel: Fall back to using vcond_mask [PR98560]

2021-01-07 Thread Richard Sandiford via Gcc-patches
Richard Biener writes: > On Wed, 6 Jan 2021, Richard Sandiford wrote: > >> PR98560 is about a case in which the vectoriser initially generates: >> >> mask_1 = a < 0; >> mask_2 = mask_1 & ...; >> res = VEC_COND_EXPR ; >> >> The vectoriser thus expects res to be calculated using vcond_mask.

Re: [PATCH] gimple-isel: Fall back to using vcond_mask [PR98560]

2021-01-06 Thread Richard Biener
On Wed, 6 Jan 2021, Richard Sandiford wrote: > PR98560 is about a case in which the vectoriser initially generates: > > mask_1 = a < 0; > mask_2 = mask_1 & ...; > res = VEC_COND_EXPR ; > > The vectoriser thus expects res to be calculated using vcond_mask. > However, we later manage to fold

[PATCH] gimple-isel: Fall back to using vcond_mask [PR98560]

2021-01-06 Thread Richard Sandiford via Gcc-patches
PR98560 is about a case in which the vectoriser initially generates: mask_1 = a < 0; mask_2 = mask_1 & ...; res = VEC_COND_EXPR ; The vectoriser thus expects res to be calculated using vcond_mask. However, we later manage to fold mask_2 to mask_1, leaving: mask_1 = a < 0; res = VEC_CON