Re: [PATCH 1/2]middle-end: Apply loop->unroll directly in vectorizer

> Am 02.06.2025 um 14:16 schrieb Tamar Christina : > >  >> >> -Original Message- >> From: Richard Biener >> Sent: Monday, May 26, 2025 2:56 PM >> To: Tamar Christina >> Cc: gcc-patches@gcc.gnu.org; nd >> Subject: RE: [PATCH 1/2]mi

Re: [PATCH 1/2]middle-end: Apply loop->unroll directly in vectorizer

;> To: Tamar Christina >>> Cc: gcc-patches@gcc.gnu.org; nd >>> Subject: RE: [PATCH 1/2]middle-end: Apply loop->unroll directly in >>> vectorizer >>> >>>> On Mon, 19 May 2025, Tamar Christina wrote: >>> >>>>>> /

RE: [PATCH 1/2]middle-end: Apply loop->unroll directly in vectorizer

> -Original Message- > From: Richard Biener > Sent: Monday, May 26, 2025 2:56 PM > To: Tamar Christina > Cc: gcc-patches@gcc.gnu.org; nd > Subject: RE: [PATCH 1/2]middle-end: Apply loop->unroll directly in vectorizer > > On Mon, 19 May

RE: [PATCH 1/2]middle-end: Apply loop->unroll directly in vectorizer

On Mon, 19 May 2025, Tamar Christina wrote: > > >/* Complete the target-specific cost calculations. */ > > >loop_vinfo->vector_costs->finish_cost (loop_vinfo->scalar_costs); > > >vec_prologue_cost = loop_vinfo->vector_costs->prologue_cost (); > > > @@ -12373,6 +12394,13 @@ vect_transf

RE: [PATCH 1/2]middle-end: Apply loop->unroll directly in vectorizer

> >/* Complete the target-specific cost calculations. */ > >loop_vinfo->vector_costs->finish_cost (loop_vinfo->scalar_costs); > >vec_prologue_cost = loop_vinfo->vector_costs->prologue_cost (); > > @@ -12373,6 +12394,13 @@ vect_transform_loop (loop_vec_info loop_vinfo, > gimple *loop_ve

RE: [PATCH 1/2]middle-end: Apply loop->unroll directly in vectorizer

On Wed, 14 May 2025, Tamar Christina wrote: > > > > > > > > > > - /* Loops vectorized with a variable factor won't benefit from > > > > > + /* Loops vectorized would have already taken into account unrolling > > specified > > > > > + by the user as the suggested unroll factor, as such we nee

RE: [PATCH 1/2]middle-end: Apply loop->unroll directly in vectorizer

> > > > > > > > - /* Loops vectorized with a variable factor won't benefit from > > > > + /* Loops vectorized would have already taken into account unrolling > specified > > > > + by the user as the suggested unroll factor, as such we need to > > > > prevent the > > > > + RTL unroller fr

RE: [PATCH 1/2]middle-end: Apply loop->unroll directly in vectorizer

On Tue, 13 May 2025, Tamar Christina wrote: > > -Original Message- > > From: Richard Biener > > Sent: Tuesday, May 13, 2025 1:59 PM > > To: Tamar Christina > > Cc: gcc-patches@gcc.gnu.org; nd > > Subject: Re: [PATCH 1/2]middle-end: Apply loop

RE: [PATCH 1/2]middle-end: Apply loop->unroll directly in vectorizer

> -Original Message- > From: Richard Biener > Sent: Tuesday, May 13, 2025 1:59 PM > To: Tamar Christina > Cc: gcc-patches@gcc.gnu.org; nd > Subject: Re: [PATCH 1/2]middle-end: Apply loop->unroll directly in vectorizer > > On Tue, 13 May 2025, Tamar Chr

Re: [PATCH 1/2]middle-end: Apply loop->unroll directly in vectorizer

On Tue, 13 May 2025, Tamar Christina wrote: > Hi All, > > Consider the loop > > void f1 (int *restrict a, int n) > { > #pragma GCC unroll 4 requested > for (int i = 0; i < n; i++) > a[i] *= 2; > } > > Which today is vectorized and then unrolled 3x by the RTL unroller due to the > use of t

[PATCH 1/2]middle-end: Apply loop->unroll directly in vectorizer

Hi All, Consider the loop void f1 (int *restrict a, int n) { #pragma GCC unroll 4 requested for (int i = 0; i < n; i++) a[i] *= 2; } Which today is vectorized and then unrolled 3x by the RTL unroller due to the use of the pragma. This is unfortunate because the pragma was intended for the