public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
* [PATCH i386 AVX512] [80/n] Extend expand_sse2_mulvxdi3.
@ 2014-10-16 12:01 Kirill Yukhin
  2014-10-16 12:34 ` Uros Bizjak
  0 siblings, 1 reply; 6+ messages in thread
From: Kirill Yukhin @ 2014-10-16 12:01 UTC (permalink / raw)
  To: Uros Bizjak; +Cc: Jakub Jelinek, Richard Henderson, GCC Patches, kirill.yukhin

Hello,
This patch extends expand_sse2_mulvxdi3.

Bootstrapped.
AVX-512* tests on top of patch-set all pass
under simulator.

Is it ok for trunk?

gcc/
	* config/i386/i386.c (ix86_expand_sse2_mulvxdi3): Extend
	expand_sse2_mulvxdi3.

--
Thanks, K

diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index 1ee947a..945bc8d 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -45667,7 +45667,19 @@ ix86_expand_sse2_mulvxdi3 (rtx op0, rtx op1, rtx op2)
   enum machine_mode mode = GET_MODE (op0);
   rtx t1, t2, t3, t4, t5, t6;
 
-  if (TARGET_XOP && mode == V2DImode)
+  if (TARGET_AVX512DQ)
+    {
+      if (mode == V8DImode)
+	emit_insn (gen_avx512dq_mulv8di3 (op0, op1, op2));
+      else if (TARGET_AVX512VL)
+	{
+	  if (mode == V4DImode)
+	    emit_insn (gen_avx512dq_mulv4di3 (op0, op1, op2));
+	  else if (mode == V2DImode)
+	    emit_insn (gen_avx512dq_mulv4di3 (op0, op1, op2));
+	}
+    }
+  else if (TARGET_XOP && mode == V2DImode)
     {
       /* op1: A,B,C,D, op2: E,F,G,H */
       op1 = gen_lowpart (V4SImode, op1);

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH i386 AVX512] [80/n] Extend expand_sse2_mulvxdi3.
  2014-10-16 12:01 [PATCH i386 AVX512] [80/n] Extend expand_sse2_mulvxdi3 Kirill Yukhin
@ 2014-10-16 12:34 ` Uros Bizjak
  2014-10-17 12:34   ` Kirill Yukhin
  0 siblings, 1 reply; 6+ messages in thread
From: Uros Bizjak @ 2014-10-16 12:34 UTC (permalink / raw)
  To: Kirill Yukhin; +Cc: Jakub Jelinek, Richard Henderson, GCC Patches

On Thu, Oct 16, 2014 at 1:55 PM, Kirill Yukhin <kirill.yukhin@gmail.com> wrote:
> Hello,
> This patch extends expand_sse2_mulvxdi3.
>
> Bootstrapped.
> AVX-512* tests on top of patch-set all pass
> under simulator.
>
> Is it ok for trunk?
>
> gcc/
>         * config/i386/i386.c (ix86_expand_sse2_mulvxdi3): Extend
>         expand_sse2_mulvxdi3.
>
> --
> Thanks, K
>
> diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
> index 1ee947a..945bc8d 100644
> --- a/gcc/config/i386/i386.c
> +++ b/gcc/config/i386/i386.c
> @@ -45667,7 +45667,19 @@ ix86_expand_sse2_mulvxdi3 (rtx op0, rtx op1, rtx op2)
>    enum machine_mode mode = GET_MODE (op0);
>    rtx t1, t2, t3, t4, t5, t6;
>
> -  if (TARGET_XOP && mode == V2DImode)
> +  if (TARGET_AVX512DQ)
> +    {
> +      if (mode == V8DImode)
> +       emit_insn (gen_avx512dq_mulv8di3 (op0, op1, op2));
> +      else if (TARGET_AVX512VL)
> +       {
> +         if (mode == V4DImode)
> +           emit_insn (gen_avx512dq_mulv4di3 (op0, op1, op2));
> +         else if (mode == V2DImode)
> +           emit_insn (gen_avx512dq_mulv4di3 (op0, op1, op2));

Should this be v2di ?

> +       }
> +    }
> +  else if (TARGET_XOP && mode == V2DImode)
>      {
>        /* op1: A,B,C,D, op2: E,F,G,H */
>        op1 = gen_lowpart (V4SImode, op1);

Please use function pointers in the added part.

Thanks,
Uros.

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH i386 AVX512] [80/n] Extend expand_sse2_mulvxdi3.
  2014-10-16 12:34 ` Uros Bizjak
@ 2014-10-17 12:34   ` Kirill Yukhin
  2014-10-17 12:45     ` Uros Bizjak
  0 siblings, 1 reply; 6+ messages in thread
From: Kirill Yukhin @ 2014-10-17 12:34 UTC (permalink / raw)
  To: Uros Bizjak; +Cc: Jakub Jelinek, Richard Henderson, GCC Patches

Hello Uroš,
On 16 Oct 14:29, Uros Bizjak wrote:
> > +         if (mode == V4DImode)
> > +           emit_insn (gen_avx512dq_mulv4di3 (op0, op1, op2));
> > +         else if (mode == V2DImode)
> > +           emit_insn (gen_avx512dq_mulv4di3 (op0, op1, op2));
> 
> Should this be v2di ?
Right, copy-and-paste :(
> 
> > +       }
> > +    }
> > +  else if (TARGET_XOP && mode == V2DImode)
> >      {
> >        /* op1: A,B,C,D, op2: E,F,G,H */
> >        op1 = gen_lowpart (V4SImode, op1);
> 
> Please use function pointers in the added part.
Done.

Updated patch in the bottom. Is it ok?

--
Thanks, K

diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index c3202c4..415e330 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -45667,7 +45667,22 @@ ix86_expand_sse2_mulvxdi3 (rtx op0, rtx op1, rtx op2)
   enum machine_mode mode = GET_MODE (op0);
   rtx t1, t2, t3, t4, t5, t6;
 
-  if (TARGET_XOP && mode == V2DImode)
+  if (TARGET_AVX512DQ)
+    {
+      rtx (*gen) (rtx, rtx, rtx);
+
+      if (mode == V8DImode)
+	gen = gen_avx512dq_mulv8di3;
+      else if (TARGET_AVX512VL)
+	{
+	  if (mode == V4DImode)
+	    gen = gen_avx512dq_mulv4di3;
+	  else if (mode == V2DImode)
+	    gen = gen_avx512dq_mulv2di3;
+	}
+      emit_insn (gen (op0, op1, op2));
+    }
+  else if (TARGET_XOP && mode == V2DImode)
     {
       /* op1: A,B,C,D, op2: E,F,G,H */
       op1 = gen_lowpart (V4SImode, op1);

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH i386 AVX512] [80/n] Extend expand_sse2_mulvxdi3.
  2014-10-17 12:34   ` Kirill Yukhin
@ 2014-10-17 12:45     ` Uros Bizjak
  2014-10-17 14:27       ` [PATCH i386 AVX512 Boostrap] " Kirill Yukhin
  0 siblings, 1 reply; 6+ messages in thread
From: Uros Bizjak @ 2014-10-17 12:45 UTC (permalink / raw)
  To: Kirill Yukhin; +Cc: Jakub Jelinek, Richard Henderson, GCC Patches

On Fri, Oct 17, 2014 at 2:32 PM, Kirill Yukhin <kirill.yukhin@gmail.com> wrote:
> Hello Uroš,
> On 16 Oct 14:29, Uros Bizjak wrote:
>> > +         if (mode == V4DImode)
>> > +           emit_insn (gen_avx512dq_mulv4di3 (op0, op1, op2));
>> > +         else if (mode == V2DImode)
>> > +           emit_insn (gen_avx512dq_mulv4di3 (op0, op1, op2));
>>
>> Should this be v2di ?
> Right, copy-and-paste :(
>>
>> > +       }
>> > +    }
>> > +  else if (TARGET_XOP && mode == V2DImode)
>> >      {
>> >        /* op1: A,B,C,D, op2: E,F,G,H */
>> >        op1 = gen_lowpart (V4SImode, op1);
>>
>> Please use function pointers in the added part.
> Done.
>
> Updated patch in the bottom. Is it ok?

OK.

Thanks,
Uros.

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH i386 AVX512 Boostrap] [80/n] Extend expand_sse2_mulvxdi3.
  2014-10-17 12:45     ` Uros Bizjak
@ 2014-10-17 14:27       ` Kirill Yukhin
  2014-10-17 14:30         ` Uros Bizjak
  0 siblings, 1 reply; 6+ messages in thread
From: Kirill Yukhin @ 2014-10-17 14:27 UTC (permalink / raw)
  To: Uros Bizjak; +Cc: Jakub Jelinek, Richard Henderson, GCC Patches

Hello,
This is fix for bootstrap failure.

Is it OK?

gcc/
	* config/i386/i386.c (ix86_expand_sse2_mulvxdi3): Refactor
	conditions to fix bootstrap.

--
Thanks, K

diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index 7040200..3ddaf3d 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -45671,21 +45671,12 @@ ix86_expand_sse2_mulvxdi3 (rtx op0, rtx op1, rtx op2)
   enum machine_mode mode = GET_MODE (op0);
   rtx t1, t2, t3, t4, t5, t6;
 
-  if (TARGET_AVX512DQ)
-    {
-      rtx (*gen) (rtx, rtx, rtx);
-
-      if (mode == V8DImode)
-	gen = gen_avx512dq_mulv8di3;
-      else if (TARGET_AVX512VL)
-	{
-	  if (mode == V4DImode)
-	    gen = gen_avx512dq_mulv4di3;
-	  else if (mode == V2DImode)
-	    gen = gen_avx512dq_mulv2di3;
-	}
-      emit_insn (gen (op0, op1, op2));
-    }
+  if (TARGET_AVX512DQ && mode == V8DImode)
+    emit_insn (gen_avx512dq_mulv8di3 (op0, op1, op2));
+  else if (TARGET_AVX512DQ && TARGET_AVX512VL && mode == V4DImode)
+    emit_insn (gen_avx512dq_mulv4di3 (op0, op1, op2));
+  else if (TARGET_AVX512DQ && TARGET_AVX512VL && mode == V2DImode)
+    emit_insn (gen_avx512dq_mulv2di3 (op0, op1, op2));
   else if (TARGET_XOP && mode == V2DImode)
     {
       /* op1: A,B,C,D, op2: E,F,G,H */

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH i386 AVX512 Boostrap] [80/n] Extend expand_sse2_mulvxdi3.
  2014-10-17 14:27       ` [PATCH i386 AVX512 Boostrap] " Kirill Yukhin
@ 2014-10-17 14:30         ` Uros Bizjak
  0 siblings, 0 replies; 6+ messages in thread
From: Uros Bizjak @ 2014-10-17 14:30 UTC (permalink / raw)
  To: Kirill Yukhin; +Cc: Jakub Jelinek, Richard Henderson, GCC Patches

On Fri, Oct 17, 2014 at 4:25 PM, Kirill Yukhin <kirill.yukhin@gmail.com> wrote:
> Hello,
> This is fix for bootstrap failure.
>
> Is it OK?
>
> gcc/
>         * config/i386/i386.c (ix86_expand_sse2_mulvxdi3): Refactor
>         conditions to fix bootstrap.

Well, OK.

Uros.

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2014-10-17 14:27 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-10-16 12:01 [PATCH i386 AVX512] [80/n] Extend expand_sse2_mulvxdi3 Kirill Yukhin
2014-10-16 12:34 ` Uros Bizjak
2014-10-17 12:34   ` Kirill Yukhin
2014-10-17 12:45     ` Uros Bizjak
2014-10-17 14:27       ` [PATCH i386 AVX512 Boostrap] " Kirill Yukhin
2014-10-17 14:30         ` Uros Bizjak

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).