public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
* [PATCH][AArch64] Fix aarch64_rtx_costs of PLUS/MINUS
@ 2015-03-04 15:38 Wilco Dijkstra
  2015-05-01 10:22 ` Marcus Shawcroft
  0 siblings, 1 reply; 6+ messages in thread
From: Wilco Dijkstra @ 2015-03-04 15:38 UTC (permalink / raw)
  To: GCC Patches

Include the cost of op0 and op1 in all cases in PLUS and MINUS in aarch64_rtx_costs.
Bootstrap & regression OK.

ChangeLog:
2015-03-04  Wilco Dijkstra  <wdijkstr@arm.com>

        * gcc/config/aarch64/aarch64.c (aarch64_rtx_costs):
        Calculate cost of op0 and op1 in PLUS and MINUS cases.

---
 gcc/config/aarch64/aarch64.c | 12 +++++-------
 1 file changed, 5 insertions(+), 7 deletions(-)

diff --git a/gcc/config/aarch64/aarch64.c b/gcc/config/aarch64/aarch64.c
index 39921a7..e22d72e 100644
--- a/gcc/config/aarch64/aarch64.c
+++ b/gcc/config/aarch64/aarch64.c
@@ -5794,6 +5794,8 @@ aarch64_rtx_costs (rtx x, int code, int outer ATTRIBUTE_UNUSED,
 	op1 = XEXP (x, 1);
 
 cost_minus:
+	*cost += rtx_cost (op0, MINUS, 0, speed);
+
 	/* Detect valid immediates.  */
 	if ((GET_MODE_CLASS (mode) == MODE_INT
 	     || (GET_MODE_CLASS (mode) == MODE_CC
@@ -5801,13 +5803,10 @@ cost_minus:
 	    && CONST_INT_P (op1)
 	    && aarch64_uimm12_shift (INTVAL (op1)))
 	  {
-	    *cost += rtx_cost (op0, MINUS, 0, speed);
-
 	    if (speed)
 	      /* SUB(S) (immediate).  */
 	      *cost += extra_cost->alu.arith;
 	    return true;
-
 	  }
 
 	/* Look for SUB (extended register).  */
@@ -5832,7 +5831,6 @@ cost_minus:
 	    *cost += aarch64_rtx_mult_cost (new_op1, MULT,
 					    (enum rtx_code) code,
 					    speed);
-	    *cost += rtx_cost (op0, MINUS, 0, speed);
 	    return true;
 	  }
 
@@ -5879,6 +5877,8 @@ cost_plus:
 	    return true;
 	  }
 
+	*cost += rtx_cost (op1, PLUS, 1, speed);
+
 	/* Look for ADD (extended register).  */
         if (aarch64_rtx_arith_op_extract_p (op0, mode))
 	  {
@@ -5900,12 +5900,10 @@ cost_plus:
 	  {
 	    *cost += aarch64_rtx_mult_cost (new_op0, MULT, PLUS,
 					    speed);
-	    *cost += rtx_cost (op1, PLUS, 1, speed);
 	    return true;
 	  }
 
-	*cost += (rtx_cost (new_op0, PLUS, 0, speed)
-		  + rtx_cost (op1, PLUS, 1, speed));
+	*cost += rtx_cost (new_op0, PLUS, 0, speed);
 
 	if (speed)
 	  {
-- 
1.9.1



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

* Re: [PATCH][AArch64] Fix aarch64_rtx_costs of PLUS/MINUS
  2015-03-04 15:38 [PATCH][AArch64] Fix aarch64_rtx_costs of PLUS/MINUS Wilco Dijkstra
@ 2015-05-01 10:22 ` Marcus Shawcroft
  2015-05-01 12:53   ` Kyrill Tkachov
  0 siblings, 1 reply; 6+ messages in thread
From: Marcus Shawcroft @ 2015-05-01 10:22 UTC (permalink / raw)
  To: Wilco Dijkstra; +Cc: GCC Patches

On 4 March 2015 at 15:37, Wilco Dijkstra <wdijkstr@arm.com> wrote:
> Include the cost of op0 and op1 in all cases in PLUS and MINUS in aarch64_rtx_costs.
> Bootstrap & regression OK.
>
> ChangeLog:
> 2015-03-04  Wilco Dijkstra  <wdijkstr@arm.com>
>
>         * gcc/config/aarch64/aarch64.c (aarch64_rtx_costs):
>         Calculate cost of op0 and op1 in PLUS and MINUS cases.

OK, we take this one instead of Kyrill's. /Marcus

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

* Re: [PATCH][AArch64] Fix aarch64_rtx_costs of PLUS/MINUS
  2015-05-01 10:22 ` Marcus Shawcroft
@ 2015-05-01 12:53   ` Kyrill Tkachov
  0 siblings, 0 replies; 6+ messages in thread
From: Kyrill Tkachov @ 2015-05-01 12:53 UTC (permalink / raw)
  To: Marcus Shawcroft, Wilco Dijkstra; +Cc: GCC Patches


On 01/05/15 11:22, Marcus Shawcroft wrote:
> On 4 March 2015 at 15:37, Wilco Dijkstra <wdijkstr@arm.com> wrote:
>> Include the cost of op0 and op1 in all cases in PLUS and MINUS in aarch64_rtx_costs.
>> Bootstrap & regression OK.
>>
>> ChangeLog:
>> 2015-03-04  Wilco Dijkstra  <wdijkstr@arm.com>
>>
>>          * gcc/config/aarch64/aarch64.c (aarch64_rtx_costs):
>>          Calculate cost of op0 and op1 in PLUS and MINUS cases.
> OK, we take this one instead of Kyrill's. /Marcus
>

I've committed this patch on Wilcos' behalf with r222676.

Thanks,
Kyrill

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

* Re: [PATCH][AArch64] Fix aarch64_rtx_costs of PLUS/MINUS
  2015-04-27 13:53 ` Kyrill Tkachov
@ 2015-04-27 14:17   ` Kyrill Tkachov
  0 siblings, 0 replies; 6+ messages in thread
From: Kyrill Tkachov @ 2015-04-27 14:17 UTC (permalink / raw)
  To: Wilco Dijkstra, 'GCC Patches'


On 27/04/15 14:53, Kyrill Tkachov wrote:
> Hi Wilco,
>
> On 27/04/15 14:43, Wilco Dijkstra wrote:
>> ping
>>
>>> -----Original Message-----
>>> From: Wilco Dijkstra [mailto:wdijkstr@arm.com]
>>> Sent: 04 March 2015 15:38
>>> To: GCC Patches
>>> Subject: [PATCH][AArch64] Fix aarch64_rtx_costs of PLUS/MINUS
>>>
>>> Include the cost of op0 and op1 in all cases in PLUS and MINUS in aarch64_rtx_costs.
>>> Bootstrap & regression OK.
>>>
>>> ChangeLog:
>>> 2015-03-04  Wilco Dijkstra  <wdijkstr@arm.com>
>>>
>>>           * gcc/config/aarch64/aarch64.c (aarch64_rtx_costs):
>>>           Calculate cost of op0 and op1 in PLUS and MINUS cases.
>>>
>>> ---
>>>    gcc/config/aarch64/aarch64.c | 12 +++++-------
>>>    1 file changed, 5 insertions(+), 7 deletions(-)
>>>
>>> diff --git a/gcc/config/aarch64/aarch64.c b/gcc/config/aarch64/aarch64.c
>>> index 39921a7..e22d72e 100644
>>> --- a/gcc/config/aarch64/aarch64.c
>>> +++ b/gcc/config/aarch64/aarch64.c
>>> @@ -5794,6 +5794,8 @@ aarch64_rtx_costs (rtx x, int code, int outer ATTRIBUTE_UNUSED,
>>>    	op1 = XEXP (x, 1);
>>>
>>>    cost_minus:
>>> +	*cost += rtx_cost (op0, MINUS, 0, speed);
>>> +
>>>    	/* Detect valid immediates.  */
>>>    	if ((GET_MODE_CLASS (mode) == MODE_INT
>>>    	     || (GET_MODE_CLASS (mode) == MODE_CC
>>> @@ -5801,13 +5803,10 @@ cost_minus:
>>>    	    && CONST_INT_P (op1)
>>>    	    && aarch64_uimm12_shift (INTVAL (op1)))
>>>    	  {
>>> -	    *cost += rtx_cost (op0, MINUS, 0, speed);
>>> -
>>>    	    if (speed)
>>>    	      /* SUB(S) (immediate).  */
>>>    	      *cost += extra_cost->alu.arith;
>>>    	    return true;
>>> -
>>>    	  }
>>>
>>>    	/* Look for SUB (extended register).  */
>>> @@ -5832,7 +5831,6 @@ cost_minus:
>>>    	    *cost += aarch64_rtx_mult_cost (new_op1, MULT,
>>>    					    (enum rtx_code) code,
>>>    					    speed);
>>> -	    *cost += rtx_cost (op0, MINUS, 0, speed);
>>>    	    return true;
>>>    	  }
>>>
>>> @@ -5879,6 +5877,8 @@ cost_plus:
>>>    	    return true;
>>>    	  }
>>>
>>> +	*cost += rtx_cost (op1, PLUS, 1, speed);
>>> +
> I don't think this is correct. In the code directly below
> (when the aarch64_rtx_arith_op_extract_p condition is true)
> we have a shift/extend operation by a constant, so we don't
> want to take into account the cost of operand 1 (which is
> the extend+shift rtx).
>
> So, I looked at this code myself recently and found that the
> wrong logic is addressed by my patch at:
> https://gcc.gnu.org/ml/gcc-patches/2015-04/msg01617.html
> which fixes the case where we forget to cost operand0 in the
> MINUS case.

Never mind, I misread the code. The case is indeed
missing the costing of operand 1, so your patch is
a superset of mine. I think it's correct, but I can't
approve. If this goes in, then I withdraw my patch at:

https://gcc.gnu.org/ml/gcc-patches/2015-04/msg01617.html


Sorry for the confusion,
Kyrill


>
> What do you think?
> Kyrill
>
>>>    	/* Look for ADD (extended register).  */
>>>            if (aarch64_rtx_arith_op_extract_p (op0, mode))
>>>    	  {
>>> @@ -5900,12 +5900,10 @@ cost_plus:
>>>    	  {
>>>    	    *cost += aarch64_rtx_mult_cost (new_op0, MULT, PLUS,
>>>    					    speed);
>>> -	    *cost += rtx_cost (op1, PLUS, 1, speed);
>>>    	    return true;
>>>    	  }
>>>
>>> -	*cost += (rtx_cost (new_op0, PLUS, 0, speed)
>>> -		  + rtx_cost (op1, PLUS, 1, speed));
>>> +	*cost += rtx_cost (new_op0, PLUS, 0, speed);
>>>
>>>    	if (speed)
>>>    	  {
>>> --
>>> 1.9.1

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

* Re: [PATCH][AArch64] Fix aarch64_rtx_costs of PLUS/MINUS
  2015-04-27 13:43 Wilco Dijkstra
@ 2015-04-27 13:53 ` Kyrill Tkachov
  2015-04-27 14:17   ` Kyrill Tkachov
  0 siblings, 1 reply; 6+ messages in thread
From: Kyrill Tkachov @ 2015-04-27 13:53 UTC (permalink / raw)
  To: Wilco Dijkstra, 'GCC Patches'

Hi Wilco,

On 27/04/15 14:43, Wilco Dijkstra wrote:
> ping
>
>> -----Original Message-----
>> From: Wilco Dijkstra [mailto:wdijkstr@arm.com]
>> Sent: 04 March 2015 15:38
>> To: GCC Patches
>> Subject: [PATCH][AArch64] Fix aarch64_rtx_costs of PLUS/MINUS
>>
>> Include the cost of op0 and op1 in all cases in PLUS and MINUS in aarch64_rtx_costs.
>> Bootstrap & regression OK.
>>
>> ChangeLog:
>> 2015-03-04  Wilco Dijkstra  <wdijkstr@arm.com>
>>
>>          * gcc/config/aarch64/aarch64.c (aarch64_rtx_costs):
>>          Calculate cost of op0 and op1 in PLUS and MINUS cases.
>>
>> ---
>>   gcc/config/aarch64/aarch64.c | 12 +++++-------
>>   1 file changed, 5 insertions(+), 7 deletions(-)
>>
>> diff --git a/gcc/config/aarch64/aarch64.c b/gcc/config/aarch64/aarch64.c
>> index 39921a7..e22d72e 100644
>> --- a/gcc/config/aarch64/aarch64.c
>> +++ b/gcc/config/aarch64/aarch64.c
>> @@ -5794,6 +5794,8 @@ aarch64_rtx_costs (rtx x, int code, int outer ATTRIBUTE_UNUSED,
>>   	op1 = XEXP (x, 1);
>>
>>   cost_minus:
>> +	*cost += rtx_cost (op0, MINUS, 0, speed);
>> +
>>   	/* Detect valid immediates.  */
>>   	if ((GET_MODE_CLASS (mode) == MODE_INT
>>   	     || (GET_MODE_CLASS (mode) == MODE_CC
>> @@ -5801,13 +5803,10 @@ cost_minus:
>>   	    && CONST_INT_P (op1)
>>   	    && aarch64_uimm12_shift (INTVAL (op1)))
>>   	  {
>> -	    *cost += rtx_cost (op0, MINUS, 0, speed);
>> -
>>   	    if (speed)
>>   	      /* SUB(S) (immediate).  */
>>   	      *cost += extra_cost->alu.arith;
>>   	    return true;
>> -
>>   	  }
>>
>>   	/* Look for SUB (extended register).  */
>> @@ -5832,7 +5831,6 @@ cost_minus:
>>   	    *cost += aarch64_rtx_mult_cost (new_op1, MULT,
>>   					    (enum rtx_code) code,
>>   					    speed);
>> -	    *cost += rtx_cost (op0, MINUS, 0, speed);
>>   	    return true;
>>   	  }
>>
>> @@ -5879,6 +5877,8 @@ cost_plus:
>>   	    return true;
>>   	  }
>>
>> +	*cost += rtx_cost (op1, PLUS, 1, speed);
>> +

I don't think this is correct. In the code directly below
(when the aarch64_rtx_arith_op_extract_p condition is true)
we have a shift/extend operation by a constant, so we don't
want to take into account the cost of operand 1 (which is
the extend+shift rtx).

So, I looked at this code myself recently and found that the
wrong logic is addressed by my patch at:
https://gcc.gnu.org/ml/gcc-patches/2015-04/msg01617.html
which fixes the case where we forget to cost operand0 in the
MINUS case.

What do you think?
Kyrill

>>   	/* Look for ADD (extended register).  */
>>           if (aarch64_rtx_arith_op_extract_p (op0, mode))
>>   	  {
>> @@ -5900,12 +5900,10 @@ cost_plus:
>>   	  {
>>   	    *cost += aarch64_rtx_mult_cost (new_op0, MULT, PLUS,
>>   					    speed);
>> -	    *cost += rtx_cost (op1, PLUS, 1, speed);
>>   	    return true;
>>   	  }
>>
>> -	*cost += (rtx_cost (new_op0, PLUS, 0, speed)
>> -		  + rtx_cost (op1, PLUS, 1, speed));
>> +	*cost += rtx_cost (new_op0, PLUS, 0, speed);
>>
>>   	if (speed)
>>   	  {
>> --
>> 1.9.1
>

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

* RE: [PATCH][AArch64] Fix aarch64_rtx_costs of PLUS/MINUS
@ 2015-04-27 13:43 Wilco Dijkstra
  2015-04-27 13:53 ` Kyrill Tkachov
  0 siblings, 1 reply; 6+ messages in thread
From: Wilco Dijkstra @ 2015-04-27 13:43 UTC (permalink / raw)
  To: 'GCC Patches'

ping

> -----Original Message-----
> From: Wilco Dijkstra [mailto:wdijkstr@arm.com]
> Sent: 04 March 2015 15:38
> To: GCC Patches
> Subject: [PATCH][AArch64] Fix aarch64_rtx_costs of PLUS/MINUS
> 
> Include the cost of op0 and op1 in all cases in PLUS and MINUS in aarch64_rtx_costs.
> Bootstrap & regression OK.
> 
> ChangeLog:
> 2015-03-04  Wilco Dijkstra  <wdijkstr@arm.com>
> 
>         * gcc/config/aarch64/aarch64.c (aarch64_rtx_costs):
>         Calculate cost of op0 and op1 in PLUS and MINUS cases.
> 
> ---
>  gcc/config/aarch64/aarch64.c | 12 +++++-------
>  1 file changed, 5 insertions(+), 7 deletions(-)
> 
> diff --git a/gcc/config/aarch64/aarch64.c b/gcc/config/aarch64/aarch64.c
> index 39921a7..e22d72e 100644
> --- a/gcc/config/aarch64/aarch64.c
> +++ b/gcc/config/aarch64/aarch64.c
> @@ -5794,6 +5794,8 @@ aarch64_rtx_costs (rtx x, int code, int outer ATTRIBUTE_UNUSED,
>  	op1 = XEXP (x, 1);
> 
>  cost_minus:
> +	*cost += rtx_cost (op0, MINUS, 0, speed);
> +
>  	/* Detect valid immediates.  */
>  	if ((GET_MODE_CLASS (mode) == MODE_INT
>  	     || (GET_MODE_CLASS (mode) == MODE_CC
> @@ -5801,13 +5803,10 @@ cost_minus:
>  	    && CONST_INT_P (op1)
>  	    && aarch64_uimm12_shift (INTVAL (op1)))
>  	  {
> -	    *cost += rtx_cost (op0, MINUS, 0, speed);
> -
>  	    if (speed)
>  	      /* SUB(S) (immediate).  */
>  	      *cost += extra_cost->alu.arith;
>  	    return true;
> -
>  	  }
> 
>  	/* Look for SUB (extended register).  */
> @@ -5832,7 +5831,6 @@ cost_minus:
>  	    *cost += aarch64_rtx_mult_cost (new_op1, MULT,
>  					    (enum rtx_code) code,
>  					    speed);
> -	    *cost += rtx_cost (op0, MINUS, 0, speed);
>  	    return true;
>  	  }
> 
> @@ -5879,6 +5877,8 @@ cost_plus:
>  	    return true;
>  	  }
> 
> +	*cost += rtx_cost (op1, PLUS, 1, speed);
> +
>  	/* Look for ADD (extended register).  */
>          if (aarch64_rtx_arith_op_extract_p (op0, mode))
>  	  {
> @@ -5900,12 +5900,10 @@ cost_plus:
>  	  {
>  	    *cost += aarch64_rtx_mult_cost (new_op0, MULT, PLUS,
>  					    speed);
> -	    *cost += rtx_cost (op1, PLUS, 1, speed);
>  	    return true;
>  	  }
> 
> -	*cost += (rtx_cost (new_op0, PLUS, 0, speed)
> -		  + rtx_cost (op1, PLUS, 1, speed));
> +	*cost += rtx_cost (new_op0, PLUS, 0, speed);
> 
>  	if (speed)
>  	  {
> --
> 1.9.1


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

end of thread, other threads:[~2015-05-01 12:53 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-03-04 15:38 [PATCH][AArch64] Fix aarch64_rtx_costs of PLUS/MINUS Wilco Dijkstra
2015-05-01 10:22 ` Marcus Shawcroft
2015-05-01 12:53   ` Kyrill Tkachov
2015-04-27 13:43 Wilco Dijkstra
2015-04-27 13:53 ` Kyrill Tkachov
2015-04-27 14:17   ` Kyrill Tkachov

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).