Re: [PATCH] tree-optimization/101186 - extend FRE with "equivalence map" for condition prediction

[Andrew MacLeod <amacleod@redhat.com>]

On 6/27/21 11:46 AM, Aldy Hernandez wrote:
>
>
> On 6/25/21 9:38 AM, Richard Biener wrote:
>> On Thu, Jun 24, 2021 at 5:01 PM Andrew MacLeod <amacleod@redhat.com> 
>> wrote:
>>>
>>> On 6/24/21 9:25 AM, Andrew MacLeod wrote:
>>>> On 6/24/21 8:29 AM, Richard Biener wrote:
>>>>
>>>>
>>>> THe original function in EVRP currently looks like:
>>>>
>>>>   =========== BB 2 ============
>>>>      <bb 2> :
>>>>      if (a_5(D) == b_6(D))
>>>>        goto <bb 8>; [INV]
>>>>      else
>>>>        goto <bb 7>; [INV]
>>>>
>>>> =========== BB 8 ============
>>>> Equivalence set : [a_5(D), b_6(D)]                 edge 2->8 provides
>>>> a_5 and b_6 as equivalences
>>>>      <bb 8> :
>>>>      goto <bb 6>; [100.00%]
>>>>
>>>> =========== BB 6 ============
>>>>      <bb 6> :
>>>>      # i_1 = PHI <0(8), i_10(5)>
>>>>      if (i_1 < a_5(D))
>>>>        goto <bb 3>; [INV]
>>>>      else
>>>>        goto <bb 7>; [INV]
>>>>
>>>> =========== BB 3 ============
>>>> Relational : (i_1 < a_5(D))                         edge 6->3 provides
>>>> this relation
>>>>      <bb 3> :
>>>>      if (i_1 == b_6(D))
>>>>        goto <bb 4>; [INV]
>>>>      else
>>>>        goto <bb 5>; [INV]
>>>>
>>>>
>>>> So It knows that a_5 and b_6 are equivalence, and it knows that i_1 <
>>>> a_5 in BB3 as well..
>>>>
>>>> so we should be able to indicate that  i_1 == b_6 as [0,0]..  we
>>>> currently aren't.   I think I had turned on equivalence mapping during
>>>> relational processing, so should be able to tag that without
>>>> transitive relations...  I'll have a look at why.
>>>>
>>>> And once we get a bit further along, you will be able to access this
>>>> without ranger.. if one wants to simply register the relations 
>>>> directly.
>>>>
>>>> Anyway, I'll get back to you why its currently being missed.
>>>>
>>>> Andrew
>>>>
>>>>
>>>>
>>> As promised.  There was a typo in the equivalency comparisons... so it
>>> was getting missed.  With the fix, the oracle identifies the relation
>>> and evrp will now fold that expression away and the IL becomes:
>>>
>>>     <bb 2> :
>>>     if (a_5(D) == b_6(D))
>>>       goto <bb 4>; [INV]
>>>     else
>>>       goto <bb 5>; [INV]
>>>
>>>     <bb 3> :
>>>     i_10 = i_1 + 1;
>>>
>>>     <bb 4> :
>>>     # i_1 = PHI <0(2), i_10(3)>
>>>     if (i_1 < a_5(D))
>>>       goto <bb 3>; [INV]
>>>     else
>>>       goto <bb 5>; [INV]
>>>
>>>     <bb 5> :
>>>     return;
>>>
>>> for the other cases you quote, there are no predictions such that if a
>>> != 0 then this equivalency exists...
>>>
>>> +  if (a != 0)
>>> +    {
>>> +      c = b;
>>> +    }
>>>
>>> but the oracle would register that in the TRUE block,  c and b are
>>> equivalent... so some other pass that was interested in tracking
>>> conditions that make a block relevant would be able to compare 
>>> relations...
>>
>> I guess to fully leverage optimizations for cases like
>>
>>    if (a != 0)
>>      c = b;
>>    ...
>>    if (a != 0)
>>      {
>>          if (c == b)
>> ...
>>      }
>>
>> That is, we'd do simplifications exposed by jump threading but
>> without actually doing the jump threading (which will of course
>> not allow all possible simplifications w/o inserting extra PHIs
>> for computations we might want to re-use).
>
> FWIW, as I mention in the PR, if the upcoming threader work could be 
> taught to use the relation oracle, it could easily solve the 
> conditional flowing through the a!=0 path.  However, we wouldn't be 
> able to thread it because in this particular case, the path crosses 
> loop boundaries.
>
> I leave it to Jeff/others to pontificate on whether the jump-threader 
> path duplicator could be taught to through loops. ??
>
> Aldy
>
This is still bouncing around in my head. I think we have the tools to 
do this better than via threading,  Ranger is now trivially capable of 
calculating when a predicate expression is true or false at another 
location in the IL. Combine this with flagging relations that are true 
when the predicate is, and that relation could be simply added into the 
oracle.

ie:

     <bb 2> :
     if (a_5(D) != 0)
       goto <bb 3>; [INV]
     else
       goto <bb 4>; [INV]

     <bb 3> :
     <bb 4> :
     # c_1 = PHI <c_6(D)(2), b_7(D)(3)>

the predicate and relations are:
     (a_5 != 0)  ->  c_1 == b_7
     (a_5 == 0) -> c_1 == c_6

then :

<bb 9> :
     # i_2 = PHI <0(4), i_12(8)>
     if (c_1 > i_2)
       goto <bb 5>; [INV]
     else
       goto <bb 10>; [INV]

     <bb 5> :                                9->5 registers c_1 > 1_2 
with the oracle
     if (a_5(D) != 0)
       goto <bb 6>; [INV]
     else
       goto <bb 8>; [INV]

     <bb 6> :
     if (i_2 == b_7(D))
       goto <bb 7>; [INV]
     else
       goto <bb 8>; [INV]
..
If we know to check the predicate list in bb_6, ranger can answer the 
question: on the branch in bb6, a_5 != 0.
This in turn means the predicated relation c_1 == b_7 can be applied to 
bb6 and register with the oracle.
Once that is done,  we already know c_1 > i_2 so we'll fold i_2 == b_7 
as [0, 0] as the equivalency between b_7 and c_1 is now applied.

So the capability is built in.. it boils down to finding the predicated 
relations we care about and knowing to apply them.
This one is pretty straightforward because the condition is exactly the 
same.   When we see a_5 != 0, a_5 is in the export list and we just 
check to see if there are any predicated flagged on a_5.  The actual 
expression could be more complicated, and it would still be able to 
answer it.  This is very similar to how the new threader finds 
threads..  It matches imports and exports.. here we mostly care just 
about the exports and figuring out what the predicates we care about are,

Anyway, theres a fit in there somewhere.  It might be something as 
straightforward as a an enhancement to the relation oracle.   I plan to 
get to some experiments eventually.

Andrew