[Bug tree-optimization/102927] Failure to optimize series of if-else to use array when possible
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102927 Andrew Pinski changed: What|Removed |Added Resolution|--- |DUPLICATE Status|NEW |RESOLVED --- Comment #8 from Andrew Pinski --- Dup of bug 40748 and others. *** This bug has been marked as a duplicate of bug 40748 ***
[Bug tree-optimization/102927] Failure to optimize series of if-else to use array when possible
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102927 --- Comment #7 from Martin Liška --- (In reply to Gabriel Ravier from comment #5) > Um, what ? How is this invalid, exactly ? Are you saying foo is faster than > baz (in which case it seems the opposite optimization should be implemented > for baz and bar), or that this optimization just won't ever be implemented > (which seems like more of a WONTFIX), or something else ? This seems kind of > odd... I say that we optimize 'foo' as 'baz' (if you adjust the minimal number of cases we consider beneficial: --param case-values-threshold=4).
[Bug tree-optimization/102927] Failure to optimize series of if-else to use array when possible
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102927 Andrew Pinski changed: What|Removed |Added Resolution|INVALID |--- Status|RESOLVED|NEW --- Comment #6 from Andrew Pinski --- Confirmed, note there might be another bug for this already filed too where I have some suggestions on how to have this fixed too.
[Bug tree-optimization/102927] Failure to optimize series of if-else to use array when possible
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102927 --- Comment #5 from Gabriel Ravier --- Um, what ? How is this invalid, exactly ? Are you saying foo is faster than baz (in which case it seems the opposite optimization should be implemented for baz and bar), or that this optimization just won't ever be implemented (which seems like more of a WONTFIX), or something else ? This seems kind of odd...
[Bug tree-optimization/102927] Failure to optimize series of if-else to use array when possible
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102927 Martin Liška changed: What|Removed |Added Resolution|--- |INVALID Status|NEW |RESOLVED --- Comment #4 from Martin Liška --- I tend to close it as invalid.
[Bug tree-optimization/102927] Failure to optimize series of if-else to use array when possible
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102927
--- Comment #3 from Martin Liška ---
> As Richi said, we do the transformation to switch only if we can expand it
> by jump table, bit test or array access.
>
> You can expand this example with:
> --param case-values-threshold=4
and we end up with array access:
$ gcc -c pr102927.c -O2 --param case-values-threshold=4
-fdump-tree-optimized=/dev/stdout
;; Function foo (foo, funcdef_no=0, decl_uid=1943, cgraph_uid=1,
symbol_order=0)
Removing basic block 5
int foo (int i)
{
int _1;
unsigned int _3;
int _4;
[local count: 1073741824]:
_3 = (unsigned int) i_2(D);
if (_3 <= 3)
goto ; [50.00%]
else
goto ; [50.00%]
[local count: 536870913]:
_4 = CSWTCH.1[_3];
[local count: 1073741824]:
# _1 = PHI <42(2), _4(3)>
return _1;
}
[Bug tree-optimization/102927] Failure to optimize series of if-else to use array when possible
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102927
--- Comment #2 from Martin Liška ---
(In reply to Gabriel Ravier from comment #0)
> int foo(int i) {
> if (i == 0)
> return 52;
> else if (i == 1)
> return 77;
> else if (i == 2)
> return 91;
> else if (i == 3)
> return 10;
> else
> return 42;
> }
>
As Richi said, we do the transformation to switch only if we can expand it by
jump table, bit test or array access.
You can expand this example with:
--param case-values-threshold=4
[Bug tree-optimization/102927] Failure to optimize series of if-else to use array when possible
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102927 Richard Biener changed: What|Removed |Added Last reconfirmed||2021-10-25 CC||marxin at gcc dot gnu.org Ever confirmed|0 |1 Status|UNCONFIRMED |NEW --- Comment #1 from Richard Biener --- I think we're doing a two-step conversion, if-to-switch and then switch conversion rather than analyzing to a common representation and then doing code generation as IIRC I suggested when if-to-switch was introduced.
