https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113426
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Last reconfirmed| |2024-01-17
Ever confirmed|0 |1
--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
We cannot analyze the evolution of 'n', so it's computation prevails inside the
loop.
<bb 3> [local count: 1063004408]:
# i_18 = PHI <i_15(5), 0(2)>
# n_lsm.10_20 = PHI <_7(5), n_lsm.10_17(2)>
# a_lsm.12_9 = PHI <_4(5), a_lsm.12_22(2)>
# b_lsm.13_8 = PHI <_6(5), b_lsm.13_23(2)>
_4 = w.2_3 + a_lsm.12_9;
_6 = b_lsm.13_8 - w.2_3;
_7 = b_lsm.13_8 + a_lsm.12_9;
i_15 = i_18 + 1;
if (i_15 != 100)
goto <bb 5>; [98.99%]
else
goto <bb 4>; [1.01%]
<bb 5> [local count: 1052266995]:
goto <bb 3>; [100.00%]
<bb 4> [local count: 10737416]:
# n_lsm.10_21 = PHI <n_lsm.10_20(3)>
(analyze_scalar_evolution
(loop_nb = 1)
(scalar = _7)
(get_scalar_evolution
(scalar = _7)
(scalar_evolution = a_lsm.12_22 + b_lsm.13_23))
)
(instantiate_scev
(instantiate_below = 2 -> 3)
(evolution_loop = 1)
(chrec = a_lsm.12_22 + b_lsm.13_23)
(res = a_lsm.12_22 + b_lsm.13_23))
(evolution_function = scev_not_known))
that's a bit odd (a_lsm.12_22 + b_lsm.13_23 is invariant).