Missed optimization in math expression: simplify calls std::pow functions

[zamazan4ik]

Bugzilla Link	35593
Version	trunk
OS	All
Blocks	#34959
CC	@rotateright

Extended Description

clang(trunk) with '--std=c++17 -O3 -march=native -ffast-math' flags for this code:

#include <cmath>

double f(double a, double b, double c)
{
    return std::pow(a, b) * std::pow(a, c);
}

generates this assembly:

f(double, double, double): # @f(double, double, double)
  sub rsp, 24
  vmovsd qword ptr [rsp + 16], xmm2 # 8-byte Spill
  vmovsd qword ptr [rsp], xmm0 # 8-byte Spill
  call pow
  vmovsd qword ptr [rsp + 8], xmm0 # 8-byte Spill
  vmovsd xmm0, qword ptr [rsp] # 8-byte Reload
  vmovsd xmm1, qword ptr [rsp + 16] # 8-byte Reload
  call pow
  vmulsd xmm0, xmm0, qword ptr [rsp + 8] # 8-byte Folded Reload
  add rsp, 24
  ret

GCC (trunk) with '--std=c++17 -O3 -march=native -ffast-math' flags generates more optimal code:

f(double, double, double):
        vaddsd  xmm1, xmm1, xmm2
        jmp     __pow_finite

[zamazan4ik]

See a lot of optimizations here: https://github.com/gcc-mirror/gcc/blob/07b69d3f1cd3dd8ebb0af1fbff95914daee477d2/gcc/match.pd

[RKSimon]

mentioned in issue #34959

[rotateright]

This was fixed along with other pow folds shown in #34943.
This was:
7736c19