Thorium AVX2 compilation error with -ffast-math

[jessy-lua]

Building on the Windows vs2019

../../third_party/opus/src/celt\arch.h(203,2): error: Cannot build libopus with -ffast-math unless FLOAT_APPROX is defined. This could result in crashes on extreme (e.g. NaN) input
#error Cannot build libopus with -ffast-math unless FLOAT_APPROX is defined. This could result in crashes on extreme (e.g. NaN) input
 ^
1 error generated.

[Alex313031]

@jessy-lua Did you modify the main build.gn and try to add -ffast-math.

[jessy-lua]

@jessy-lua Did you modify the main build.gn and try to add -ffast-math.

didn't modify anything

[jessy-lua]

this issue presents only when i compiling with AVX2 build folder. tried two times

[Alex313031]

@jessy-lua What CPU is your build system using.

[jessy-lua]

@jessy-lua What CPU is your build system using.

Intel 12th gen i9