Can't build MAME 2016 -bgfx issue

[treaves]

I'm trying to build this using the current version of RetroPie, 4.3.3. I had the same issue with 4.3.1. I'm building on an ODroid XU4. I'm using the setup script to try and build it. In the log file, there are a lot of errors relating to bgfx (output below). I can build the seemingly generic 'mame' (no year/version) with no issue, and an update of retro pie, which checks all of the base out and rebuilds works fine.

Any help appreciated.

Archiving liblua.a...
Compiling 3rdparty/bgfx/src/bgfx.cpp...
In file included from ../../../../../3rdparty/bx/include/bx/float4_t.h:17:0,
                 from ../../../../../3rdparty/bx/include/bx/float4x4_t.h:9,
                 from ../../../../../3rdparty/bgfx/src/bgfx_p.h:129,
                 from ../../../../../3rdparty/bgfx/src/bgfx.cpp:13:
../../../../../3rdparty/bx/include/bx/float4_neon.h: In function ‘float bx::float4_x(bx::float4_t)’:
../../../../../3rdparty/bx/include/bx/float4_neon.h:99:47: error: too many arguments to function ‘float __builtin_neon_vget_lanev4sf(__vector(4) float, int)’
   return __builtin_neon_vget_lanev4sf(_a, 0, 3);
                                               ^
<built-in>: note: declared here
../../../../../3rdparty/bx/include/bx/float4_neon.h: In function ‘float bx::float4_y(bx::float4_t)’:
../../../../../3rdparty/bx/include/bx/float4_neon.h:104:47: error: too many arguments to function ‘float __builtin_neon_vget_lanev4sf(__vector(4) float, int)’
   return __builtin_neon_vget_lanev4sf(_a, 1, 3);
                                               ^
<built-in>: note: declared here
../../../../../3rdparty/bx/include/bx/float4_neon.h: In function ‘float bx::float4_z(bx::float4_t)’:
../../../../../3rdparty/bx/include/bx/float4_neon.h:109:47: error: too many arguments to function ‘float __builtin_neon_vget_lanev4sf(__vector(4) float, int)’
   return __builtin_neon_vget_lanev4sf(_a, 2, 3);
                                               ^
<built-in>: note: declared here
../../../../../3rdparty/bx/include/bx/float4_neon.h: In function ‘float bx::float4_w(bx::float4_t)’:
../../../../../3rdparty/bx/include/bx/float4_neon.h:114:47: error: too many arguments to function ‘float __builtin_neon_vget_lanev4sf(__vector(4) float, int)’
   return __builtin_neon_vget_lanev4sf(_a, 3, 3);
                                               ^
<built-in>: note: declared here
../../../../../3rdparty/bx/include/bx/float4_neon.h: In function ‘bx::float4_t bx::float4_ild(uint32_t, uint32_t, uint32_t, uint32_t)’:
../../../../../3rdparty/bx/include/bx/float4_neon.h:147:66: error: ‘__builtin_neon_vreinterpretv4sfv4si’ was not declared in this scope
   const float4_t result = __builtin_neon_vreinterpretv4sfv4si(tmp);
                                                                  ^
../../../../../3rdparty/bx/include/bx/float4_neon.h: In function ‘bx::float4_t bx::float4_isplat(uint32_t)’:
../../../../../3rdparty/bx/include/bx/float4_neon.h:169:66: error: ‘__builtin_neon_vreinterpretv4sfv4si’ was not declared in this scope
   const float4_t result = __builtin_neon_vreinterpretv4sfv4si(tmp);
                                                                  ^
../../../../../3rdparty/bx/include/bx/float4_neon.h: In function ‘bx::float4_t bx::float4_itof(bx::float4_t)’:
../../../../../3rdparty/bx/include/bx/float4_neon.h:181:65: error: ‘__builtin_neon_vreinterpretv4siv4sf’ was not declared in this scope
   const _i32x4_t itof   = __builtin_neon_vreinterpretv4siv4sf(_a);
                                                                 ^
../../../../../3rdparty/bx/include/bx/float4_neon.h:182:58: error: ‘__builtin_neon_vcvtv4si’ was not declared in this scope
   const float4_t result = __builtin_neon_vcvtv4si(itof, 1);
                                                          ^
../../../../../3rdparty/bx/include/bx/float4_neon.h: In function ‘bx::float4_t bx::float4_ftoi(bx::float4_t)’:
../../../../../3rdparty/bx/include/bx/float4_neon.h:189:56: error: ‘__builtin_neon_vcvtv4sf’ was not declared in this scope
   const _i32x4_t ftoi   = __builtin_neon_vcvtv4sf(_a, 1);
                                                        ^
../../../../../3rdparty/bx/include/bx/float4_neon.h:190:67: error: ‘__builtin_neon_vreinterpretv4sfv4si’ was not declared in this scope
   const float4_t result = __builtin_neon_vreinterpretv4sfv4si(ftoi);
                                                                   ^
../../../../../3rdparty/bx/include/bx/float4_neon.h: In function ‘bx::float4_t bx::float4_add(bx::float4_t, bx::float4_t)’:
../../../../../3rdparty/bx/include/bx/float4_neon.h:197:43: error: too many arguments to function ‘__vector(4) float __builtin_neon_vaddv4sf(__vector(4) float, __vector(4) float)’
   return __builtin_neon_vaddv4sf(_a, _b, 3);
                                           ^
<built-in>: note: declared here
../../../../../3rdparty/bx/include/bx/float4_neon.h: In function ‘bx::float4_t bx::float4_sub(bx::float4_t, bx::float4_t)’:
../../../../../3rdparty/bx/include/bx/float4_neon.h:202:43: error: too many arguments to function ‘__vector(4) float __builtin_neon_vsubv4sf(__vector(4) float, __vector(4) float)’
   return __builtin_neon_vsubv4sf(_a, _b, 3);
                                           ^
<built-in>: note: declared here
../../../../../3rdparty/bx/include/bx/float4_neon.h: In function ‘bx::float4_t bx::float4_mul(bx::float4_t, bx::float4_t)’:
../../../../../3rdparty/bx/include/bx/float4_neon.h:207:43: error: ‘__builtin_neon_vmulv4sf’ was not declared in this

[caver01]

@treaves Ow! Ow. My eyes! It burns! Help!