Scons: refactor cycles kernel code to avoid building the AVX kernel with

compilers that don't support it.

CMake still needs to updated to work the same for consistency, but this should
fix the OS X buildbot at least.

1 files changed, 40 insertions, 37 deletions

diff --git a/intern/cycles/SConscript b/intern/cycles/SConscript
index c9465ada950..3b7ca7d3b8f 100644
--- a/intern/cycles/SConscript
+++ b/intern/cycles/SConscript
@@ -75,52 +75,55 @@ else:
 if env['OURPLATFORM'] in ('win32-vc', 'win32-mingw', 'linuxcross', 'win64-vc', 'win64-mingw'):
     incs.append(env['BF_PTHREADS_INC'])
 
-# optimized kernel
-sse2_cxxflags = Split(env['CXXFLAGS'])
-sse3_cxxflags = Split(env['CXXFLAGS'])
-sse41_cxxflags = Split(env['CXXFLAGS'])
-avx_cxxflags = Split(env['CXXFLAGS'])
+# optimized kernel. we compile the kernel multiple times with different
+# optimization flags, at runtime it will choose the optimal kernel
+kernel_flags = {}
 
 if env['OURPLATFORM'] == 'win32-vc':
     # there is no /arch:SSE3, but intrinsics are available anyway
-    sse2_cxxflags.append('/arch:SSE /arch:SSE2 -D_CRT_SECURE_NO_WARNINGS /fp:fast /Ox /Gs-'.split())
-    sse3_cxxflags.append('/arch:SSE /arch:SSE2 -D_CRT_SECURE_NO_WARNINGS /fp:fast /Ox /Gs-'.split())
-    sse41_cxxflags.append('/arch:SSE /arch:SSE2 -D_CRT_SECURE_NO_WARNINGS /fp:fast /Ox /Gs-'.split())
-    avx_cxxflags.append('/arch:SSE /arch:SSE2 -D_CRT_SECURE_NO_WARNINGS /fp:fast /Ox /Gs-'.split()) #/arch:AVX for VC2012 and above
-elif env['OURPLATFORM'] == 'win64-vc':
-    sse2_cxxflags.append('-D_CRT_SECURE_NO_WARNINGS /fp:fast /Ox /Gs-'.split())
-    sse3_cxxflags.append('-D_CRT_SECURE_NO_WARNINGS /fp:fast /Ox /Gs-'.split())
-    sse41_cxxflags.append('-D_CRT_SECURE_NO_WARNINGS /fp:fast /Ox /Gs-'.split())
-    avx_cxxflags.append('-D_CRT_SECURE_NO_WARNINGS /fp:fast /Ox /Gs-'.split()) #/arch:AVX for VC2012 and above
-else:
-    sse2_cxxflags.append('-ffast-math -msse -msse2 -mfpmath=sse'.split())
-    sse3_cxxflags.append('-ffast-math -msse -msse2 -msse3 -mssse3 -mfpmath=sse'.split())
-    sse41_cxxflags.append('-ffast-math -msse -msse2 -msse3 -mssse3 -msse4.1 -mfpmath=sse'.split())
-    avx_cxxflags.append('-ffast-math -msse -msse2 -msse3 -mssse3 -msse4.1 -mavx -mfpmath=sse'.split())
+    kernel_flags['sse2'] = '/arch:SSE /arch:SSE2 -D_CRT_SECURE_NO_WARNINGS /fp:fast /Ox /Gs-'
+    kernel_flags['sse3'] = kernel_flags['sse2']
+    kernel_flags['sse41'] = kernel_flags['sse3']
 
-optim_defs = defs[:]
+elif env['OURPLATFORM'] == 'win64-vc':
+    # /arch:AVX only available from visual studio 2012
+    kernel_flags['sse2'] = '-D_CRT_SECURE_NO_WARNINGS /fp:fast /Ox /Gs-'
+    kernel_flags['sse3'] = kernel_flags['sse2']
+    kernel_flags['sse41'] = kernel_flags['sse3']
 
-cycles_avx = cycles.Clone()
-avx_sources = [path.join('kernel', 'kernel_avx.cpp')]
-if env['OURPLATFORM'] == 'darwin' and env['C_COMPILER_ID'] == 'gcc' and  env['CCVERSION'] >= '4.6': # use Apple assembler for avx , gnu-compilers do not support it ( gnu gcc-4.6 or higher case )
-    cycles_avx.BlenderLib('bf_intern_cycles_avx', avx_sources, incs, optim_defs, libtype=['intern'], priority=[10], cxx_compileflags=avx_cxxflags, cc_compilerchange='/usr/bin/clang', cxx_compilerchange='/usr/bin/clang++')
+    if env['MSVC_VERSION'] in {'11.0', '12.0'}:
+        kernel_flags['avx'] = kernel_flags['sse41'] + ' /arch:AVX'
 else:
-    cycles_avx.BlenderLib('bf_intern_cycles_avx', avx_sources, incs, optim_defs, libtype=['intern'], priority=[10], cxx_compileflags=avx_cxxflags)
-
-cycles_sse41 = cycles.Clone()
-sse41_sources = [path.join('kernel', 'kernel_sse41.cpp')]
-cycles_sse41.BlenderLib('bf_intern_cycles_sse41', sse41_sources, incs, optim_defs, libtype=['intern'], priority=[10], cxx_compileflags=sse41_cxxflags)
-
-cycles_sse3 = cycles.Clone()
-sse3_sources = [path.join('kernel', 'kernel_sse3.cpp')]
-cycles_sse3.BlenderLib('bf_intern_cycles_sse3', sse3_sources, incs, optim_defs, libtype=['intern'], priority=[10], cxx_compileflags=sse3_cxxflags)
-
-cycles_sse2 = cycles.Clone()
-sse2_sources = [path.join('kernel', 'kernel_sse2.cpp')]
-cycles_sse2.BlenderLib('bf_intern_cycles_sse2', sse2_sources, incs, optim_defs, libtype=['intern'], priority=[10], cxx_compileflags=sse2_cxxflags)
+    # -mavx only available with relatively new gcc/clang
+    kernel_flags['sse2'] = '-ffast-math -msse -msse2 -mfpmath=sse'
+    kernel_flags['sse3'] = kernel_flags['sse2'] + ' -msse3 -mssse3'
+    kernel_flags['sse41'] = kernel_flags['sse3'] + ' -msse4.1'
+
+    if (env['C_COMPILER_ID'] == 'gcc' and env['CCVERSION'] >= '4.6') or (env['C_COMPILER_ID'] == 'clang' and env['CCVERSION'] >= '3.1'):
+        kernel_flags['avx'] = kernel_flags['sse41'] + ' -mavx'
+
+for kernel_type in kernel_flags.keys():
+    defs.append('WITH_KERNEL_' + kernel_type.upper())
+
+for kernel_type in kernel_flags.keys():
+    kernel_source = path.join('kernel', 'kernel_' + kernel_type + '.cpp')
+    kernel_cxxflags = Split(env['CXXFLAGS'])
+    kernel_cxxflags.append(kernel_flags[kernel_type].split())
+    kernel_defs = defs[:]
+    kernel_env = cycles.Clone()
+
+    if env['OURPLATFORM'] == 'darwin' and env['C_COMPILER_ID'] == 'gcc' and  env['CCVERSION'] >= '4.6':
+        # use Apple assembler for avx , gnu-compilers do not support it ( gnu gcc-4.6 or higher case )
+        kernel_env.BlenderLib('bf_intern_cycles_' + kernel_type, [kernel_source], incs, kernel_defs,
+            libtype=['intern'], priority=[10], cxx_compileflags=kernel_cxxflags,
+            cc_compilerchange='/usr/bin/clang', cxx_compilerchange='/usr/bin/clang++')
+    else:
+        kernel_env.BlenderLib('bf_intern_cycles_' + kernel_type, [kernel_source], incs, kernel_defs,
+            libtype=['intern'], priority=[10], cxx_compileflags=kernel_cxxflags)
 
 cycles.BlenderLib('bf_intern_cycles', sources, incs, defs, libtype=['intern'], priority=[0], cxx_compileflags=cxxflags)
 
+# OSL shaders
 if env['WITH_BF_CYCLES_OSL']:
     oso_files = SConscript(['kernel/shaders/SConscript'])
     cycles.Depends("kernel/osl/osl_shader.o", oso_files)