[Bf-committers] Problem compiling cycles+cuda on windows / slow GPU rendering

Yousef Hurfoush bat3a at msn.com
Sun Jan 22 23:01:52 CET 2012


here as will, windows 7 x64, scons, msvcsp1 2008, svn 43619, the same error:

the log:
ge_videotex.lib(Texture.obj) : warning LNK4217: locally defined symbol __GLEW_AR
B_texture_non_power_of_two imported in function "struct _object * __cdecl Textur
e_refresh(struct Texture *,struct _object *)" (?Texture_refresh@@YAPEAU_object@@
PEAUTexture@@PEAU1@@Z)
tmpxft_00000f88_00000000-3_kernel.cudafe1.gpu
Install file: "Z:\Development\blender\build\win64-patched\bin\blenderplayer.exe"
 as "Z:\Development\blender\install\win64-patched\blenderplayer.exe"
tmpxft_00000f88_00000000-10_kernel.cudafe2.gpu
./z:\development\blender\blender\intern\cycles\kernel\svm\svm_texture.h(45): War
ning: Pointer parameters must be inlined, so overriding noinline attribute on '_
Z7voronoi6float318NodeDistanceMetricfPfPS_'
z:\development\blender\blender\intern\cycles\kernel\svm/svm.h(154): Warning: Poi
nter parameters must be inlined, so overriding noinline attribute on '_Z14svm_ev
al_nodesP13KernelGlobalsP10ShaderData10ShaderTypefi'
C:/Users/CRAX/AppData/Local/Temp/tmpxft_00000f88_00000000-11_kernel.cpp3.i(0): W
arning: Optimizing huge function kernel_cuda_path_trace because Olimit has been
overridden;
        compiler may run out of memory or run very slowly
   Creating library Z:\Development\blender\build\win64-patched\bin\blender.lib a
nd object Z:\Development\blender\build\win64-patched\bin\blender.exp
ge_videotex.lib(Texture.obj) : warning LNK4217: locally defined symbol __GLEW_AR
B_texture_non_power_of_two imported in function "struct _object * __cdecl Textur
e_refresh(struct Texture *,struct _object *)" (?Texture_refresh@@YAPEAU_object@@
PEAUTexture@@PEAU1@@Z)
bf_intern_cycles.lib(device_cuda.obj) : warning LNK4217: locally defined symbol
__glewBufferData imported in function "public: virtual void __cdecl ccl::CUDADev
ice::pixels_alloc(class ccl::device_memory &)" (?pixels_alloc at CUDADevice@ccl@@UE
AAXAEAVdevice_memory at 2@@Z)
bf_intern_cycles.lib(device_cuda.obj) : warning LNK4217: locally defined symbol
__glewBindBuffer imported in function "public: virtual void __cdecl ccl::CUDADev
ice::pixels_alloc(class ccl::device_memory &)" (?pixels_alloc at CUDADevice@ccl@@UE
AAXAEAVdevice_memory at 2@@Z)
bf_intern_cycles.lib(device_cuda.obj) : warning LNK4217: locally defined symbol
__glewGenBuffers imported in function "public: virtual void __cdecl ccl::CUDADev
ice::pixels_alloc(class ccl::device_memory &)" (?pixels_alloc at CUDADevice@ccl@@UE
AAXAEAVdevice_memory at 2@@Z)
bf_intern_cycles.lib(device_cuda.obj) : warning LNK4217: locally defined symbol
__glewUnmapBuffer imported in function "public: virtual void __cdecl ccl::CUDADe
vice::pixels_copy_from(class ccl::device_memory &,int,int,int)" (?pixels_copy_fr
om at CUDADevice@ccl@@UEAAXAEAVdevice_memory at 2@HHH at Z)
bf_intern_cycles.lib(device_cuda.obj) : warning LNK4217: locally defined symbol
__glewMapBuffer imported in function "public: virtual void __cdecl ccl::CUDADevi
ce::pixels_copy_from(class ccl::device_memory &,int,int,int)" (?pixels_copy_from
@CUDADevice at ccl@@UEAAXAEAVdevice_memory at 2@HHH at Z)
bf_intern_cycles.lib(device_cuda.obj) : warning LNK4217: locally defined symbol
__glewDeleteBuffers imported in function "public: virtual void __cdecl ccl::CUDA
Device::pixels_free(class ccl::device_memory &)" (?pixels_free at CUDADevice@ccl@@U
EAAXAEAVdevice_memory at 2@@Z)
bf_intern_cycles.lib(device_cuda.obj) : warning LNK4217: locally defined symbol
__glewBindBufferARB imported in function "public: virtual void __cdecl ccl::CUDA
Device::draw_pixels(class ccl::device_memory &,int,int,int,int,int,int,bool)" (?
draw_pixels at CUDADevice@ccl@@UEAAXAEAVdevice_memory at 2@HHHHHH_N at Z)
Trying to remove existing py bundle.
Unpacking '../lib/win64/release/python32.tar.gz' to '..\install\win64-patched\2.
61\python\lib'
Install file: "Z:\Development\blender\build\win64-patched\bin\blender.exe" as "Z
:\Development\blender\install\win64-patched\blender.exe"
C:/Users/CRAX/AppData/Local/Temp/tmpxft_00000f88_00000000-11_kernel.cpp3.i(0): #
## Compiler Error (user routine 'kernel_cuda_path_trace') during Code_Expansion
phase:
### Out of memory in MEM_POOL_Realloc
nvopencc ERROR: C:/CUDA/v4.0/bin/../open64/lib//be.exe returned non-zero status
1
scons: *** [Z:\Development\blender\build\win64-patched\intern\cycles\kernel\kern
el_sm_10.cubin] Error 2
scons: building terminated because of errors.


Regards
Yousef Harfoush

bat3a at msn.com



> Date: Sun, 22 Jan 2012 21:30:59 +0200
> From: kalast at gmail.com
> To: bf-committers at blender.org
> Subject: Re: [Bf-committers] Problem compiling cycles+cuda on windows / slow GPU rendering
> 
> Same here on windows 7, I opened the system manager to check the
> memory use and indeed it seems to surpass 2 GB. I am not sure this can
> be "fixed". Inevitably the cycles kernel will get bigger and will
> require more RAM. It's interesting though that 4.1 can bite through
> this. About rendering speed I can't know since I haven't tested with
> 4.1 here. I hope this is indeed a "bug" and can be fixed, if not,
> waiting for the 4.1 normal release.
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers
 		 	   		  


More information about the Bf-committers mailing list