Memory overflow when compiling Cuda binaries

Compiling, libraries, modules, coding guidelines and porting

Moderators: jesterKing, stiv

Post Reply
Posts: 0
Joined: Tue Jan 22, 2013 10:30 pm

Memory overflow when compiling Cuda binaries

Post by epst » Tue Jan 22, 2013 10:42 pm

Hi everyone.
I have an error while try to compile Blender with Cuda binaries. The error does not let me sleep at night :)

Code: Select all

[ 78%] Generating kernel_sm_11.cubin
/opt/src/blender/intern/cycles/kernel/svm/svm.h(182): Warning: Pointer parameters must be inlined, so overriding noinline attribute on '_Z14svm_eval_nodesP13KernelGlobalsP10ShaderData10ShaderTypefi'
/tmp/tmpxft_000054f5_00000000-10_kernel.cpp3.i(0): Warning: Optimizing huge function kernel_cuda_path_trace because Olimit has been overridden;
        compiler may run out of memory or run very slowly
nvopencc ERROR: /usr/local/cuda-5.0/open64/lib//be died due to signal 9
nvopencc ERROR: Probably caused by running out of swap space -- check /var/log/messages
make[2]: *** [intern/cycles/kernel/kernel_sm_10.cubin] Error 1
make[2]: *** Ожидание завершения заданий...
I set WITH_CYCLES_CUDA_BINARIES to ON, and add sm_11 in CYCLES_CUDA_BINARIES_ARCH section in CMakeLists.txt

Debian unstable (amd64) Cuda toolkit 5.0, last nvidia drivers 313, GeForce 8800 GT support CUDA 1.1, Python 3.3

Post Reply