nvcc stalls in be's "Verify Live-Range phase"

[ebfe]

code which works on linux (cuda 2.0b2) seems to stall nvcc; i need to interrupt nvcc with ctrl+c.

nvcc -V

nvcc: NVIDIA ® Cuda compiler driver

Copyright © 2005-2007 NVIDIA Corporation

Built on Fri_Feb__8_14:07:34_PST_2008

Cuda compilation tools, release 1.1, V0.2.1221

nvcc -v -Xptxas “-v -maxrregcount=42” -c cpyrit_cuda.cu

#$ SPACE=

#$ MODE=DEVICE

#$ HERE=/usr/local/cuda/bin

#$ THERE=/usr/local/cuda/bin

#$ TOP=/usr/local/cuda/bin/…

#$ PATH=/usr/local/cuda/bin/…/open64/bin:/usr/local/cuda/bin/…/bin:/sw/bin:/sw/sbin:/usr/local/cuda/bin:/Library/Frameworks/Python.framework/Versions/Current/bin:/usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin:/usr/X11/bin:/usr/X11R6/bin

#$ INCLUDES=“-I/usr/local/cuda/bin/…/include”

#$ LIBRARIES=  “-L/usr/local/cuda/bin/…/lib” -lcudart -Xlinker -rpath -Xlinker /usr/local/cuda/lib/

#$ CUDAFE_FLAGS=

#$ OPENCC_FLAGS=

#$ PTXAS_FLAGS=

#$ gcc -E -x c++ -DCUDA_FLOAT_MATH_FUNCTIONS -DCUDA_NO_SM_11_ATOMIC_INTRINSICS  “-I/usr/local/cuda/bin/…/include”  -I. -D__CUDACC__ -C  -include “cuda_runtime.h” -m32 -malign-double -o “/tmp/tmpxft_00000a8c_00000000-2_cpyrit_cuda.ii” “cpyrit_cuda.cu”

#$ cudafe --m32 --diag_error=host_device_limited_call -tused  --gen_c_file_name “/tmp/tmpxft_00000a8c_00000000-0_cpyrit_cuda.c” --stub_file_name “/tmp/tmpxft_00000a8c_00000000-0_cpyrit_cuda.stub.c” --stub_header_file_name “/tmp/tmpxft_00000a8c_00000000-0_cpyrit_cuda.stub.h” --gen_device_file_name “/tmp/tmpxft_00000a8c_00000000-0_cpyrit_cuda.gpu” --include_file_name /tmp/tmpxft_00000a8c_00000000-1_cpyrit_cuda.c “/tmp/tmpxft_00000a8c_00000000-2_cpyrit_cuda.ii”

#$ gcc -E -x c -DCUDA_FLOAT_MATH_FUNCTIONS -DCUDA_NO_SM_11_ATOMIC_INTRINSICS  “-I/usr/local/cuda/bin/…/include”  -I. -D__CUDACC__ -C  -m32 -malign-double -o “/tmp/tmpxft_00000a8c_00000000-3_cpyrit_cuda.i” “/tmp/tmpxft_00000a8c_00000000-0_cpyrit_cuda.gpu”

#$ cudafe --m32 --c  --gen_c_file_name “/tmp/tmpxft_00000a8c_00000000-4_cpyrit_cuda.c” --stub_file_name “/tmp/tmpxft_00000a8c_00000000-4_cpyrit_cuda.stub.c” --stub_header_file_name “/tmp/tmpxft_00000a8c_00000000-4_cpyrit_cuda.stub.h” --gen_device_file_name “/tmp/tmpxft_00000a8c_00000000-4_cpyrit_cuda.gpu” --include_file_name /tmp/tmpxft_00000a8c_00000000-1_cpyrit_cuda.c “/tmp/tmpxft_00000a8c_00000000-3_cpyrit_cuda.i”

#$ gcc -E -x c -DCUDA_FLOAT_MATH_FUNCTIONS -DCUDA_NO_SM_11_ATOMIC_INTRINSICS  “-I/usr/local/cuda/bin/…/include”  -I. -D__CUDABE__  -m32 -malign-double -o “/tmp/tmpxft_00000a8c_00000000-5_cpyrit_cuda.i” “/tmp/tmpxft_00000a8c_00000000-4_cpyrit_cuda.gpu”

#$ filehash --skip-cpp-directives -s "-O3 -v -maxrregcount=42  " “/tmp/tmpxft_00000a8c_00000000-5_cpyrit_cuda.i” > “/tmp/tmpxft_00000a8c_00000000-6_cpyrit_cuda.hash”

#$ nvopencc  -TARG:sm_10  -m32 “/tmp/tmpxft_00000a8c_00000000-5_cpyrit_cuda.i”  -o “/tmp/tmpxft_00000a8c_00000000-7_cpyrit_cuda.ptx”

^CSignal: Interrupt in Global Optimization – Verify Live-Range phase.

nvopencc ERROR: signal Interrupt caught, stop processing

the compiler stalls in the last nvopencc line with 100% cpu consumption and no further output. it work’s just fine when compiling for device emulation on mac os.

[mfatica]

CUDA 2.0 should be on CUDA Zone today or tomorrow.
There is also a Mac version, it should fix your problem.

[ebfe]

this is good news, thanks very much