I am trying to compile the CUDA PPC code on a cluster. When running 'make gpu' in the PPC_CUDA directory, I get the following error:
$ make gpu
nvcc ppc.cu -Xptxas=-v -arch=sm_50 -O2 --use_fast_math --compiler-options=-O2,--fast-math -o ppc 2>&1 | grep -v assuming
pro.cu(58): error: identifier "int_as_float" is undefined
return int_as_float(tmp|0x3f800000)-1.0f;
^
1 error detected in the compilation of "ppc.cu".
int_as_float is used outside the scope of the #ifdef XCPU, but it seems like 'pro.cu' is included after #define XCPU in 'ppc.cu', so I don't see why there should be any errors. Adding #define XCPU to the top of pro.cu introduces more errors.
I am trying to compile the CUDA PPC code on a cluster. When running 'make gpu' in the PPC_CUDA directory, I get the following error:
int_as_float
is used outside the scope of the#ifdef XCPU
, but it seems like 'pro.cu' is included after#define XCPU
in 'ppc.cu', so I don't see why there should be any errors. Adding#define XCPU
to the top ofpro.cu
introduces more errors.