Object cuda_computeCapability(Object self, int nparts, int *argcv,
Object *argv, int flags) {
cuInit(0);
int deviceCount = 0;
cuDeviceGetCount(&deviceCount);
if (deviceCount == 0) {
raiseError("No CUDA devices found");
}
CUdevice cuDevice;
int major, minor;
cuDeviceComputeCapability(&major, &minor, cuDevice);
return alloc_Float64(major + minor / 10.0);
}
Object cuda_cores(Object self, int nparts, int *argcv,
Object *argv, int flags) {
cuInit(0);
int deviceCount = 0;
cuDeviceGetCount(&deviceCount);
if (deviceCount == 0) {
raiseError("No CUDA devices found");
}
CUdevice cuDevice;
int mpcount;
cuDeviceGetAttribute(&mpcount, CU_DEVICE_ATTRIBUTE_MULTIPROCESSOR_COUNT,
cuDevice);
int major, minor;
cuDeviceComputeCapability(&major, &minor, cuDevice);
mpcount *= coreMultiplicand(major, minor);
return alloc_Float64(mpcount);
}
Object cuda_FloatArray_at(Object self, int nparts, int *argcv,
Object *argv, int flags) {
int n = integerfromAny(argv[0]);
struct CudaFloatArray *a = (struct CudaFloatArray *)self;
return alloc_Float64(a->data[n]);
}
Object math_abs(Object self, int nparams, int *argcv, Object *argv, int flags) {
return alloc_Float64(fabs(*(double*) argv[0]->data));
}
Object math_pi(Object self, int nparams, int *argcv, Object *argv, int flags) {
return alloc_Float64((double)3.141592653589793);
}
Object math_random(Object self, int nparams, int *argcv, Object *argv, int flags) {
return alloc_Float64((double)rand() / RAND_MAX);
}
