Htykuut

I'm using Numba module for Python, namely, the @vectorize decorator.

The signature is the following:

@vectorize(['float64(int64,float64,float64,int64,int64,float64)'], target='cuda', nopython=True)

def calculate(a, b, c, d, e, f):

    # calculations here...

    return result

Here the output is float64 and inputs are of various types.

As for the usage, I pass single numbers for parameters a-e and a vector for the last parameter f. So, it should basically parallelize computations for the elements of that vector.

The problem is, the code works for cpu and parallel targets but not for cuda.

Is there something wrong with the signature? Or the way I pass parameters?

The error is:

File "path-to-my-file.py", line 7, in <module>

   @vectorize(['float64(int64,float64,float64,int64,int64,float64)'], target='cuda', nopython=True)

File "C:UsersChantoAppDataLocalContinuumanaconda3libsite-packagesnumbanpyufuncdecorators.py", line 118, in wrap

   vec = Vectorize(func, **kws)

File "C:UsersChantoAppDataLocalContinuumanaconda3libsite-packagesnumbanpyufuncdecorators.py", line 38, in __new__

   return imp(func, identity=identity, cache=cache, targetoptions=kws)

File "C:UsersChantoAppDataLocalContinuumanaconda3libsite-packagesnumbanpyufuncdeviceufunc.py", line 354, in __init__

   assert not targetoptions



AssertionError