Ah, indeed after some more debugging it turns out that the error was being
caused by my `grid' argument -- the entries were floats (should have been
cast to int), and were causing the exception. For anyone interested,
here's a simple working example (using only pycuda) of passing in an int:
import pycuda.driver as cuda
import pycuda.autoinit
from pycuda.compiler import SourceModule
import numpy as np
a = np.random.randn(4,4)
a = a.astype(np.float32)
a_gpu = cuda.mem_alloc(a.nbytes)
# cast the DeviceAllocation to int, and wrap as numpy type
intptr = int(a_gpu)
intptr = np.uint64(intptr)
print intptr
print hex(intptr)
cuda.memcpy_htod(a_gpu, a)
mod = SourceModule("""
__global__ void doublify(float *a)
{
int idx = threadIdx.x + threadIdx.y*4;
a[idx] *= 2;
}
""")
block = (4,4,1)
grid = (1,1)
func = mod.get_function("doublify")
# func(a_gpu, block=block, grid=grid)
# pass in the numpy integer
func(intptr, block=block, grid=grid)
a_doubled = np.empty_like(a)
cuda.memcpy_dtoh(a_doubled, a_gpu)
print a_doubled
print a
print a_doubled/a
On Fri, Sep 27, 2013 at 3:34 PM, Andreas Kloeckner
<[email protected]>wrote:
> Sam Preston <[email protected]> writes:
>
> > Hi all,
> >
> > I would like to use pycuda to write a few kernels to interoperate with a
> > larger cuda/python library I'm already using. I can get the raw device
> > memory address as a python int, and from searching past threads on the
> > mailing list it sounds like I should be able to pass this as an argument
> to
> > my kernel. However everything I try seems to generate an error -- I'm
> > currently doing something like this:
> >
> > .
> > .
> > .
> > myfunc = mod.get_function("myfunc")
> > # myob.get() returns memory address as int
> > addr = np.uint32(myob.get())
> > myfunc(addr, block=block, grid=grid)
>
> Can you post a full reproducing example? Just passing a (sized) integer
> (to a kernel function) *should* work. For the rest of the PyCUDA API,
> you should probably use just a plain Python int.
>
> Andreas
>
--
--------------------
J. Samuel Preston
Research Assistant
Scientific Computing and Imaging Institute
University of Utah
_______________________________________________
PyCUDA mailing list
[email protected]
http://lists.tiker.net/listinfo/pycuda
