{
int num_steps = 1000000;
double step = 1.0/num_steps;
int i = get_global_id(0);
double x = (i - 0.5) * step;
integrals[i] = 4.0 / (1.0 + x * x);
}"""
context = cl.create_some_context()
# Print out device info
deviceinfo.output_device_info(context.devices[0])
# Create a command queue
queue = cl.CommandQueue(context)
# Create the compute program from the source buffer
# and build it
program = cl.Program(context, kernelsource).build()
#Create empty vector of h_i that will be used to store the output
h_i = numpy.empty(LENGTH).astype(numpy.float32)
# Create the output (i) array in device memory
d_i = cl.Buffer(context, cl.mem_flags.READ_WRITE, h_i.nbytes)
int i = get_global_id(0);
if (i < count)
c[i] = a[i] + b[i];
}
"""
#------------------------------------------------------------------------------
# Main procedure
# Create a compute context
# Ask the user to select a platform/device on the CLI
context = cl.create_some_context()
# Print out device info
deviceinfo.output_device_info(context.devices[0])
# Create a command queue
queue = cl.CommandQueue(context)
# Create the compute program from the source buffer
# and build it
program = cl.Program(context, kernelsource).build()
# Create a and b vectors and fill with random float values
h_a = numpy.random.rand(LENGTH).astype(numpy.float32)
h_b = numpy.random.rand(LENGTH).astype(numpy.float32)
# Create an empty c vector (a+b) to be returned from the compute device
h_c = numpy.empty(LENGTH).astype(numpy.float32)
# Create the input (a, b) arrays in device memory and copy data from host