sv650

Convolution example

Discussion created by sv650 on Jan 2, 2010
Latest reply on Jan 4, 2010 by MicahVillmow

Hi

I am trying to use the convolution example - I am able to run the example and produce correct results for part 1 - in part 2 (where you use the GPU) all of my answers are wrong from GPU.  The GPU is returning values of -999.999 for all grid values.

I am using an Nvidia (GT 9600, single precision) hardware - in order to get it compile on the machine, I changed the device context query to GPU as CPU is not yet supported as a OCL device in Nvidia's current stable SDK.  I think this is due to single vs double precision issue, but may be I am wrong - can some one give me any help please?

This is the only bit I changed from the main.cpp file on the ATI website.

oclHandles.context = clCreateContextFromType(NULL,
                                                CL_DEVICE_TYPE_CPU,
                                                NULL,
                                                NULL,
                                                &resultCL);

to

oclHandles.context = clCreateContextFromType(NULL,
                                                CL_DEVICE_TYPE_GPU,
                                                NULL,
                                                NULL,
                                                &resultCL);

sv650

Outcomes