Will the Intel HD Graphics OpenCL compiler support "1 work item" work groups that are float8 vectors?
clEnqueueNDRangeKernel may fail on kernel with 2D local arrays but succeed with 1D local arrays and manual index computing.
For example, the following matrix multiplication kernel fails with CL_OUT_OF_RESOURCES if USE_2D is defined and succeedes otherwise.
Matricies are [24, 72] * [24, 72]T = [24, 24] and BLOCK_SIZE = 24.
On I7-4770 (HD4600) and I7-4860EQ (HD5200) we install Win7 x64 with driver 22.214.171.12496, then install Intel SDK for OpenCL Application 2013 r3 x64. In SDK Kernel Builder Menu 'Device info', it only shows CPU support OpenCL 1.2, no GPU support information.
For I3-3120M and i5-3427U (HD4000) with same graphics driver and OpenCL SDK version, there is GPU support info show up. See attached pics.
I have discovered what I consider to be a bug in the reference tracking for cl_command_queue instances that leads to a segmentation fault under a reasonable usage scenario.
The cl_command_queue reference count is not incremented when a cl_event instance is created. It should be.
I'm having trouble debugging my offload executable in windows, and I suspect there may be a cache someplace that needs to be purged. If I create a new solution/project with the same name as my real executable, but instead only do "hello world":
Would you provide sample(Tutorial) for OpenCL 2.0 which include SVM and Pipes, you know the product of Intel® SDK for OpenCL™ Applications 2014 can support the OpenCL2.0, and we begin to use it on inter machine, but it did not provide a sample? Thank you!
I find out that main memory also stores a copy of data allocated on MIC Card when handling big data on MIC Card.
For example, given below codes, the program uses 1G RAM when pausing at "Press key to continue 1".
However after loading the 1G emtpy data to MIC Card, it uses 2G RAM at "Press key to continue 2". Meanwhile 1G of MIC Card's memory is also occupied.
At "Press key to continue 3", when the memory on MIC is free, the program occupies 1G RAM again.