compute-runtime: Unable to get working on Broadwell i7 5600U, or missing steps?
I have built and installed without any error (following exactly the steps from https://github.com/intel/compute-runtime#building) on my laptop (ubuntu 17.10, Broadwell i7-5600U), but clinfo returns 0 platform found. Checking clinfo calls with strace shows it is properly picking-up /opt/intel/opencl/libigdrcl.so.
Is Broadwell supported right now? Readme states it is https://github.com/intel/compute-runtime#supported-platforms, but the “GenX” naming is unclear to me, and https://ark.intel.com/products/85215/Intel-Core-i7-5600U-Processor-4M-Cache-up-to-3_20-GHz reports “5th gen”. And the GPU is Intel HD Graphics 5500, whose naming would be consistent with 5th gen, but it’s not listed on https://www.intel.com/content/www/us/en/architecture-and-technology/visual-technology/graphics-overview.html.
So, am I trying to get it working on an unsupported platform, or did I missed anything to get that working ? Thanks!
About this issue
- Original URL
- State: closed
- Created 6 years ago
- Comments: 97 (37 by maintainers)
@lissyx I added few additional improvements for Kmd Notify mechanism (0e41bc7). Performance should be improved significantly.
Pointer that is being passed to clEnqueueWriteBuffer call is allocated in read_only memory region. Such pointer cannot be used on the GPU with the current implementation of Kernel Mode driver, that results in exec failure. This is also why doing the copy on the CPU is successful here, as CPU has no problems in reading from this memory. We are currently working on a solution for this problem.
One small clarification, long compilation happens inside the OpenCL driver as a part of the setup. Execution on the GPU takes around 5 seconds. One way to mitigate this heavy compilation cost is to cache compiled binaries and re-use them in subsequent application runs, but that needs to be developed as a part of the actual application.
This flag only changes how writeBuffer operation is being handled, instead of going to GPU it uses CPU for transfer. It is unlikely that it affects performance so significantly, therefore I think problem is somewhere else. I think your investigation may provide valuable feedback.
You did it correctly. Manual build of Khronos ICD Loader is required. Neo CMake buildsystem won’t build it automatically.