New update for OpenCL™ runtime to Xeon and Xeon Phi is available - Check it out

New update for OpenCL™ runtime to Xeon and Xeon Phi is available - Check it out

OpenCL runtime 14.2 with support to CPU and Intel® Xeon Phi™ coprocessor is now available. Designed to support high performance applications with Intel’s server line of processors. This release provides new features for better performance and wider support for SPIR 1.2.

What’s new?

  • SPIR 1.2 is now conformant on both CPU and Intel Xeon Phi coprocessor.
  • Support for offline kernel compilation to the final executable binary is now available on both CPU and Xeon Phi. This enables a faster build time in application run-time.
  • Improved kernel invocation time on Intel Xeon Phi coprocessor device.
  • More optimized compiler vectorizer
  • New environment variable CL_CONFIG_CPU_VECTORIZER_MODE
  • User logger for API tracing and debugging functional failures in OpenCL applications

You can download the new release directly at: OpenCL™ Drivers and Runtimes for Intel® Architecture.

Regards,

Arnon

6 posts / 0 new
Last post
For more complete information about compiler optimizations, see our Optimization Notice.

Hi, dunno if this is the right to place to report a little problem, but I updated the Windows 64 bit OpenCL CPU-only runtime version from 14.1 to 14.2 and now I'm getting the following warning when I initialize the OpenCL context:

WARNING: Linking two modules of different data layouts!
WARNING: Linking two modules of different target triples: C:\Program Files (x86)\Common Files\Intel\OpenCL\bin\common\clbltfnshared.rtl: 'i686-pc-win32-elf' and 'x86_64-pc-win32-elf'

The warning appears only when I run my application in 64 bit. In spite of the warning, the application works as expected. This is the first time I see this message since it never popped up with the previous version. Thanks!

 

Hi Achille,

Thank you for reporting the issue. It will be fixed by the next release.
Meanwhile, this message should be simply ignored - it's just a warning which doesn't affect any functionality (as you correctly mentioned).

Thanks,
Yuri

Hi,

seeing SPIR 2.0 provisional extension being released I hope once Intel ships a OpenCL 2.0 runtime, beta support for SPIR 2.0 is already there,..

 

 

 

 

Don't want to create a new thread for this

This is a suggestion for machines that don't have OpenCL driver installed (because they don't have a GPU)

It's sometimes difficult on linux for us as we do OpenCL when there is a GPU or CPU only (no OpenGL when there is no GPU or user selects to run in CPU) and we do plugins and end users context totally varies (inclduing flavor of linux). You would think I could simply just not have the library load when I don't need it, that aside...

My comment is your driver is a lot less convenient to install than the AMD package for your own CPU when there is no GPU (like on a XEON and we have a lot of render farm clients on that).

I have to say to customers things like:  Find the OpenCL drivers at https://software.intel.com/en-us/articles/opencl-drivers
Then they come back it suddenly it has the following dependencies:
                yum install numactl
                yum install redhat-lsb-core
                intel mpss from: https://software.intel.com/en-us/articles/intel-manycore-platform-software-stack-mps
uh?  (this is this week with a CentOS 6.5 client)

Thank you,

(Also I don't see a way to press a checkbox to subscribe to this thread :) ).

Thanks for the information, 

We aware of some install issues, and we will work to improve that.

Arnon

Leave a Comment

Please sign in to add a comment. Not a member? Join today