Intel Compute Runtime / IGC Shifts To LLVM Clang 10

The Intel Graphics Compiler (IGC) and now in turn the Intel Compute Runtime have updated their compiler stack against the newly released LLVM Clang 10.0.

LLVM 10.0 was released at the end of March with many new features. Intel's open-source crew has punctually updated their graphics compiler code-base against LLVM 10.0 and now beginning with today's Intel Compute Runtime release and their reference binaries now have the updated compiler at play.

It will be interesting to benchmark this Intel Compute Runtime release with OpenCL and oneAPI Level Zero to see if their upgrade from LLVM 9 to LLVM 10 is yielding any performance benefits in their graphics compute stack. In any case, it's great to see them continuing to remain close to upstream compared to many LLVM-derived projects often trailing new releases by months.

With this Intel Compute Runtime 20.14.16441 release they have also now enabled performance counters support both for OpenCL and oneAPI Level Zero usage.

Those wanting to try out the new Compute Runtime for the flagship open-source OpenCL + oneAPI L0 experience, see this GitHub release page for all the details and Ubuntu binaries along with the source code.