Intel's OpenCL Beignet Implementation Expands Support For Doubles
Intel's Beignet open-source project that provides an OpenCL implementation for Iris/HD Graphics on Linux has landed many commits over night.
The 70+ commits are about expanding support for the double-precision floating-point data type.
It appears the support for doubles in Beignet is now much more vetted but have yet to see any Beignet developers make any formal comment on the patches or in the mailing list. The latest Beignet patches can be found in Git as we await the project's next release. It was just last week as well that they landed LLVM 4.0 support for their back-end.
The 70+ commits are about expanding support for the double-precision floating-point data type.
It appears the support for doubles in Beignet is now much more vetted but have yet to see any Beignet developers make any formal comment on the patches or in the mailing list. The latest Beignet patches can be found in Git as we await the project's next release. It was just last week as well that they landed LLVM 4.0 support for their back-end.
Add A Comment