Intel CM Compiler Updated For New Platforms, Including Ponte Vecchio "XT"
Intel on Tuesday released CM Compiler 1.0.119 as their LLVM-based open-source compiler implementing their C for Metal programming language. The CM Compiler goes back to their HD graphics days for offering a new GPU kernel programming language. If the C For Metal compiler doesn't ring a bell for you among all the different GPU computing efforts these days even within Intel alone, the former 01.org project page describes it as "a programming language that allows for creation of high-performance compute and media kernels for Intel® GPUs using explicit SIMD programming model. CM is based on restricted C++ specification, with extensions to support new SIMD constructs and data types, inline assembly, and access to architecture-specific hardware features."
The new Intel CM Compiler 1.0.119 release adds in platform support for Rocket Lake, Alder Lake, DG2 / Alchemist, PVC, and PVC XT. The Rocket Lake support is well overdue and Alder Lake is already shipping, but at least Xe HPG DG2/Alchemist support is ready ahead of those cards shipping. There is also the Xe HPC "Ponte Vecchio" as part of this enablement too.
What is perhaps most interesting is the "PVCXT" addition too as at least from my open-source/Linux side monitoring I don't recall much/any talk around Ponte Vecchio XT. This compiler work pegs Ponte Vecchio XT as a B stepping (or Revision 3 and higher) while Ponte Vecchio (non-XT) as stepping A. We'll see if this XT variant is just some minor engineering difference or if there is something more to come.
The CM Compiler 1.0.119 release is available from GitHub in source form as well as Ubuntu and RHEL/CentOS binary packages. Besides all the new platform support, this update brings new language macros for checking on additional hardware capabilities and a variety of fixes.
1 Comment
