Intel Sends Out Meteor Lake & Sierra Forest Patches For The GCC Compiler

Written by Michael Larabel in Intel on 14 October 2022 at 06:21 AM EDT. 7 Comments
INTEL
Intel has today sent out new compiler patches as they work to get their next-generation processors all set for the open-source GNU Compiler Collection (GCC), including the introduction of some new x86_64 instruction set extensions coming with the high core count Sierra Forest processors.

First up, Intel engineer Haochen Jiang sent out a patch adding the Meteor Lake support to the GCC compiler. That came along with adding the overdue Raptor Lake support too as part of a separate patch today. Though it's not particularly exciting as both the RPL and MTL code paths are just following the existing Intel Alder Lake (ADL) paths within the GCC compiler. So no new instruction differences nor any cost table differences or other tuning changes, at least for now.

Much more interesting is the patches sent out today by Intel for enabling Sierra Forest with the GCC compiler. Sierra Forest is Intel's E-core-only Xeon server processors expected out in 2024. Sierra Forest with just using the lower-power E cores is aiming for high core count, energy efficient servers in the data center to ultimately compete with the likes of AMD's upcoming Bergamo and the high core count AArch64 CPUs like from Ampere Computing.

The Sierra Forest compiler patches sent out today beyond enabling the "sierraforest" compiler target enable new instructions to be found with the 2024-timed E cores. The new instructions plumbed for the GCC compiler with Sierra Forest being the initial supporter are AVX-IFMA, AVX-VNNI-INT8, AVX-NE-CONVERT, and CMPccXADD.

In the AVX-512 world there is already IFMA (Integer Fused Multiply Add) for fused multiply add of integers while AVX-IFMA is adding similar support for non-AVX-512 usage. AVX-VNNI-INT8 is for VNNI INT8 AVX usage, AVX-NE-CONVERT for loading a BF16 FP element from m16 and converting to FP32, and CMPccXADD for "Compare and Add if Condition is Met".


Intel recently updated their x86_64 ISA programming reference manual with these new instructions and confirming they are being introduced with Sierra Forest and Grand Ridge. Intel Grand Ridge meanwhile is Intel's successor to the Snow Ridge processor as a higher core count Atom SoC.

It's good seeing Intel continuing in their timely push of support for new x86_64 instructions coming with their next-generation processors and getting the "meteorlake" and "sierraforest" compiler targeting in place. Meanwhile we are still waiting for AMD to post their Zen 4 "znver4" compiler patches... Today these Intel patches were just sent out in patch form but there still is time for these patches to be mainlined ahead of the GCC 13 compiler release due out early next year. GCC 13 in turn will then be found in Ubuntu 23.10 this time next year and in other Linux distributions, which should lead to this improved compiler support being found out-of-the-box by the time Sierra Forest data center processors are shipping.
Related News
About The Author
Michael Larabel

Michael Larabel is the principal author of Phoronix.com and founded the site in 2004 with a focus on enriching the Linux hardware experience. Michael has written more than 20,000 articles covering the state of Linux hardware support, Linux performance, graphics drivers, and other topics. Michael is also the lead developer of the Phoronix Test Suite, Phoromatic, and OpenBenchmarking.org automated benchmarking software. He can be followed via Twitter, LinkedIn, or contacted via MichaelLarabel.com.

Popular News This Week