Intel Launches 4th Gen Xeon Scalable "Sapphire Rapids", Xeon CPU Max Series

Intel also provided a slide with their analysis of Sapphire Rapids against AMD's Genoa CPUs. It's heavily accelerator focused as well as showing the areas where Intel Xeon CPUs have provided features long before AMD, such as with the seven year headstart on AVX-512.

Due to needing more time for the software ecosystem to take advantage of accelerators, the full advantages of Sapphire Rapids won't be realized at launch. It will be very interesting though as more upstream open-source software begins making use of the new IP and benchmarking the performance benefits.

Intel can be heavily applauded for all of the software investments, particularly around Linux and open-source software... From those that read Phoronix, you know about the immense software contributions made by Intel over the years and the work that remains ongoing. Going back to 2021 is when Intel added the initial Sapphire Rapids support to the GCC compiler and began plumbing AMX. In comparison, AMD Zen 4 (znver4) compiler patches only began arriving after the EPYC Genoa launch -- an unfortunately common trend of AMD compiler support only being published post-launch where Intel aims to get things out well early. Already in upstream GCC and LLVM Clang, Intel has Emerald Rapids, Granite Rapids, Sierra Forest, Meteor Lake, Grand Ridge, and other new support premiering at the same time as the Znver4 target. Intel's software work in so many areas can be praised and is also all the more important as they open the door on these accelerators.

Intel has also been contributing a lot to the Linux kernel around TDX support, so it came as a surprise that it will be "limited" this generation but at least that upstream kernel support is there for when there is more broad TDX availability and can be used today by the cloud partners with TDX access.