This week, Intel provided more information regarding the Software Defined Silicon (SDSi) capability of its Xeon Scalable processors for the next generation, as well as the formal name of this technology. With this technology, system administrators will be able to pay extra to enable specialised accelerators included inside Intel’s 4th Generation Xeon Scalable “Sapphire Rapids” CPUs. The technology will subtly go by the name “Intel On Demand.”
Updates to Intel’s SDSi patches, which were incorporated into Linux 5.18 this week, provide more information about this capability than the patches themselves, according to Phoronix (opens in new tab). Intel’s 4th Generation Xeon Scalable Sapphire Rapids(opens in new tab) CPUs (and likely their successors) will be able to perform the following thanks to a piece of software called Intel On Demand:
We do not yet know which features Intel will let activating after a purchase, which is the biggest mystery surrounding the Intel On Demand system. We are aware that Sapphire Rapids from Intel features a number of specialised acceleration technologies, though. In order to speed up particular tasks, the list also contains Intel Data Streaming Accelerator (DSA), Dynamic Load Balancer (DLB), Intel In-Memory Analytics Accelerator (IAA), and Intel QuickAssist Technology (QAT).
Meanwhile, the fact that the software will need to discover which capabilities are physically supported by a processor and hide those not supported means that not all Intel Xeon Scalable ‘Sapphire Rapids’ processors will be created equal. Some CPU models may not gain support for certain features even by using Intel On Demand software.
Not all users will simultaneously need AMX, DLB, DSA, IAA, and QAT. But when Intel formally introduces its next-generation Xeon Scalable CPUs on January 10, it will undoubtedly announce which features will be enabled by default on all SKUs and which must be configured via the IOD software.