Intel announces Xeon 6 performance core processor

     Intel officially released the Xeon 6 performance core processor for the data center market, with 3 computing chip units and up to 128 cores, which is a significant increase in the number of cores compared to the previous generation. While performance is improved, the average TCO is reduced by 30% compared to the previous generation. The Xeon 6900P series processor (codenamed Granite Rapids-AP) was announced.

     For the cloud computing field, Xeon 6 processors have improved the number of cores per channel, single-core performance, and per-watt performance compared to the previous generation of processors, which can help cloud service providers reduce computing costs. Compared to fifth-generation Intel Xeon scalable processors, Xeon 6 processors have up to twice the number of cores per path, up to 1.2 times the average single-core performance improvement, up to 1.6 times the average performance improvement per watt, and an average TCO savings of 30% for the same level of performance

     In the face of data center requirements to improve data transfer efficiency, today announced the Xeon 6900P series processor has a series of performance updates. The processor is equipped with 128 cores, supports DDR5 memory up to 6400MT per second, MRDIMM memory up to 8800MT per second, six UPI 2.0 links (up to 24 GT per second), 96 PCIe 5.0 or 64 CXL 2.0 channels, 504MB L3 cache, and more. Intel Matrix Extensions (Intel AMX) with support for FP16 data formats provide MRDIMM options for memory bandwidth-sensitive workloads such as AI and scientific computing, with new support for CXL 2.0.

     At this event, Intel eco-partners shared their solutions based on Xeon 6 performance core products. Among them, Alibaba Cloud cubic architecture and Xeon processor continue to iterate evolution, launched a generation of Panjiu computing server, and the ninth generation of Alibaba Cloud Intel platform enterprise computing instance products based on Xeon 6 performance core; Superfusion brings its first memory pool to support CXL2.0+, allowing for flexible allocation and sharing of memory to fully improve system performance. Volcano Engine has been optimized for architecture and performance based on Xeon 6 performance core processors, doubled the number of CPU cores, and adopted a dual-single-path architecture design, so that its fourth-generation cloud server instance has greatly improved computing density and application performance while ensuring stability.