Intel updated its Xeon roadmap yesterday at a virtual DCAI (Data Center and AI Investor Webinar), confirming that it is currently working on a 144-core Xeon processor called Sierra Forrest and is expected to arrive in the first half of 2024. . Intel hopes to launch another pair of Xeon products, Emerald Rapids (Q4 2023) and Granite Rapids, shortly after Sierra Forest.
Another surprise is Intel’s adoption of performance/efficiency cores at the data center level. We know that AMD will launch a 128 core Bergamo CPU based on the Zen 4c architecture later this year.
Both P cores and E cores do not exist in the same silicon (such as consumer CPUs). It also has its own set of challenges. Both have to deal with fixed power and transistor budgets.
A few Power cores with all bells and whistles (like AVX-512) clocked at higher speeds are just as many with more efficiency cores with all bells and whistles (like single thread) at less clock speeds generates/consumes
Then cache (memory closest to the core), latency (moving data to and from different parts of the CPU), chip yield (how many usable silicon chips are produced), available memory bandwidth wide (up to main system memory). etc.
Does it matter to anyone else? Rumors of a 120-core Xeon CPU We finally went to bed yesterday, and while this race makes for funny headlines, it’s not a big deal for non-hyperscalers. You can’t (edit: in a few years you might be able to bid on Ebay or buy from Aliexpress). Systems that are meaningless to the end user and cost hundreds of thousands of dollars per part (including new breeds of DDR memory modules).
The high core count, low power strategy adopted by both Intel and AMD will help contain the Arm threat at this time. There are at least half a dozen competitors who want to grab most of the pie (Huawei, Nvidia, Ampere) or reduce their reliance on Intel and AMD (Amazon, Tencent, Google).
Graviton on AWS has 64 cores, but is expected to grow depending on market demand. Ampere already has 128 cores, and Arm’s Neoverse Poseidon platform, announced in September 2020, aims to offer higher core densities (higher than N1’s 128 core limit) when deployed.
Increased core count is important for applications such as web hosting (virtual private servers or VPS), virtual desktops (VDI), and other applications that don’t consume a lot of power and don’t mind living in a multi-tenant environment .
websitebuildersnow
GIPHY App Key not set. Please check settings