Nehalem (microarchitecture)

Nehalem (microarchitecture)
Name	Nehalem
Designer	Intel
Bits	64-bit
Introduced	November 2008
Design	Superscalar, Out-of-order execution
Predecessor	Penryn
Successor	Sandy Bridge

Nehalem (microarchitecture). Nehalem is the codename for a microarchitecture developed by Intel as the successor to the Penryn-based Core 2 processors. First launched in November 2008, it represented a major architectural overhaul, re-integrating a memory controller onto the CPU die and introducing a new point-to-point interconnect called QuickPath Interconnect. The architecture formed the foundation for the first-generation Core i7, Core i5, and Xeon 5500 series processors, marking a significant shift in Intel's product strategy and performance leadership.

Overview

The development of Nehalem was led by Intel under the direction of architects like Glenn Hinton and was a direct response to the competitive pressure from AMD's K10 architecture and its integrated memory controller. It was manufactured on Intel's then-new 45 nm high-k metal gate process technology. The design philosophy moved away from the shared front-side bus model of previous NetBurst and Core architectures, embracing a more modular, scalable approach. Key initial products included the desktop Core i7-900 series for the LGA 1366 socket and the server-oriented Xeon 5500 series.

Features

Nehalem introduced several landmark features that defined modern Intel processors. The most significant was the integration of a memory controller supporting DDR3 SDRAM, which drastically reduced memory latency. It also debuted QuickPath Interconnect (QPI), a high-speed point-to-point link that replaced the aging front-side bus for processor-to-processor and processor-to-I/O Hub communication. Other major innovations included the return of Simultaneous multithreading (marketed as Hyper-threading) to the Core brand, an improved power management system known as Turbo Boost, and a new multi-level shared cache hierarchy with an inclusive L3 cache.

Processor cores

The Nehalem core was an evolution of the Penryn core but with substantial enhancements. Each core featured a new 4-wide decoder and an improved out-of-order execution engine. The re-addition of Simultaneous multithreading allowed each physical core to handle two threads, effectively doubling the thread count on models like the Core i7-920. The core also incorporated new SSE4.2 instructions, accelerating specific string and XML processing tasks. Core variants ranged from high-performance quad-core designs for the Bloomfield desktop chips to scalable multi-socket configurations for the Gainestown Xeon processors.

Memory and I/O

The memory subsystem was radically redesigned. The on-die memory controller supported triple-channel DDR3 SDRAM on high-end desktop platforms like LGA 1366, providing significantly higher bandwidth. For communication, the QuickPath Interconnect (QPI) operated at speeds up to 6.4 GT/s, linking the CPU to other processors and the Platform Controller Hub (PCH), which itself consolidated functions previously handled by the northbridge. This modular I/O architecture, separating memory and processor interconnect from general I/O, provided greater scalability for server and workstation platforms like the Tylersburg platform.

Performance and reception

Upon release, Nehalem-based processors like the Core i7-965 Extreme Edition received widespread critical acclaim for delivering substantial performance gains, particularly in multi-threaded applications and memory-intensive workloads. Reviewers from publications like AnandTech and Tom's Hardware highlighted the massive leap in memory bandwidth and the effectiveness of Hyper-threading and Turbo Boost. The architecture restored Intel's clear performance leadership across consumer and enterprise segments, decisively outpacing contemporary offerings from AMD, such as the Phenom II. It set new standards for server efficiency with the Xeon 5500 series.

Successors and variants

Nehalem was followed by the Westmere microarchitecture, a "tick" in Intel's tick-tock model that shrunk the design to a 32 nm process and introduced support for AES-NI instructions. The direct architectural successor was Sandy Bridge, which introduced a new ring bus interconnect and integrated graphics. Several notable variants of Nehalem were produced, including the budget-oriented Lynnfield and Clarkdale processors for the LGA 1156 socket, and the high-end desktop Gulftown six-core Core i7-980X, which was based on the Westmere shrink.

Category:Intel microarchitectures Category:2008 introductions