Intel Xeon

Overview

The purpose of this HEXUS.help guide is to delineate the features present in Intel's quad-core Xeon processors and why they're a good fit for all business users looking towards upgrading their server or workstation environment.

Introduction

Intel introduced the industry's first quad-core processor for mainstream servers on November 14th, 2006. Codenamed Clovertown, the quad-core 5300-series Xeons are based on Intel's advanced Core microarchitecture and can be implemented in both single- and dual-processor systems.

The architecture - what is a quad-core Xeon?

With the release of Intel's Core microarchitecture, Intel is now utilising the same basic processors for both its high-end consumer-level and mainstream server/workstation environments. The main architectural difference between the quad-core 5300-series Xeons and their Core 2 Quad consumer counterparts is in the ability to run in a multi-socket system and to support a faster Front-Side Bus. For quad-core models - and this applies to both desktop and server models - each physical CPU is, effectively, a couple of dual-core models that sit on a single package.

In a little more detail, quad-core Xeons take in a couple of dual-core Xeon dies that fall under the Woodcrest design and place them side-by-side on the same processor package, such that both dies communicate with one another and the system via, and only via, a shared FSB. What you'll also notice is that Intel is using the same fundamental building blocks on its mobile CPUs, so the Core architecture is a top-to-bottom design.

Each quad-core Xeon, then, takes the goodness that is present in the Intel Core microarchitecture, doubles it and places them together, thereby offering four cores in one physical processor.

The two dual-core dies benefit from the performance-enhancing features endemic to the Core microarchitecture series, including a wide dynamic execution architecture, where the processor is architected to fetch, dispatch, execute and retire up to four full instructions simultaneously, offering a 33% boost over, say, all other three-issue CPUs that make up the competing market.

Core microarchitecture CPUs are also the beneficiaries of Intel's Smart Memory Access and Advanced Smart Cache technologies, whereby each of the quad-core Xeon's processors can dynamically allocate their 4MiB if L2 cache between individual cores on the same die, although, of course, not between the two physical dies that make up an Intel quad-core Xeon.

Other Intel enhancements include Virtualisation Technology, and 64-bit processing support, along with the ability to process a 128-bit SIMD instruction in a single clock cycle, rather than requiring the incumbent two clocks that other current generations employ.

Further, and particularly important to the mainstream server market, Intel's quad-core Xeon architecture is designed with intelligent power capability that power-gates parts of the CPU which aren't in use; there's little need for four cores running at full power in single-threaded applications, for example.

Models and chipsets

As with Woodcrest-based Xeons from which they're based upon, Intel's quad-core Xeons are packaged in an LGA771 form factor, making them a drop-in upgrade for most current Woodcrest-supporting chipsets. The current line-up of mainstream server quad-core CPUs is as follows:

Intel also introduced 3200-series quad-core Xeons on January 7th, 2007. These are identical to the 5300-series in most aspects but only support single-CPU operation. The X3210 and X3220 run at 2.13GHz and 2.40GHz, respectively, and both support a maximum 1066MHz FSB. They only differ from the desktop Kentsfield dual-core processors by being packaged in a different form factor: LGA771 vs. LGA775.

With power requirements at the forefront of most large-scale I.T. deployment teams' minds, Intel has packed in significant processing power with an energy-efficient design, thanks to a low-ish TDP and, at peak performance, four efficient cores concurrently handling computational load. Considered as a platform, a 2P Intel quad-core Xeon solution provides a decent performance-per-watt metric, even if it is somewhat hampered by the wattage-eating FB-DIMMs that are required on supporting chipsets.

Intel provides support with its 5000P/V/X chipsets that can host two quad-core Xeon processors, thereby offering an 8-core solution from a single board. All the chipsets support dedicated, independent FSBs for each physical processor, too.

The Intel 5000P chipset is designed for performance servers and supports all quad-core Xeons. Further, it provides three PCI-Express x8 links and, in common with the remainder of the 5000-series, support for Fully Buffered DDR2 DIMMs with a maximum, effective speed of 667MHz. It also supports Intel's 6700PX 64-bit PCI hub. The 5000V chipset is designed for the value server and replicates most of the 'P's feature-set but provides only a single PCI-Express x8 link.

For the workstation market, Intel offers the 5000X chipset that, in addition to the 5000V's single PCI-Express x8 link, also adds support for a regular x16 link, suitable for running, at full speed, any of today's PCI-Express-based graphics cards.

Usage and summary

Due to the multi-core nature of Intel's 5300-series Xeon processors any application(s) that are heavily threaded will see significant benefits in performance when run on servers/workstations that can take full advantage of the quad cores present in each processor package, with the benefits more pronounced if utilised in a 2P environment and with software that can concurrently execute on each of the 8 available cores.

Intel made rapid gains in server-level processor performance with the introduction of its Woodcrest dual-core CPUs. That progress has been further underscored with the availability of quad-core processors based on the same technology.