| Order a report

A Guide To Server Processors

Third Edition

Published May 2014

Authors: Mike Demler, David Kanter and Kevin Krewell

Single License: $3,995 (single copy, one user)
Corporate License: $5,995

Ordering Information



Examining the Processors Powering Scalable Computing

The market for server processors is changing, creating openings for new vendors. With the emergence of mega data centers and cloud computing, server economics no longer focus on capital expenses alone. Demand for ultimate performance from a single processor has been replaced by a balanced view of capital and operating costs. Performance per watt and performance per watt per dollar are the new metrics driving purchasing decisions in large data centers. Physical density is also growing in importance, driving greater scalability and new form factors such as microservers that pack more nodes into precious rack space.

The market is moving to a new era of where backward compatibility is less important than before and innovation takes the front seat. Intel and AMD—the incumbent vendors—continue to innovate and advance their Xeon and Opteron designs, respectively. Integration, microarchitecture advances, and process technology are the primary factors when evolving these x86 processors. But new entrants are eyeing cloud-computing environments as an opening for radically different architectures or more power-efficient CPU architectures.

Product Information Tempered By In-Depth Analysis

This report covers processors designed specifically for servers. We provide detailed coverage of Intel’s Xeon E3, E5 and E7 product lines as well as its new Atom products for microservers. We cover AMD’s Opteron family, including Opteron X for microservers and the company’s new ARM processor. Other ARM-compatible products include AppliedMicro’s X-Gene, Broadcom’s Vulcan, and Cavium’s Thunder. Because other vendors entering this market are using ARM-based designs, we provide coverage of ARM’s intellectual-property cores including the 64-bit Cortex-A57. We examine what went wrong at pioneering firms such as Calxeda, Marvell, and Tilera and speculate about potential market entries by Qualcomm and Samsung. This edition also continues our coverage of coprocessors (or accelerators) for high-performance computing (HPC), including Intel’s Xeon Phi, Nvidia’s Tesla, and AMD’s FirePro.

This report analyzes each vendor and each product, probing their strengths and weaknesses and presenting key details in a consistent, easy to compare fashion. We examine processor performance, integration, power dissipation, and overall system design. Where possible, we also look at the vendors’ roadmap.

Make Informed Decisions

As the leading vendor of technology analysis for microprocessors, The Linley Group has the expertise to deliver a comprehensive look at these technologies. Our analysts use their broad experience to deliver the technical and strategic information you need to make informed business decisions. And in case you are not familiar with all of the concepts involved in processor and server designs, the report includes several introductory chapters that define and describe terms such as superscalar, multithreading, pipelines, and virtualization.

This report is written for:

  • OEMs that need to make strategic vendor selections
  • ODMs supplying cloud-computing and HPC customers
  • Data-center architects looking at alternative platforms
  • Marketing and engineering staff at companies that sell other server components
  • Financial analysts who desire a detailed analysis and comparison of both incumbent and new vendors

What's New in This Edition

“A Guide to Server Processors” has been extensively updated to include the latest vendor disclosures.

Here are some of the many changes you will find:

  • Coverage of many new products from Intel, including Atom C-series (Avoton), Xeon E5/E7 v2 (Ivy Bridge), Xeon E3 v3 (Haswell), and Xeon Phi (Knights Landing)
  • New coverage of AMD’s Bulldozer-based Opteron 6300 processors, the low-end Opteron X, and its first ARMv8 processor (Seattle)
  • New coverage of Broadcom’sVulcan CPU for future server processors
  • Updated coverage of AppliedMicro’s X-Gene processor, the server industry’s first ARMv8 product
  • Updated coverage of Cavium’s Project Thunder, a multicore ARMv8 design built off of the company’s successful Octeon architecture
  • Updated coverage of ARM’s Cortex-A57 and fabric IP
  • New coverage of AMD’s FirePro accelerators for high-performance computing
  • Extensive updates to company information, roadmaps, and analysis
  • Forecast for ARM and x86 server processors through 2016

The market for server processors is changing, creating openings for new vendors. With the emergence of mega data centers and cloud computing, server economics no longer focuses on capital expenses alone. Demand for ultimate performance from a single processor has been replaced by a balanced view of capital and operating costs. Performance per watt and performance per watt per dollar are the new metrics driving purchasing decisions in large data centers. Physical density is also growing in importance, driving greater scalability and new form factors such as microservers that pack more nodes into precious rack space.

The market is moving to a new era where backward compatibility is less important than before and innovation takes the front seat. Intel and AMD—the incumbent vendors—continue to innovate and advance their Xeon and Opteron designs, respectively. Integration, microarchitecture advances, and process technology are the primary factors when evolving these x86 processors. But new entrants are eyeing cloud-computing environments as an opening for radically different architectures and more-power-efficient CPU designs. Because the merchant server-processor market exceeds $8 billion, success requires taking only a few percentage points of share from Intel.

Having reached practical power limits, server-processor designers are increasing performance primarily by adding cores rather than increasing clock speeds. Mainstream x86 processors now offer 16 cores per chip, while ARM vendors are planning 32-core processors. Intel has recently refreshed its entire lineup, moving to 22nm FinFET technology to reduce cost and power. By moving to a finer-geometry process, vendors get more transistors in the same die area and power envelope. This additional transistor count can be used to add CPUs or to increase cache sizes. Larger caches increase performance by absorbing DRAM latency, which is not decreasing as rapidly as processor performance is growing.

For low-cost processors targeting microservers, a single-chip design is now common. ARM vendors such as AppliedMicro bring this SoC model from their embedded products, and both AMD and Intel have responded with SoC products of their own. Intel’s Avoton processor, for example, integrates eight lightweight Atom CPUs along with the traditional north-bridge and south-bridge logic as well as four Gigabit Ethernet ports.

Intel also offers a broad line of higher-performance server processors. These devices include, for single-socket servers, the Xeon E3 v3 family based on the company’s newest Haswell CPU design. For the popular dual-socket configuration, Intel offers the new Xeon E5 v2, using the same 22nm process as the Xeon E3 but the slightly older Ivy Bridge CPU. This product, code-named Ivytown, scales to 12 cores. For top-end performance in systems with four or eight coherent processors, the Xeon E7 v2 delivers up to 15 cores at 2.8GHz in a 155W thermal envelope. By adding reliability, availability, and serviceability (RAS) features to the E7 line, Xeon now serves mission-critical designs that formerly required Itanium (IA-64) processors.

AMD’s Opteron chips have fallen far behind Xeon in both performance and market share. Using a 32nm SOI process, they generally deliver about half the performance per watt of comparable 22nm Xeon processors. The new Piledriver-based Opterons do little to improve on this short¬coming. For microservers, AMD also offers the Opteron X1150, based on its low-end Jaguar CPU, that falls well behind Intel’s Avoton. AMD’s products are most competitive for applications that don’t need maximum CPU performance because they are I/O or memory bound.

After initial enthusiasm over 32-bit processors from vendors such as Calxeda, Marvell, and Tilera, data-center operators have standardized on the new 64-bit ARMv8 instruction set, leaving out these RISC pioneers. The first ARMv8 server processors will enter production in 2014, led by AppliedMicro’s X-Gene. This eight-core product could outperform Intel’s eight-core Avoton, but the 40nm X-Gene is likely to use considerably more power. AppliedMicro is already sampling a 28nm X Gene that should reduce the power gap.

AMD, ironically, is the other leader in the ARMv8 race. As part of its “ambidextrous” strategy, the company is sampling a processor with eight Cortex-A57 CPUs that is due for production by the end of 2014. This SoC design should better match up against Avoton in power efficiency. Until these ARMv8 vendors release full specifications and performance scores for their products, however, a detailed comparison against Intel’s products is impossible.

Cavium is also developing an ARMv8 processor, code-named Thunder, scheduled for production in 2015. Broadcom has announced a high-end ARMv8 CPU known as Vulcan that could approach Xeon in performance, but it is even further out in time. Other ARMv8 vendors such as Qualcomm and Samsung could jump into the server market over time, creating additional competition for Intel.

Having 96% market share, Intel faces no effective competition in many segments of the server market and is able to name its own prices. This situation is good for Intel shareholders but bad for Intel customers. These customers are actively supporting new competitors, as well as AMD, in hopes of putting pressure on Intel to moderate its processor prices. ARM competition has already forced Intel to increase its pace of innovation, introducing Atom-based SoCs to the microserver market. It may take a few more years, but ARM vendors are likely to eat into Intel’s market share and return balance to the server-processor market.

List of Figures
List of Tables
About the Authors
About the Publisher
Preface
Executive Summary
1 Processor Technology
Processor Basics
Central Processing Unit (CPU)
Caches
MMUs and TLBs
Bus Bandwidth
CPU Microarchitecture
RISC Versus CISC
Endianness
Scalar and Superscalar
Instruction Reordering
Pipelining and Penalties
Branch Prediction
Server Processors and Technologies
What Is a Server Processor?
Multicore
Multithreading
System Buses
Memory Subsystem
PCI Express
Server Benchmarks
SPEC Benchmarks
TPC Benchmarks
VMmark
HPL
ApacheBench
2 Instruction Sets
x86 Instruction Set
Background
Initial Instruction Set
ISA Extensions
ARM Instruction Set
Background
Initial Instruction Set
ARMv7
ARMv8
3 Server System Technology
Basic Server Architecture
Main Memory
System-Logic Chipset
Baseboard-Management Controller
Multisocket System Design
Storage
RAID
Storage Interfaces
High-Performance Computing
InfiniBand
RDMA Over Ethernet
MPI and OFED
Networking
Storage Networking
Form Factors
Operating Systems
Windows Server
Linux Server
Virtualization
Hypervisor Software
4 Technology and Market Trends
Technology Trends
x86 Versus ARM
SoC Integration
The Main-Memory Bottleneck
Microservers
System Fabric
Scale Up Versus Scale Out
Cloud-Computing Workloads
High-Performance Computing
Market Outlook
Cloud Computing
Open Compute
Market Forecast and Segmentation
Processor Revenue and ASP
Addressable Market for ARM
Market Share
5 Intel
Company Background
Product-Line Overview
Key Features and Performance
Atom-Based Processors
Haswell-Based Xeon Processors
Ivy Bridge-Based Xeon Processors
Itanium Processors
Internal Architecture
Ivy Bridge and Haswell
Silvermont
System Design
Xeon E3 v3
Atom C2500
Xeon E5 v2
Xeon E7 v2
Product Roadmap
Atom and Xeon E3
Xeon E5
Xeon E7
Itanium
Conclusions
6 AMD
Company Background
Key Features and Performance
Mainstream Server Processors
Microserver Processors
Internal Architecture
Bulldozer and Piledriver CPUs
Jaguar CPU
ARM Processor
System Design
Opteron System Design
A1100 System Design
Product Roadmap
Conclusions
7 AppliedMicro
Company Background
Key Features and Performance
Design Details
Product Roadmap
Conclusions
8 ARM
Company Background
Key Features and Performance
Internal Architecture
System-on-a-Chip Design
Development Tools
Product Roadmap
Conclusions
9 Cavium
Company Background
Key Features and Performance
Conclusions
10 HPC Coprocessor Vendors
Intel Xeon Phi
Company Background
Key Features and Performance
Internal Architecture
Programming Model and Tools
Product Roadmap
Conclusions
Nvidia Tesla
Company Background
Key Features and Performance
Design Details
Product Roadmap
Conclusions
AMD FirePro S10000
Company Background
Key Features and Performance
Design Details
Product Roadmap
Conclusions
11 Other Vendors
Broadcom
Company Background
Key Features
Conclusions
Calxeda
Marvell
Qualcomm
Samsung
Tilera
12 Processor Comparisons
Microserver Processors
Performance
Integration
Single-Socket Platforms
Performance
Integration
Two-Socket Platforms
Performance
Integration
Four-Socket Platforms
Performance
Integration
13 Conclusions
Market Outlook
Near-Term Trends
ARM vs. Intel
Long-Term Trends
Vendor Outlook
Intel
AMD
AppliedMicro
Other ARM Vendors
HPC Coprocessors
Closing Thoughts
Appendix: Further Reading
Index
Figure 3-1. Typical single-processor server architecture.
Figure 3-2. Typical multisocket server architecture.
Figure 3-3. Rack-mount servers and a standard-size rack.
Figure 3-4. IBM's BladeCenter H.
Figure 3-5. Typical blade-server architecture.
Figure 4-1. Dell PowerEdge C5000 microserver and HP Moonshot.
Figure 4-2. Server-processor shipment forecast and segmentation.
Figure 4-3. Server-processor revenue by form factor.
Figure 4-4. Available market opportunity for non-x86 processors.
Figure 4-5. Server-processor revenue share for x86, 2012-2013.
Figure 5-1. Intel server-processor roadmap.
Figure 5-2. Block diagram of Intel Ivy Bridge microarchitecture.
Figure 5-3. Block diagram of Intel Ivytown processor.
Figure 5-4. Block diagram of Intel Silvermont microarchitecture.
Figure 5-5. Server design based on Intel Xeon E3-1200 v3.
Figure 5-6. Server design based on Intel Atom C2570.
Figure 5-7. Dual-socket server design based on Intel Xeon E5-2600.
Figure 5-8. Four-socket server design based on Intel Xeon E7 v2.
Figure 6-1. Block diagram of AMD Bulldozer/Piledriver CPU module.
Figure 6-2. Block diagram of AMD Bulldozer/Piledriver microarchitecture.
Figure 6-3. Block diagram of AMD Opteron X server chip.
Figure 6-4. Block diagram of AMD Opteron A1100 server chip.
Figure 6-5. AMD Opteron 4300 two-socket system design.
Figure 6-6. AMD Opteron 6300 four-socket system design.
Figure 7-1. Block diagram of AppliedMicro X-Gene CPU.
Figure 7-2. Block diagram of AppliedMicro X-Gene processor.
Figure 8-1. Block diagram of ARM Cortex-A57 microarchitecture.
Figure 8-2. Block diagram of ARM Cortex-A57 in an SoC.
Figure 9-1. Conceptual block diagram of Cavium Thunder.
Figure 10-1. Intel Xeon Phi coprocessor card.
Figure 10-2. Microarchitecture of Intel Xeon Phi core.
Figure 10-3. Conceptual block diagram of Xeon Phi coprocessor.
Figure 10-4. Block diagram of Tesla GK110 SMX array.
Figure 10-5. Block diagram of GCN compute unit.
Table 1-1. Selected SPEC benchmarks.
Table 5-1. Summary of Intel x86 server portfolio.
Table 5-2. Key parameters for selected single-socket Intel server processors.
Table 5-3. Key parameters for selected dual-socket Intel Xeon processors.
Table 5-4. Key parameters for selected multisocket Intel Xeon processors.
Table 5-5. Key parameters for selected Intel Itanium 9500-series processors.
Table 6-1. Key parameters for AMD Piledriver-based Opteron processors.
Table 6-2. Key parameters for selected AMD Opteron processors.
Table 6-3. Key parameters for AMD microserver processors.
Table 6-4. Key parameters for AMD SR56x0 north-bridge chips.
Table 6-5. Key parameters for AMD SP5100 south-bridge chip.
Table 8-1. Key parameters for ARM Cortex-A15 and Cortex-A57.
Table 10-1. Key parameters for Intel Xeon Phi coprocessor cards.
Table 10-2. Key parameters for Nvidia Tesla coprocessor cards.
Table 10-3. Key parameters for AMD FirePro S-series coprocessor cards.
Table 11-1. Server-processor vendors and products.
Table 12-1. Comparison of microserver processors.
Table 12-2. Comparison of high-performance single-socket processors.
Table 12-3. Comparison of processors for dual-socket servers.
Table 12-4. Comparison of processors for four-socket servers.