AMD 3800 - Processor - 1 x Athlon 64 Instruction sheet

Publication #: 30579 Revision: 3.74
Issue Date: March 2007
Mark W. Welker
Johann Pais
Advanced Micro Devices, Inc.
One AMD Place
Sunnyvale, CA 94088
AMD Processor
Performance
Evaluation Guide

© 2003 - 2007 Advanced Micro Devices, Inc. All rights reserved.
The contents of this document are provided in connection with
Advanced Micro Devices, Inc. (“AMD”) products. AMD makes no repre-
sentations or warranties with respect to the accuracy or completeness
of the contents of this publication and reserves the right to make
changes to specifications and product descriptions at any time without
notice. The information contained herein may be of a preliminary or
advanced nature and is subject to change without notice. No license,
whether express, implied, arising by estoppel or otherwise, to any intel-
lectual property rights is granted by this publication. Except as set forth
in AMD’s Standard Terms and Conditions of Sale, AMD assumes no lia-
bility whatsoever, and disclaims any express or implied warranty, relat-
ing to its products including, but not limited to, the implied warranty of
merchantability, fitness for a particular purpose, or infringement of any
intellectual property right.
AMD’s products are not designed, intended, authorized or warranted for
use as components in systems intended for surgical implant into the
body, or in other applications intended to support or sustain life, or in
any other application in which the failure of AMD’s product could create
a situation where personal injury, death, or severe property or environ-
mental damage may occur. AMD reserves the right to discontinue or
make changes to its products at any time without notice.
Trademarks
AMD, the AMD Arrow logo, AMD Athlon, and combinations thereof, Cool’n’Quiet and 3DNow!, are trademarks
of Advanced Micro Devices, Inc.
HyperTransport is a licensed trademark of the HyperTransport Technology Consortium.
Microsoft and Windows are registered trademarks and Windows Vista is a trademark of Microsoft Corporation.
3DMark and Futuremark are registered trademarks of Futuremark Corporation.
PCI-X is a registered trademark and PCIe is a trademark of the PCI-SIG.
NVIDIA is a registered trademark and nForce is a trademark of NVIDIA, Inc.
Business Winstone, Content Creation Winstone, and Winstone are registered trademarks of Ziff Davis
Publishing Holdings Inc. in the U.S. and other countries.
BAPCO and SYSmark are registered trademarks of Business Applications Performance Corporation.
3DMark and Futuremark are registered trademarks of Futuremark Corporation.
Pentium is a registered trademark of Intel Corporation.
Other product names and company names used in this publication are for identification purposes only and
may be trademarks of their respective companies.
Tests performed without independent verification by DivXNetworks, Inc., its parents, subsidiaries, and
affiliates. DivXNetworks, Inc., its parents, subsidiaries, and affiliates make no representation or warranty as
to the results of the tests.
Test results have not been verified by PC World, and neither PC World nor International Data Group, Inc.
makes any representations or warranties as to the accuracy of the test results.

AMD Processor Performance Evaluation Guide30579 Rev. 3.74 March 2007
Contents 4
Contents
Revision History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
About This Document . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Audience. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12
Life of Document . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12
AMD64 Single Socket Processor Architecture . . . . . . . . . . . . . . . . . . . 13
AMD Quad FX Platform with
Dual Socket Direct Connect Architecture . . . . . . . . . . . . . . . . . . . . . . . 17
Benchmarking Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Benchmark Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18
Benchmarking System Configuration . . . . . . . . . . . . . . . . . . . . . . . . . 21
Test System Configurations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21
Windows Vista™ Installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .22
BIOS Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .22
RAID Configuration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .24
Operating System Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . .24
Windows® XP Professional Installation . . . . . . . . . . . . . . . . . . . . . . . . . .27
BIOS Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .27
RAID Configuration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .28
Operating System Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . .29
Driver Installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31
Video Card Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32
Benchmark Installation and Testing . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Windows Vista, 64-Bit Applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . .34
CINEBENCH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .34
Crafty. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .34
Panorama Factory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .34
POV-Ray . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .34

AMD Processor Performance Evaluation Guide 30579 Rev. 3.74 March 2007
5Contents
Windows Vista™, 32-Bit Applications . . . . . . . . . . . . . . . . . . . . . . . . . . .35
3DMark™06 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35
Cakewalk Sonar. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35
CINEBENCH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .36
Crafty. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .36
Dr. DivX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .36
POV-Ray . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .36
Vegas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .36
Windows® XP Professional, 32-Bit Applications . . . . . . . . . . . . . . . . . . . .37
3DMark06 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .37
Cakewalk Sonar. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .37
CINEBENCH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .38
Crafty. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .38
Dr. DivX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .38
High Performance Gaming and Multimedia Experience
(City of Villains and Windows Media Encoder 9) . . . . . . . . . . . . . . . . . .38
POV-Ray . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .39
Vegas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .39
Benchmarking Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Overall Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .40
Windows Vista, 64-Bit Applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . .41
Windows Vista, 32-Bit Applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . .44
Windows XP Professional, 32-Bit Applications . . . . . . . . . . . . . . . . . . . . .49

AMD Processor Performance Evaluation Guide 30579 Rev. 3.74 March 2007
6List of Figures
List of Figures
Figure 1. Dual-Core Processor Architecture . . . . . . . . . . . . . . . . . . . . . . . .14
Figure 2. HyperTransport™ Technology Block Diagram . . . . . . . . . . . . . . . .16
Figure 3. AMD Quad FX Platform with
Dual Socket Direct Connect Architecture . . . . . . . . . . . . . . . . . . .17
Figure 4. Overall Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .40
Figure 5. Overall Performance, Windows Vista™, 64-Bit Applications . . . . . .41
Figure 6. CINEBENCH Performance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .42
Figure 7. Crafty Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .42
Figure 8. Panorama Factory Performance . . . . . . . . . . . . . . . . . . . . . . . . .43
Figure 9. POV-Ray Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .43
Figure 10. Overall Performance, Windows Vista, 32-Bit Applications. . . . . . . .44
Figure 11. 3DMark™06 Performance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .45
Figure 12. Cakewalk Sonar Performance . . . . . . . . . . . . . . . . . . . . . . . . . . .45
Figure 13. CINEBENCH Performance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .46
Figure 14. Crafty Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .46
Figure 15. Dr. DivX Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .47
Figure 16. POV-Ray Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .47
Figure 17. Vegas Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .48
Figure 18. Overall Performance, Windows® XP Professional,
32-Bit Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .49
Figure 19. 3DMark06 Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .50
Figure 20. Cakewalk Sonar Performance . . . . . . . . . . . . . . . . . . . . . . . . . . .50
Figure 21. CINEBENCH Performance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .51
Figure 22. Crafty Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .51
Figure 23. Dr. DivX Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .52
Figure 24. High Performance Gaming and Multimedia Performance . . . . . . . .52
Figure 25. Panorama Factory Performance . . . . . . . . . . . . . . . . . . . . . . . . .53
Figure 26. POV-Ray Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .53
Figure 27. Vegas Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .54

AMD Processor Performance Evaluation Guide30579 Rev. 3.74 March 2007
List of Tables 7
List of Tables
Table 1. Memory Bandwidth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13
Table 2. Effect of Memory Bandwidth on Performance . . . . . . . . . . . . . . . .15
Table 3. Effect of Memory Latency on Performance . . . . . . . . . . . . . . . . . .15
Table 4. Benchmark Application Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . .20
Table 5. AMD Athlon™ 64 Single Socket Dual-Core Processor System . . . . .21
Table 6. AMD Quad FX Platform with
Dual Socket Direct Connect Architecture. . . . . . . . . . . . . . . . . . . .22
Table 7. Performance Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .55

AMD Processor Performance Evaluation Guide 30579 Rev. 3.74 March 2007
8Revision History
Revision History
Date Revision Description
March 2007 3.74 Incorporated legal and grammatic edits.
February 2007 3.73 Updated for the AMD Athlon™ 64 FX-62, AMD Athlon 64 FX-70,
AMD Athlon 64 FX-72, and AMD Athlon 64 FX-74 dual-core processors.
Replaced Office Productivity, Digital Media, and Gaming categories with
categories for Windows Vista™ with 64-bit applications, Windows Vista
with 32-bit applications, and Windows® XP Professional with 32-bit
applications.
Removed the following benchmarks:
• BAPCO® SYSmark® 2004 SE Office Productivity
• Business Winstone® 2004, Version 1.01*
• Business Winstone 2004 Multitasking, Version 1.01
• Worldbench
• Remote Collaboration Scenario
• Travel-Ready Scenario
• BAPCO SYSmark 2004 SE Internet Content Creation
• Content Creation Winstone® 2004, Version 1.01*
• iTunes, Version 6.0.4
• Protected High Definition Viewing Scenario
•3DMark
™05, Build 1.2.0 (Hardware and Software)
• City of Villains, Version 10.2
•SeriousSamII
• Half-Life 2, Version 1.0.1.0
• Quake 4, Version 1.04
• Unreal Tournament 2004, Version 3369
• Tom Clancy’s Splinter Cell Chaos Theory, Version 1.2b
• Far Cry, Version 1.3.1
•Doom3,Version1.1
Updated the following benchmarks:
• Vegas, Version 6.0 (to 7.0)
• Crafty, Version 19.19 (to four-thread version)
Added the following benchmarks:
•CINEBENCH9.5
• POV-Ray, version 3.7beta-14
• High Performance Gaming and Multimedia Scenario
Updated application information in Chapter 3.
Changed the recommended system configurations in Chapter 4.
Updated Chapter 5 to reflect changes in system configuration and tests.
Updated performance graphics in Chapter 6.
Made minor grammatic changes to improve readability.
*Business Winstone, Content Creation Winstone, and Winstone are
registered trademarks of Ziff Davis Publishing Holdings Inc. in the U.S.
and other countries.

AMD Processor Performance Evaluation Guide30579 Rev. 3.74 March 2007
Revision History 9
June 2006 3.72 Updated document format to comply with latest AMD standards.
Updated for the AMD Athlon™ 64 FX-62 processor, the AMD Athlon 64
processor 5000+, and the AMD Athlon 64 processor X2 3800+, relative
to the Intel Pentium® D 950 processor and the Intel Pentium EE 955
processor.
Added information about DDR2 memory latency to Chapter 2.
Removed the following benchmarks:
• WinRAR, version 3.42
• RazorLAME, version 1.1.5
•POV-Ray,version3.7.4
• Microsoft® Movie Maker, version 5.1
• 3DMark®03, build 3.6.0 (hardware and software)
• Pain Killer, version 1.64
• Quake III, version 1.11 (replaced by Quake 4)
• Return to Castle Wolfenstein Enemy Territory, version 2.60
• Star Wars- Jedi Knight II: Jedi Outcast, version 1.04
Updated the following benchmarks:
• SYSMark® 2004, version 1.02, patch 2
updated to SYSMark 2004 SE
• Unreal Tournament 2004, version 3355
updated to version 3369
Added the following benchmarks:
• iTunes, version 6.0.4
• Cakewalk Sonar, version 5
• Protected High Definition Viewing Scenario
• 3DMark06, build 1.0.2 (hardware and software)
•SeriousSamII
• City of Villains, version 10.2
• Quake 4, version 1.04 (replaces Quake 3)
Updated application information in Chapter 3.
Changed the recommended system configurations in Chapter 4.
Updated procedures in Chapter 5 to reflect changes in system
configuration and addition/removal of tests.
Removed individual data tables, increased graph size, and added data
summary table to Chapter 6.
Removed Appendix A Listings because listing source files are available
directly from AMD.
June 2005 3.71 Corrected title for Travel Ready Scenario.
June 2005 3.70 Updated for the AMD Athlon™ 64 FX-57 processor and the
AMD Athlon 64 processor 4800+ relative to the Intel Pentium®4 550
processor, which operates at 3.8 GHz.
Removed the Performance Analysis (64-bit) section, as these tests will
be shown in a separate document.
Removed obsolete processor information.
October 2004 3.61 Incorporated documentation edits.
Date Revision Description

AMD Processor Performance Evaluation Guide 30579 Rev. 3.74 March 2007
10 Revision History
October 2004 3.60 Updated for the AMD Athlon™ 64 FX-55 processor and the
AMD Athlon 64 processor 4000+ relative to the Intel Pentium®4 550
processor, which operates at 3.4 GHz.
Removed obsolete processor information.
Added the following tests to the standard benchmarking suite:
• Dr. DivX, version 1.0.6 (Replaces RawAVI to MPEG2 and Xmpeg)
• Return to Castle Wolfenstein Enemy Territory, version 2.60
(Replaces Return to Castle Wolfenstein)
• FarCry, version 1.3.1
•FarCrypier
Note:The two FarCry benchmarks were combined in revision 3.70.
• Painkiller, version 1.64
Within the Performance Analysis test suite, 64-bit versus 32-bit test
results have been combined with the 32-bit and 64-bit results, where
applicable.
Note:These changes are obsolete, as the Performance Analysis test
suite has been removed from this document.
June 2004 3.50 Updated to reflect the AMD Athlon™ 64 FX-53 (939) processor and the
3700+ and 3800+ processors relative to the Intel Pentium®4 Extreme
Edition 3.4 GHz and the Pentium 4 3.4 GHz processors.
Removed obsolete processor information.
Added the following tests to the Performance Analysis test suite:
• Table 64, “Panorama Factory Ver. 3.1 64-Bit Benchmark” on
page 69
• Table 65, “Crafty Ver. 19.12 64-Bit Benchmark” on page 69
• Table 64, “Panorama Factory Ver. 3.1 Benchmark Results” on
page 72
• Table 65, “Crafty Ver. 19.12 64-Bit Benchmark Results” on page 72
Note:These changes are obsolete, as the Performance Analysis test
suite has been removed from this document.
March 2004 3.43 Updated legal attribution for various benchmarks.
March 2004 3.41 As of revision 3.60, these tables are obsolete.
Updated the following tables:
• Table 35 on page 63
• Table 45 on page 66
• Table 48 on page 68
• Table 51 on page 69
• Table 77 on page 78
March 2004 3.40 Replaced the obsolete AMD Athlon™ 64 FX-51 processor information
with the AMD Athlon 64 FX-53 processor. This change affects Table 3 on
page 24 and each benchmark result.
Replaced the older Intel Pentium® 4 3.2 GHz configuration and
performance data with the Intel Pentium 4 3.2 GHz Extreme Edition
Processor. This change affects Table 4 on page 22 and each benchmark
result.
Replaced the benchmark result tables with graphs and corresponding
tables.
Date Revision Description

AMD Processor Performance Evaluation Guide30579 Rev. 3.74 March 2007
Revision History 11
January 2004 3.32 Corrected instructions for “Ziff Davis Media Inc. Business Winstone®
2004” on page 32, “Ziff Davis Media, Inc. Business Multitasking
Winstone® 2004” on page 33 and “Ziff Davis Media Inc.’s Content
Creation Winstone® 2004” on page 39.*
Moved 64-Bit performance results from non-optimized rows to
optimized rows in Table 7 on page 49 and Table 8 on page 51.
*Business Winstone, Content Creation Winstone, and Winstone are
registered trademarks of Ziff Davis Publishing Holdings Inc. in the U.S.
and other countries.
Note:These changes are obsolete, as the Performance Analysis test
suite has been removed from this document.
January 2004 3.31 Updated performance results for Table 7 on page 49 and Table 8 on
page 51.
Corrected minor typos throughout.
December 2003 3.30 Updated to reflect 3400+ launch.
Figure 3 was removed.
Additional instructions were added for the DivX Encoder for 64-Bit
installation and run. Now refer to “Mini-GZIP” on page 61.
December 2003 3.25 Removed Revision bars.
December 2003 3.24 On Page 15, removed references to WinACE, because it is no longer
tested.
Within “Operating System Configuration” on page 27, added
instructions to skip steps 12 and 13 because they do not apply if
Microsoft® Windows® is not yet installed. Instead, skip to step 14.
On page 30 added notations that ASUS and MSI drivers are applicable
only to their respective motherboards.
December 2003 3.23 Updated benefits for 64-bit processing in “64-bit processing” on
page 16.
Note:This change is now obsolete, as the Performance Analysis test
suite has been removed from this document.
Corrected attribution in “WinZip Computing WinZip 8.1” on page 20.
November 2003 3.22 Added figure label to Figure 3 on page 46. Corrected two column format
balancing in various locations.
November 2003 3.21 Applied new document template.
October 2003 3.2 Revision to Table 3 on page 24 to correct memory manufacturer.
Revision to update configuration steps for To install the video clip to use
for DivX Encoder on page 45.
September 2003 3.1 Revision to include NVIDIA® video driver and ASUS chipset installation.
September 2003 3.0 Initial Public Release
Date Revision Description

About This Document 12
AMD Processor Performance Evaluation Guide30579 Rev. 3.74 March 2007
About This Document
This guide describes AMD processor performance test methodology and presents
performance test results based on that methodology.
Audience
The guide provides information for those interested in evaluating the performance
of AMD64 technology, with particular emphasis on members of the hardware review
community.
Life of Document
This document provides information about the performance of these processors:
• AMD Athlon™ 64 FX-62 dual-core processor
• AMD Athlon 64 FX-70 dual-core processor
• AMD Athlon 64 FX-72 dual-core processor
• AMD Athlon 64 FX-74 dual-core processor
This document may become obsolete or may be revised as new speed grades
become available.

AMD64 Single Socket Processor Architecture 13
AMD Processor Performance Evaluation Guide30579 Rev. 3.74 March 2007
AMD64 Single Socket Processor Architecture
Detailed knowledge of AMD 64-bit processor architecture is not required to perform
optimal benchmarking. However, the benchmarks demonstrate the advantages of
key architectural features. This overview provides information about those features
and shows how the benchmarks demonstrate the exceptional performance of AMD
processors.
AMD 64-bit processors include the following architectural improvements specifically
designed to increase the number of instructions per clock (IPC).
• AMD64 Technology
When the AMD64 Instruction Set Architecture is utilized, 64-bit mode offers:
– Support for 64-bit operating systems that provide full, transparent, and
simultaneous 32-bit and 64-bit platform application multitasking.
– A physical address space that supports up to 1 TB of installed RAM, shattering the
4 GB RAM barrier on current x86 systems.
– Sixteen 64-bit general-purpose integer registers, four times as much general-
purpose register space for applications and device drivers as traditional x86
architectures.
– Sixteen 128-bit XMM registers for enhanced multimedia performance, double the
register space of current SSE/SSE2/SSE3 implementations.
• An integrated DDR2 memory controller (see Figure 1).
– The integrated controller reduces memory latency and increases overall system
performance.
– Whencomparing platformswith different typesof memory, testmemory bandwidth
and latency first. The results help to clarify the sometimes surprising results of
more complex, application-based benchmarks.
Memory is marked/marketed on bandwidth. As Table 1 shows, during tests,
DDR2-800 memory can provide up to 35% more bandwidth than DDR1-400.
Table 1. Memory Bandwidth
AMD Athlon™ 64 FX-62
Dual-Core Processor,
DDR2-800
AMD Athlon 64 FX-60
Processor, DDR1-400
Sciencemark 2.0
Bandwidth (MB/s) 6779.3 4472.6
Sciencemark 2.0
Latency (ns) 46.5 52.5

AMD Processor Performance Evaluation Guide 30579 Rev. 3.74 March 2007
14 AMD64 Single Socket Processor Architecture
Figure 1. Dual-Core Processor Architecture
&RRO¶Q¶4XLHW WHFKQRORJ\
IRUTXLHWHURSHUDWLRQDQG
ORZHUSRZHUFRVWV
 5HFRJQL]HG E\WKH86
(QYLURQPHQWDO3URWHFWLRQ
$JHQF\ IRUDGYDQFHPHQW
RIHQHUJ\HIILFLHQW
FRPSXWHUWHFKQRORJ\
+\SHU7UDQVSRUWWHFKQRORJ\
DFFHOHUDWHVWKHV\VWHP
EXVIRUKLJKVSHHG
,2FRPPXQLFDWLRQ
 *%VRIDYDLODEOH
ELGLUHFWLRQDOV\VWHP
 *+]%LGLUHFWLRQDO
$0' $WKORQ 
'XDO&RUH3URFHVVRUV
DUHWUXHGXDOFRUH
 ,QGHSHQGHQW/FDFKH
 ,QWHOOLJHQW6\VWHP5HTXHVW
4XHXHDQGLQWHJUDWHG
&URVVEDU6ZLWFK
7KHLQGXVWU\¶VILUVWWUXHRQGLH
GXDOFRUH[SURFHVVRU
 ,QWHUFRUHFRPPXQLFDWLRQ
DW&38VSHHG
LUHFW ' DFFHVVWRPHPRU\
FRQWUROOHUDQG
+\SHU7UDQVSRUW
WHFKQRORJ\OLQNV
(QKDQFHG9LUXV3URWHFWLRQ
IRU:LQGRZV ;363
 'HVLJQHG WRKHOSSUHYHQW
WKHVSUHDGRIFHUWDLQ
YLUXVHVOLNH06%ODVWHU DQG
6ODPPHU
 'HVLJQHGWRUHGXFHWKH
FRVWDQGGRZQWLPH
DVVRFLDWHGZLWK VLPLODU
YLUXVHV DQGLPSURYHVWKH
SURWHFWLRQRIFRPSXWHUV
DQGSHUVRQDOLQIRUPDWLRQ
DJDLQVWFHUWDLQ 3&YLUXVHV
$0'$UFKLWHFWXUHGHOLYHUV
OHDGLQJHGJHVRIWZDUH
SHUIRUPDQFH
 %RWKKLJKSHUIRUPDQFHDQG
 DQGELW FRPSXWLQJ
 ([SHULHQFHGXDOFRUH
SHUIRUPDQFHWRGD\ZLWK QR
KDUGZDUHFKDQJHV
$0'$WKORQ6LQJOH6RFNHW
'XDO&RUH3URFHVVRU$UFKLWHFWXUH
5HSODFHV$GGUHVV'DWDDQG&RQWURO%XV
''50HPRU\
&RQWUROOHU
+\SHU7UDQVSRUW
7HFKQRORJ\
&URVVEDU6ZLWFK
EDQGZLGWK
/
'DWD
&DFKH

 6\VWHPVFDQLQFOXGH
PXOWLSOHOLQNV
$0'
3URFHVVRU
&RUH

$0'
3URFHVVRU
&RUH

6\VWHP5HTXHVW4XHXH






/
&DFKH

/
&DFKH

/
'DWD
&DFKH

/
,QVWUXFWLRQ
&DFKH

/
,QVWUXFWLRQ
&DFKH


AMD Processor Performance Evaluation Guide30579 Rev. 3.74 March 2007
AMD64 Single Socket Processor Architecture 15
For discrete graphics configurations, memory bandwidth has a smaller impact on
end performance. For example, changing from dual-channel to single-channel
memory (halving the marketed bandwidth) on an otherwise identical system
reduces performance only 2-3% (see Table 2).
Memory latency is more critical to performance. As Table 3 shows, an
AMD Athlon™ 64 FX-62 dual-core processor with DDR2-800 memory has
approximately 4% better performance than an AMD Athlon 64 FX-60 dual-core
processor with DDR1-400 memory. In general, each 5-10% reduction in latency
improves performance by 1-2%.
• An advanced HyperTransport™ technology link (see Figure 2)
– Dramatically improves I/O bandwidth, enabling faster access to peripherals such
as hard disk drives, USB 2.0, and gigabit Ethernet cards.
– Data compression benchmarks illustrate higher processor performance due to a
reduced I/O interface throttle.
• Large first-level (L1) and second-level (L2) on-die caches
– 128 KB of L1 cache and 1 MB of L2 cache allow the AMD Athlon 64 processor to
excel at performing matrix calculations on arrays.
– Benchmarks that use intensive, large matrix calculations benefit from having the
entire matrix available in the L2 cache.
• Processor core clock-for-clock improvements
– Larger translation look-aside buffers (TLB) with reduced latencies.
– A global history counter with fourtimes the number of bimodal counters as seventh-
generation processors, to improve branch prediction.
– These features drive IPC improvements and provide a more efficient pipeline for
CPU-intensive applications.
– CPU-intensive games such as Unreal Tournament benefit from these core
improvements.
Table 2. Effect of Memory Bandwidth on Performance
AMD Athlon™ 64
Processor 3800+,
Dual-Channel Memory
AMD Athlon 64
Processor 3800+,
Single-Channel Memory
Office Productivity 101.3% 100.0%
Digital Media 101.9% 100.0%
Computer Gaming 104.1% 100.0%
Overall 102.4% 100.0%
Table 3. Effect of Memory Latency on Performance
AMD Athlon™ 64 FX-62
Dual-Core Processor,
DDR2-800
AMD Athlon 64 FX-60
Processor,
DDR1-400
Office Productivity 103.5% 100.0%
Digital Media 105.6% 100.0%
Computer Gaming 104.5% 100.0%
Overall 104.6% 100.0%

AMD Processor Performance Evaluation Guide 30579 Rev. 3.74 March 2007
16 AMD64 Single Socket Processor Architecture
Figure 2. HyperTransport™ Technology Block Diagram
• The SSE3 instruction set and 3DNow!™ Professional (SSE and 3DNow!
Enhanced) support all industry-standard x86 32-bit instruction set
extensions.
• 64-bit processing
– A 64-bit address and data set enables processing in the terabyte space.
– Microsoft Windows®XP Professional 64-Bit Edition for 64-Bit Extended Systems
supports up to 32 GB of RAM and up to 16 TB of virtual memory.
– Gamers can preload entire three-dimensional worlds into memory for a fully
immersive experience.
– Home video enthusiasts can easily edit video recordings, with professional-quality
results.
– The 64-bit space is designed to bring home the digital experience.
• The first true on-die dual core x86 PC processor
– Inter-core communication at CPU speed
– Direct access to memory controller and HyperTransport™technology link
HyperTransport™
Tunnel
IEEE 1394
Graphics
Adapter
EIDE
Ethernet
SMBus
AC-97
USB
LPC
DDR2 Memory
• High-bandwidth serial interconnect
• Extensive data protection/management
• Integrated power management
• Building block structure
• Up to N devices supported in chain
HyperTransport™ I/O links
Slot Slot Slot
AMD64
Processor
HyperTransport
I/O Hub
AGP/PCIe™
PCI 3.0/PCI-X®2.0 Bus
Device 0
Device 1
Device N
12.8 GB/Sec

AMD Processor Performance Evaluation Guide30579 Rev. 3.74 March 2007
AMD Quad FX Platform with Dual Socket Direct Connect Architecture 17
AMD Quad FX Platform with
Dual Socket Direct Connect Architecture
The AMD Quad FX Platform with Dual Socket Direct Connect Architecture (see
Figure 3) is a two-socket, four-core processing solution that features high-
bandwidth processor-to-processor communication between matched pairs of
AMD Athlon™ 64 FX-70, FX-72, and FX-74 dual-core processors.
Figure 3. AMD Quad FX Platform with
Dual Socket Direct Connect Architecture
The platform enhances the personal computing experience by exploiting the full
potential of the latest multi-threaded applications running under Windows Vista™.
Currently-shipping motherboards include the following features.
• Twelve SATA RAID interfaces
•TwoGigabitEthernet (GbE) interfaces
• Ten USB 2.0 interfaces
• Two IEEE 1394 interfaces
• UAA HD Audio, External CODEC
• SP/PP/PS2 connections
• Optical/Coax SPDIF connections
128-Bit
Dual-Channel
Unbuffered DDR2 DIMM Unbuffered DDR2 DIMM
Unbuffered DDR2 DIMM Unbuffered DDR2 DIMM
Coher
Link (
NVIDIA®
nForce 680a SLI
MCP
SATA
(12) GbE (2)
USB2
(10)
IEEE
1394
(2)
LPC
×8 PCIe (×16 Slot)
×8 PCIe (×16 Slot)
×1 PCIe Slot
×16 PCIe™ Slot
×16 PCIe Slot
PCI Slot
Super
I/O BIOS
Flash
128-Bit
Dual-Channel

Benchmarking Methodology 18
AMD Processor Performance Evaluation Guide30579 Rev. 3.74 March 2007
Benchmarking Methodology
This chapter describes the benchmarks used to generate the performance scores
shown in this guide. The benchmarks are chosen with the following points in mind.
• AMD Athlon™ 64 processors can run multiple tasks very efficiently.
• Systems based on AMD Athlon 64 processors can take advantage of unique
architectural features to deliver outstanding performance for media creation
and playback.
• The high performance of AMD Athlon 64 processors can significantly enhance
the three-dimensional display capabilities of a gaming system1.
AMD recommends these benchmarks for proper, balanced, real-world performance
analysis.
Benchmark Description
Two versions of the Microsoft Windows® operating system are used to run three
sets of benchmark applications.
• 64-bit Windows Vista™ running 64-bit applications
• 64-bit Windows Vista running 32-bit applications
• 32-bit Windows® XP Professional running 32-bit applications
For some applications, both a 64-bit version and a 32-bit version are run. The
descriptions that follow indicate which versions are used. Table 4 summarizes the
sets of applications.
• MAXON Computer GmbH, CINEBENCH 9.5
– Multiprocessor video performance benchmarking2
– Tested as a 64-bit application running under Windows Vista, a 32-bit application
running under Windows Vista, and a 32-bit application running under Windows XP
Professional
• Robert M. Hyatt, Crafty, Version 19.19 (Four-Thread Version)
– High-level, computation-intensive chess
– Tested as a 64-bit application running under Windows Vista, a 32-bit application
running under Windows Vista, and a 32-bit application running under Windows XP
Professional
• Smoky City Design, LLC, Panorama Factory, Version 4.4
– Photo stitching
– Tested as a 64-bit application running under Windows Vista and a 32-bit application
running under Windows XP Professional

AMD Processor Performance Evaluation Guide30579 Rev. 3.74 March 2007
Benchmarking Methodology 19
• Persistence of Vision Raytracer Pty. Ltd., POV-Ray 3.7beta-14
– 3D graphic creation
– Tested as a 64-bit application running under Windows Vista, a 32-bit application
running under Windows Vista, and a 32-bit application running under 32-bit
Windows XP Professional
• Futuremark Inc., 3DMark™06, Build 1.0.2
– 3D game performance benchmarking
– Only the CPU tests are run2
– Tested as a 32-bit application running under Windows Vista™and a 32-bit
application running under Windows®XP Professional
• Twelve Tone Systems, Inc., Cakewalk Sonar, Version 5
– Music composition
– Tested as a 32-bit application running under Windows Vista and a 32-bit application
running under Windows XP Professional
• DivX, Inc., Dr. DivX, Version 2.0 (DivX Codec 6.4)
– Video encoding
– Tested as a 32-bit application running under Windows Vista and a 32-bit application
running under Windows XP Professional
• Sony Corporation of America, Vegas, Version 7.0
– Video file conversion
– Tested as a 32-bit application running under Windows Vista and a 32-bit application
running under Windows XP Professional
• High Performance Gaming and Multimedia Experience operating scenario
(NCsoft Corporation, City of Villains, Version 10.2 and
Microsoft, Inc., Windows Media Encoder, Version 9.00.00.2980)
– High-performance online gaming1and video encoding
– City of Villains is launched and on-line game play is initiated
– Windows Media Encoder converts an MPEG2 file to MPEG4 format
– Tested as a 32-bit application running under Windows XP Professional
1. Computer gaming benchmarks may reveal the limitations of a graphics card and may not truly represent relative
processor performance.
2. The term “Central Processing Unit (CPU)” often refers to devices with a single processing unit. AMD uses the term
“processor” to designate a computing device that contains one or more processing units, or “cores”. The term
“multiprocessor” refers to a computer system that contains more than one core, either in a single device or in multiple
devices. The CINEBENCH benchmark and the 3DMark06 CPU benchmark can test single or multiprocessor systems.

AMD Processor Performance Evaluation Guide 30579 Rev. 3.74 March 2007
20 Benchmarking Methodology
Table 4. Benchmark Application Sets
Windows Vista™, 64-bit Applications
CINEBENCH 9.5
Crafty 19.19 (four-thread version)
Panorama Factory 4.4
POV-Ray 3.7beta-14
Windows Vista™, 32-bit Applications
3DMark™06 1.02 (CPU)
Cakewalk Sonar 5
CINEBENCH 9.5
Crafty 19.19 (four-thread version)
Dr.DivX 2.0 (DivX Codec 6.4)
POV-Ray 3.7beta-14
Vegas 7.0
Windows® XP Professional, 32-bit Applications
3DMark06 1.02 (CPU)
Cakewalk Sonar 5
CINEBENCH 9.5
Crafty 19.19 (4-thread version)
Dr.DivX 2.0) (DivX Codec 6.4)
High Performance Gaming and Multimedia Experience
Panorama Factory 4.4
POV-Ray 3.7beta-14
Vegas 7.0

Benchmarking System Configuration 21
AMD Processor Performance Evaluation Guide30579 Rev. 3.74 March 2007
Benchmarking System Configuration
This chapter describes the system configurations used for benchmarking and
provides setup procedures for the AMD Athlon™ 64 processors in this document.
Test System Configurations
Systems that are configured as described in the following tables are most likely to
demonstrate optimal system performance.
Table 5. AMD Athlon™ 64 Single Socket Dual-Core Processor System
Component Manufacturer Model Description
Processor AMD AMD Athlon™ 64 FX-62
Dual-Core Processor 2.8 GHz clock frequency
Operating
System Microsoft® Windows® XP Professional
Windows Vista™ Version 2002 (Service Pack 2)
RC2
Motherboard ASUS M2N32-SLI
Deluxe NVIDIA® nForce 590 SLI MCP
SATA Driver 5.10.2600.652
BIOS ASUS Crashfree BIOS 3 Version 3092
Hard Drive
(2) Western Digital Raptor WD1500ADFD SATA RAID 0 10k RPM 150 GB
RAM
(4) Corsair CM2X1024-8500SD 1 GB PC8500 DDR2 DIMM 800 MHz
Video Card
(2) NVIDIA 7950 GX2 (Quad SLI mode)
1600x1200 1 GB GDDR3 Onboard RAM, Video Driver 6.14.10.9371