Skip to main content

Move Up to IBM Power Systems

Power up - Power6 has 2 to 3 times less performance per core compared to the competition, and is designed to reduce downtime.

IBM POWER6™ processor-based servers are designed to allow you to get the most out of your IT investment. The combination of leadership performance, energy efficiency, flexible virtualization features and RAS features designed to maximize application availability can enable you to take back control of your IT infrastructure.

Tab navigation

Tab navigation

Power performance

Compare highend servers

IT managers have to grapple with the issue of server sprawl and the accompanying high cost of management combined with the explosion in energy use in the datacenter. One solution that many clients are implementing is server consolidation. The IBM Power™ 595 server, exploiting IBM PowerVM™ technology, is the epitome of a consolidation server because of its outstanding performance and capacity.

With the fastest processor in the industry, the 5.0GHz POWER6 processor, and outstanding memory and I/O bandwidth, the Power 595 has the capacity to support huge consolidation efforts that other servers just can't handle.

We ran a variety of benchmarks, and the answer is the same for all of them. The IBM Power™ 595 server outperforms systems from HP & Sun with 2 times the cores and outperforms systems with the same number of cores by more than 2 to 1.

The Power 595 server can offer you this fabulous performance for your UNIX® applications.

Highend Systems Performance1
  SPECint_rate2006 SPECfp_rate2006 SPECjbb2005 2-tier SAP SD
64-core IBM Power 595 2,080 2,110 3,272,999 35,400
128-core HP Integrity Superdome 1,650 1,480 2,054,864 30,000
64-core HP Integrity Superdome 824 745 No Benchmark 9,265
128-core Sun SPARC Enterprise M9000 1,290 1,230 No Benchmark No Benchmark
64-core Sun SPARC Enterprise M9000 650 636 No Benchmark No Benchmark
144-core Sun Fire E25K 1,230 No Benchmark 2,105,264 10,175

The 64-core Power 595 server outperformed the 64-core HP Integrity Superdome server by at least 2 ½ to 1 and the 64-core Sun SPARC Enterprise M9000 server by over 3 to 1 on processor intensive benchmarks. Even when the Superdome and M9000 servers were tested with twice the number of cores, the Power 595 server was faster by 18% up to as much as 71%. 71% faster than a system with double the cores is a very impressive result, but that is what you can expect form the IBM Power 595 server, very impressive results.

These benchmarks show the outstanding systems performance of the Power 595, but sometimes overall systems performance is not the most important factor, especially in a consolidated environment. Some applications are more bound by memory constraints, either based on the quantity of memory available or the memory bandwidth. Others are constrained by I/O bandwidth. When we compare those characteristics, the Power 595 is equally impressive.

  Power 595 Superdome M9000
Cores 64 128 128
Memory (GB) 4096 2048 2048
Memory BandWidth (GB/s) 1376 273.1 737
I/O BandWidth (GB/s) 640 172.8 234
       
Memory (GB) per core 64 16 16
Memory BandWidth (GB/s) per core 21.50 2.13 5.76
I/O BandWidth (GB/s) per core 10.00 1.35 1.83

Besides delivering two times the processor performance per core, the Power 595 also supports 4 times the memory per core, over 3 ½ times the memory bandwidth per core and over 5 times the I/O bandwidth per core of the Integrity Superdome and the M9000 highend systems.2 The tremendous capacity of the IBM Power 595 server can help you run challenging applications in every partition.

The Power 595 server is the fastest system with the fastest processor, the most memory capability, and I/O bandwidth. It is no wonder that it is a prime target for highly consolidated environments.

Compare midrange servers

The workhorse UNIX® database and application servers have to deliver cost effective, reliable performance for mission-critical applications. Using advanced features like virtualization, IBM POWER6™ processor performance, and mainframe-inspired reliability, availability, and serviceability (RAS) characteristics, the IBM Power™ 570 server epitomizes what today's mid-range server should be.

The POWER6 processor technology and advanced system design of the Power 570 server create a system with leadership-balanced performance. The very fast processor, up to 4.7 GHz, with system bus, I/O bandwidth, and large memory capacity that grows with the number of processors leads to the astounding number of leadership commercial and technical benchmarks.

The result is the very best performance for a mid-range server, and the very best performance per core of a mid-range server.

We compare performance because the exceptional speed of the Power 570 system is what allows you to meet your end user's service level objectives. And this is designed to be done with fewer cores which helps drive down your cost and improve system manageability.

The following table summarizes the best mid-range server performance result for the IBM Power 570 server versus Sun and HP 32-core UNIX servers. The results show that the Power 570 server has the very best performance of any midrange server.

Midrange Systems Performance3
  TPC-C4 SPECint_rate20065 SPECfp_rate20066 SPECjbb20057 2-tier SAP SD8
IBM Power 570 1,616,162 478 430 691,975 8,000
HP Integrity rx8640   416 371    
Sun SPARC
Enterprise M8000
  352 333 408,621 7,300

The Power 570 server leads competitive midrange systems in capacity by far. However that is only one measure of performance. Better performance per core helps drive cost savings in hardware acquisition, software licensing, and a variety of operational areas. The following table shows the Power 570 server as compared to Sun and HP 32-core UNIX® servers.

Midrange Systems Performance per Core9
  TPC-C SPECint_rate 2006 SPECfp_rate 2006 SPECjbb 2005 2-tier SAP SD
IBM Power 570 101,010 30 27 43,248 500
HP Integrity rx8640   13 12    
Sun SPARC Enterprise M8000   11 10 12,769 228

With performance per core more than double the midrange competitors, the Power 570 server can enable you to reduce the number of servers and/or the size of servers you require to do your production. The PowerVM™ feature of the Power 570 server can also help you increase the utilization of your servers which could enable you to even further reduce the number. This combination could have a major impact on lowering your cost of server acquisition and your total cost of ownership for your IT environment.


Compare Entry Servers

As entry servers become faster, more customers find them the right performance for mission-critical applications. They also are often the right servers for distributed applications in stores and branch banks. The same POWER6 technology that makes the Power 570 the fastest midrange server helps make the IBM Power 550 Express system the fastest UNIX entry server. The comparison to the 8-core HP Integrity rx6600 and the 16-core Sun SPARC Enterprise M5000 servers help demonstrate this. In fact, the Power 550 Express server has such excellent performance that its result for the demanding TPC-C benchmark is even better than the best HP 9000 Superdome result ever posted.10

Entry Systems Performance11
  TPC-C12 SPECint_rate 200613 SPECfp_rate 200614 SPECjbb 200515 2-tier SAP SD16
IBM Power 550 Express 629,159 213 178 333,779 3,104
HP Integrity rx6600 372,140 102 71 158,174 2,150
Sun SPARC Enterprise M5000   158   192,055  

The advantage of the Power 550 Express server is even more evident when you look at performance per core.

Entry Systems Performance per core17
  TPC-C SPECint_rate 2006 SPECfp_rate 2006 SPECjbb 2005 2-tier SAP SD
IBM Power 550 Express 78,645 27 22 41,722 388
HP Integrity rx6600 46,518 13 9 19,772 269
Sun SPARC Enterprise M5000   10   12,003  

Compare Blade Servers

Blade servers provide a form factor designed to optimize management and operations for scale out applications. The servers are plugged into a chassis that provides power, cooling, management and many optional features shared by all the blade servers. This form factor can enable you to increase compute density, reduce overall power consumption, and eliminate expensive and hard to manage cabling. The shared features and the management capabilities can combine with those savings to reduce the total cost of ownership for scale out applications.

The IBM BladeCenter® JS22 Express combines the attributes of the industry-leading IBM BladeCenter chassis with the performance of POWER6 processor technology. This performance can enhance the overall cost advantage by reducing the number of blades required. The following table compares the performance of 4-core blade servers. The JS22 Express has an advantage of at least 30% in SPECint_rate2006, and 40% for both SPECfp_rate2006 and SPEC OMPM2001. Since there was no 4-core SPARC processor-based server results reported, the results for 8-core servers were reported in this table and the JS22 Express result was still 7% better in SPECint_rate2006 than the SPARC processor-based server with twice the number of cores and more than 20% better on SPECfp_rate2006 and SPEC OMPM2001 than the SPARC processor-based servers with twice the number of cores.


4-core Blade Server Performance18 SPECint_rate2006 SPECfp_rate2006 SPEC OMPM2001
POWER6 (IBM BladeCenter JS22 Express) 84.7 75 19688
Xeon 65 45.5 8497
Opteron 58.9 52.5 13868
Itanium 53.9 48.8 6886
SPARC 78.6 62.3 16208

The next table documents the best performance per core for each technology for 2- or 4-core blade servers. Some very popular software is licensed based on the number of cores. Having the best performance per core can lead to lower software costs enhancing total cost of ownership. In this case, the comparison included both 2- and 4-core blade servers since both are popular for scale out applications. Once again the JS22 Express has a 30% advantage in peak throughput per core using SPECint_rate2006 results and 40% advantage in the other benchmarks.

2 or 4-core Blade Server Performance/Core19 SPECint_rate20067 SPECfp_rate20068 SPEC OMPM20019
POWER6 (IBM BladeCenter JS22 Express) 21.2 18.8 4922
Xeon 16.3 11.4 2124
Opteron 14.8 13.3 3467
Itanium 13.5 12.2 1722
SPARC 9.8 7.8 2026

One of the main reasons to use blade servers is to increase compute density. Server sprawl is a widely recognized issue for scale out applications. One way to contain sprawl is to increase the density. Therefore the next table looks at performance per 42U rack for the technologies. Each blade technology has multiple chassis options. For this measurement, the chassis which provided the greatest density was chosen for the comparison. The JS22 Express provided better than 20% more compute density based on SPECint_rate2006, 25% more compute density based on SPECfp_rate2006, and more than 60% greater compute density based on SPEC OMPM2001.

2 or 4-core Blade Server Performance/Rack20 SPECint_rate2006 SPECfp_rate2006 SPEC OMPM2001
POWER6 (IBM BladeCenter JS22 Express) 4743 4200 1102528
Xeon 3910 3335 NA 
Opteron 3770 3360 665664
Itanium 1725 1562 27544
SPARC 3773 2990 680736

So if you need absolute performance, performance per core to minimize software licensing, or performance per rack to help control server sprawl, then the IBM BladeCenter JS22 Express is a computing solution you should be considering.



1Competitive benchmark results reflect results published as of March 24, 2008. The results are the best IBM Power 595 64-core server result, the best 128-core and 64-coreSun SPARC Enterprise M9000 server results, the best 144-core Sun SPARC E25K server result, and the best 128-core and 64-core HP Integrity Superdome results. Cells were left blank if there is no posted result for the benchmark. The SPECint_rate2006, SPECfp_rate2006 and SPECjbb2005 results can be found at www.spec.org. The SAP SD 2-tier results can be found at www.sap.com.

SPEC® and the benchmark names SPECrate®, SPECint®, and SPECjbb® are registered trademarks of the Standard Performance Evaluation Corporation. For the latest SPEC benchmark results, visit http://www.spec.org

Highend SPECint_rate2006 results:
System Name Cores Chips Cores / Chip Threads / Core Base Peak
IBM Power 595 64 32 2 2   2080
HP Integrity Superdome (1.6GHz/24MB Dual-Core Intel Itanium 2) 128 64 2 1 1530 1650
HP Integrity Superdome (1.6GHz/24MB Dual-Core Intel Itanium 2) 64 32 2 1 770 824
Sun SPARC Enterprise M9000 128 64 2 2 1110 1290
Sun SPARC Enterprise M9000 64 32 2 2 553 650

Highend SPECfp_rate2006 results:
Test Sponsor System Name Cores Chips Cores / Chip Threads / Core Peak
IBM Corporation IBM Power 595 64 32 2 2 2110
Hewlett-Packard Company HP Integrity Superdome (1.6GHz/24MB Dual-Core Intel Itanium 2) 128 64 2 1 1480
Hewlett-Packard Company HP Integrity Superdome (1.6GHz/24MB Dual-Core Intel Itanium 2) 64 32 2 1 745
Sun Microsystems Sun SPARC Enterprise M9000 128 64 2 2 1230
Sun Microsystems Sun SPARC Enterprise M9000 64 32 2 2 636

Highend SPECjbb2005 results:
HW Company Name HW System Name # JVM Instances Cores Chips HW Threading bops bops / JVM
IBM Corporation IBM Power 595 32 64 32 Yes 3,272,999 102281
Hewlett-Packard Company HP Integrity SD64B Server 32 128 64 No 2054864 64215
Sun Microsystems, Inc. Sun Fire E25K 72 144 72 N/A 2105264 29240

Highend systems 2-tier SAP SD benchmark results
Central Server Operating System - Release RDBMS Release ERP Release Number of Benchmark Users Chips Cores Threads Certification Number
IBM Power 595       35,400 32 64 128 TBD
HP Integrity Superdome-32 Windows Server 2003 Datacenter Edition SQL Server 2005 2005 9,265 32 64 128 2007020
HP Integrity SD64B HP-UX 11iV3 Oracle 10g 2005 30,000 64 128 256 2006089
Sun Fire Model E25k Solaris 9 Oracle 9i 4.7 10,175 72 0 0 2004039

2 Comparisons between the IBM Power™ 595 server, the HP Integrity Superdome Server and the Sun SPARC Enterprise M9000 server. System data for HP from the HP Superdome Datasheet available at www.hp.com. System data for Sun from the Sun SPARC Enterprise M9000 Datasheet available at www.sun.com. Data is current as of March 26, 2008.

3Competitive benchmark results reflect results published as of March 24, 2008. The results are the best IBM Power 570 16-core server result, the best Sun SPARC Enterprise M8000 server 32-core result, and the best HP Integrity rx8640 server 32-core result. Cells were left blank if there is no posted result for the benchmark. The TPC-C results can be found at www.tpc.org. The SPECint_rate2006, SPECfp_rate2006 and SPECjbb2005 results can be found at www.spec.org. The SAP SD 2-tier results can be found at www.sap.com.

SPEC® and the benchmark names SPECrate®, SPECint®, and SPECjbb® are registered trademarks of the Standard Performance Evaluation Corporation. For the latest SPEC benchmark results, visit http://www.spec.org

4 The highest TPC-C results are in the following table.


* Hardware Vendor System tpmC Price/tpmC System Availability Date Submitted
IBM IBM Power 570 1,616,162 3.54 US $ 11/21/07 05/21/07

The IBM Power 570 has 8-processor chips, 16-cores, and 2-threads per core.

5 The SPECint_rate2006 results are:

Test Sponsor System Name Base Copies Cores Chips Cores / Chip Threads/Core Base Peak
Hewlett-Packard Company HP Integrity rx8640 (1.6 GHz/24 MB Dual-Core Intel Itanium 2) 32 32 16 2 1 385 416
IBM Corporation IBM System p 570 (4.7 GHz, 16 core) 32 16 8 2 2 410 478
Sun Microsystems Sun SPARC Enterprise M8000 63 32 16 2 2 298 352

6SPECfpRate_2006 results are:

Test Sponsor System Name Base Copies Cores Chips Cores / Chip Threads/Core Base Peak Peak / Core
Hewlett-Packard Company HP Integrity rx8640 (1.6 GHz/24 MB Dual-Core Intel Itanium 2) 32 32 16 2 1 385 416 13
Sun Microsystems Sun SPARC Enterprise M8000 63 32 16 2 2 298 352 11
IBM Corporation IBM System p 570 (4.7 GHz, 16-core) 32 16 8 2 2 410 478 29.88

7SPECjbb2005 results for the best bops for 4- to 32-core servers of each technology are:

HW Company Name HW System Name # JVM Instances Cores Chips HW Threading bops bops/JVM
IBM Corporation IBM System p 570 (4.7 GHz, 16 core) 8 16 8 Yes 691975 86497
Sun Microsystems, Inc. Sun SPARC Enterprise M8000 16 32 16 No 440207 27513

8SAP SD 2-tier results for the best Benchmark Users for servers with 4- to 32-cores are:

Date of Certification (mm/dd/yyyy) Technology Partner Number of Benchmark Users Central Server (*) Additional Specifications Chips Cores Threads Certification Number
5/21/2007 IBM 8000 IBM System p 570 8 16 32 2007039
4/17/2007 Sun Microsystems 7300 Sun SPARC Enterprise Server Model M8000 16 32 64 2007026

9The midrange systems performance per core is calculated by dividing the values in the midrange systems performance table by the number of cores in each benchmarked system. All the IBM Power 570 server benchmarks in the table used 16-cores. All the results for the HP Integrity rx8640 and the Sun SPARC Enterprise M8000 used 32-cores.

10The comparison of the IBM Power 550 Express server with the best performing HP 9000 Superdome is shown in the table below. The Power 550 had 4-processor chips, 8-cores, and 2-threads per core. The HP Superdome had 64-processor chips, 64-cores, and 1 thread per core.

* Hardware Vendor System tpmC Price/tpmC System Availability Date Submitted
HP HP Superdome – PA-RISC/875 MHz-64p/64c 541,673 11.66 US $ 01/30/04 07/30/03
IBM IBM System p 550 Express Model 8204-E8A 629,159 2.49 US $ 04/20/08 03/20/08

HP 9000 Superdome Enterprise server comparison utilizes 64 cores - 875 MHz PA-RISC 8700 processors 256 gb of memory. Power consumption figures of 1400W for the IBM Power 550 and 16,920W for the HP 9000 Superdome were based on the maximum rates published by IBM and HP, respectively. For complete comparison claim details go to ibm.com/takebackcontrol/benchmark

11Competitive benchmark results reflect results published as of March 24, 2008. The results are the best IBM Power 550 8-core server result, the best Sun SPARC Enterprise M5000 server 16-core result, and the best HP Integrity rx6600 server 8-core result. Cells were left blank if there is no posted result for the benchmark. The TPC-C results can be found at www.tpc.org. The SPECint_rate2006, SPECfp_rate2006 and SPECjbb2005 results can be found at www.spec.org. The SAP SD 2-tier results can be found at www.sap.com.

12The TPC-C results are:

* Hardware Vendor System tpmC Price/tpmC System Availability Date Submitted
HP HP Integrity rx6600 372,140 1.81 US $ 06/11/07 06/11/07
IBM IBM Power 550 Express 629,159 2.49 US $ 04/20/08 03/20/08

The HP Integrity rx6600 server had 4-processor chips, 8-cores, and 2-threads per core.
The IBM Power 550 Express had 4-processor chips, 8-cores, and 2-threads per core.

13The SPECint_rate2006 results are:

Test Sponsor System Name Copies Cores Chips Cores / Chip Threads/Core Base Peak
Hewlett-Packard Company HP Integrity rx6600 (1.6GHz/24MB Dual-Core Intel Itanium 2) 8 8 4 2 1 94.7 102
IBM Corporation IBM System p 550 (4.2 GHz, 8 core, RedHat) 16 8 4 2 2 182 213
Sun Microsystems Sun SPARC Enterprise M5000 32 16 8 2 2 134 158

14The SPECfp_rate2006 results are:

Test Sponsor System Name Copies Cores Chips Cores / Chip Threads/Core Base Peak
Hewlett-Packard Company HP Integrity rx6600 (1.6GHz/24MB Dual-Core Intel Itanium 2) 8 8 4 2 1 69.1 71.4
IBM Corporation IBM System p 550 (4.2 GHz, 8 core) 16 8 4 2 2 156 178
Sun Microsystems Sun SPARC Enterprise M5000 32 16 8 2 2 133 140

15The SPECjbb2005 results are:

HW System Name # JVM Instances Cores Chips HW Threading bops bops/JVM
HP Integrity rx6600 Server 4 8 4 Yes 158174 39544
IBM System p 550 (4.2 GHz, 8 core) 4 8 4 Yes 333779 83445

16The results for 2-tier SAP SD are

Date of Certification (mm/dd/yyyy) Technology Partner Number of Benchmark Users Central Server (*) Additional Specifications Chips Cores Threads Certification Number
11/27/2006 HP 2150 HP Integrity rx6600 4 8 16 2006083
1/29/2008 IBM 3104 IBM System p 550 4 8 16 2008002

17The entry systems performance per core is calculated by dividing the values in the midrange systems performance table by the number of cores in each benchmarked system. All the IBM Power 550 Express server benchmarks in the table used 8-cores. All the results for the HP Integrity rx6600 used 8-cores. The results for the Sun SPARC Enterprise M5000 used 16-cores.

18The SPECint_rate2006 results for peak performance for 4-core blade servers of each technology are:

Test Sponsor System Name Base Processor (Cores) Chips Cores/ Chip Cores Threads/ Core Base Peak
IBM Corporation IBM BladeCenter JS22 (4.0 GHz, 4-core, POWER6) 8 4 2 2 4 2 77.2 84.7
Sun Microsystems Sun Blade T6320 (SPARC) 63 8 1 8 8 8 73.1 78.6
Sun Microsystems Sun Blade X6250 (Opteron) 4 4 2 2 4 1 53.8 65
Hewlett-Packard Company Proliant BL460c (3.0 GHz, Intel Xeon processor X5365) 4 4 1 4 4 1 54 61.1
Hewlett-Packard Company HP Integrity BL860c (1.6 GHz/18 MB Dual-Core Intel Itanium 2) 4 4 2 2 4 1 49.7 53.9

Since there was not a 4-core result for a SPARC processor-based server, the 8-core result for the Sun Blade T6320 was used.

The best performance per core for from 2 to 4-core blade servers of each technology based on SPECint_rate2006 is:

Test Sponsor System Name Base Base Processor (Cores) Chips Cores/ Chip Cores Threads/ Core Base Peak Peak/ Core
IBM Corporation IBM BladeCenter JS22 (4.0 GHz, 4-core, POWER6) 8 4 2 2 4 2 77.2 84.7 21.18
Sun Microsystems Sun Blade X6250 (Opteron) 4 4 2 2 4 1 53.8 65 16.25
Hewlett-Packard Company Proliant BL460c (3.0 GHz, Intel Xeon processor X5365) 4 4 1 4 4 1 54 61.1 15.28
Hewlett-Packard Company HP Integrity BL860c (1.6 GHz/18 MB Dual-Core Intel Itanium 2) 4 4 2 2 4 1 49.7 53.9 13.48
Sun Microsystems Sun Blade T6320 (SPARC) 63 8 1 8 8 8 73.1 78.6 9.83

Since there was not a 2- or 4-core result for a SPARC processor-based server, the 8-core result for the Sun Blade T6320 was used.

The best performance per 42U rack for from 2- to 4-core blade servers of each technology based on SPECint_rate2006 is:

Test Sponsor System Name Base Base Processor (Cores) Chips Cores/ Chip Cores Threads/ Core Base Peak Peak/ Core
IBM Corporation IBM BladeCenter JS22 (4.0 GHz, 4-core, POWER6) 8 4 2 2 4 2 77.2 84.7 4743
Hewlett-Packard Company Proliant BL460c (3.00 GHz, Intel Xeon processor X5365) 4 4 1 4 4 1 54 61.1 3910
Sun Microsystems Sun Blade T6320 (SPARC) 63 8 1 8 8 8 73.1 78.6 3773
Hewlett-Packard Company ProLiant BL465c (3.0 GHz AMD Opteron 2222) 4 4 2 2 4 1 54.4 58.9 3770
Hewlett-Packard Company HP Integrity BL860c (1.6 GHz/18 MB Dual-Core Intel Itanium 2) 4 4 2 2 4 1 49.7 53.9 1725

Since there was not a 2 or 4-core result for a SPARC processor-based server, the 8-core result for the Sun Blade T6320 was used.

19The SPECfp_rate2006 results for peak performance of 4-core blade servers for each technology are:

Test Sponsor System Name Base Cores Chips Cores/ Chip Threads/ Core Base Peak
IBM Corporation IBM BladeCenter JS22 (4.0 GHz, 4-core, POWER6) 8 4 2 2 2 65.7 75
Sun Microsystems Sun Blade T6320 (SPARC) 63 8 1 8 8 58.1 62.3
Hewlett-Packard Company ProLiant BL465c (3.0 GHz AMD Opteron 2222) 4 4 2 2 1 51.6 52.5
Hewlett-Packard Company HP Integrity BL860c (1.6 GHz/18 MB Dual-Core Intel Itanium 2) 4 4 2 2 1 46.9 48.8
Fujitsu Siemens Computers PRIMERGY BX620 S3, Intel Xeon processor 5160, 3.0 GHz 4 4 2 2 1 43.9 45.5

Since there was not a 4-core result for a SPARC processor-based server, the 8-core result for the Sun Blade T6320 was used.

The best performance per core for from 2- to 4-core blade servers of each technology based on SPECfp_rate2006 is:

Test Sponsor System Name Base Cores Chips Cores/ Chip Threads/ Core Base Peak Peak/ Core
IBM Corporation IBM BladeCenter JS22 (4.0 GHz, 4-core, POWER6) 8 4 2 2 2 65.7 75 18.8
Hewlett-Packard Company ProLiant BL465c (3.0 GHz AMD Opteron 2222) 2 2 1 2 1 26 26.5 13.3
Hewlett-Packard Company HP Integrity BL860c (1.6 GHz/18 MB Dual-Core Intel Itanium 2) 4 4 2 2 1 46.9 48.8 12.2
Fujitsu Siemens Computers PRIMERGY BX620 S3, Intel Xeon processor 5160, 3.0 GHz 4 4 2 2 1 43.9 45.5 11.4
Sun Microsystems Sun Blade T6320 (SPARC) 63 8 1 8 8 58.1 62.3 7.8

Since there was not a 2 or 4-core result for a SPARC processor-based server, the 8-core result for the Sun Blade T6320 was used.

The best performance per 42U rack for from 2- to 4-core blade servers of each technology based on SPECfp_rate2006 is:

Test Sponsor System Name Base Cores Chips Cores/ Chip Threads/ Core Base Peak Peak/ Core
IBM Corporation IBM BladeCenter JS22 (4.0 GHz, 4-core, POWER6) 8 4 2 2 2 65.7 75 4200
Hewlett-Packard Company ProLiant BL465c (3.0 GHz AMD Opteron 2222) 2 2 1 2 1 26 26.5 1696
Hewlett-Packard Company HP Integrity BL860c (1.6 GHz/18 MB Dual-Core Intel Itanium 2) 4 4 2 2 1 46.9 48.8 1562
Fujitsu Siemens Computers PRIMERGY BX620 S3, Intel Xeon processor 5160, 3.0 GHz 4 4 2 2 1 43.9 45.5 2730
Sun Microsystems Sun Blade T6320 (SPARC) 63 8 1 8 8 58.1 62.3 2990

Since there was not a 2 or 4-core result for a SPARC processor-based server, the 8-core result for the Sun Blade T6320 was used.

20The SPEC OMPM2001 results for peak performance of 4-core blade servers for each technology are:

Tester Name System Name Cores Chips Max OpenMP Threads Base Peak
IBM IBM BladeCenter JS22 (4.0 GHz, 4-core, POWER6) 4 2 8 17980 19688
AMD Supermicro X7DBR-8+ (Intel 64-bit Xeon 5160) 4 2 4 8408 8497
Sun Microsystems, Santa Clara Sun Blade X6220 (Opteron) 4 2 4 13345 13868
Hewlett-Packard Company hp server rx7620 (1500 MHz Itanium 2) 4 4 4 6555 6886
Sun Microsystems, Palo Alto Sun SPARC Enterprise T5120 8 1 64 14399 16208

There were no blade server results for Xeon, Itanium or SPARC. Therefore the Supermicro Motherboard result was used for Xeon. The HP Integrity 7620 (mid-range 10U rack server) result was used for Itanium. The Sun SPARC Enterprise T5120 (1U Rack server) was used for SPARC.

The best performance per core for from 2- to 4-core blade servers of each technology based on SPEC OMPM2001 is:

Tester Name System Name Cores Chips Max OpenMP Threads Base Peak Peak/ Core
IBM IBM BladeCenter JS22 (4.0 GHz, 4-core, POWER6) 4 2 8 17980 19688 4922
AMD Supermicro X7DBR-8+ (Intel 64-bit Xeon 5160) 4 2 4 8408 8497 2124
Sun Microsystems, Santa Clara Sun Blade X6220 4 2 4 13345 13868 3467
Hewlett-Packard Company hp server rx7620 (1500 MHz Itanium 2) 4 4 4 6555 6886 1722
Sun Microsystems, Palo Alto Sun SPARC Enterprise T5120 8 1 64 14399 16208 2026

There were no blade server results for Xeon, Itanium or SPARC. Therefore the Supermicro Motherboard result was used for Xeon. The HP Integrity 7620 (mid-range 10U rack server) result was used for Itanium. The Sun SPARC Enterprise T5120 (1U Rack server) was used for SPARC.

The best performance per rack for from 2- to 4-core blade servers of each technology based on SPEC OMPM2001 is:

Tester Name System Name Cores Chips Max OpenMP Threads Base Peak Peak / Rack
IBM IBM BladeCenter JS22 (4.0 GHz, 4-core, POWER6) 4 2 8 17980 19688 1102528
AMD Supermicro X7DBR-8+ (Intel 64-bit Xeon 5160) 4 2 4 8408 8497  
Sun Microsystems, Santa Clara Sun Blade X6220 4 2 4 13345 13868 665664
Hewlett-Packard Company hp server rx7620 (1500 MHz Itanium 2) 4 4 4 6555 6886 27544
Sun Microsystems, Palo Alto Sun SPARC Enterprise T5120 8 1 64 14399 16208 680736

There were no blade server results for Xeon, Itanium or SPARC. Therefore the Supermicro Motherboard result was used for Xeon. The HP Integrity 7620 (mid-range 10U rack server) result was used for Itanium. The Sun SPARC Enterprise T5120 (1U Rack server) was used for SPARC

21The comparison of the IBM Power 550 Express server with the best performing HP 9000 Superdome is shown in the table below. The Power 550 had 4-processor chips, 8-cores, and 2-threads per core. The HP Superdome had 64-processor chips, 64-cores, and 1 thread per core.

* Hardware Vendor System tpmC Price/tpmC System Availability Date Submitted
HP HP Superdome – PA-RISC/875 MHz-64p/64c 541,673 11.66 US $ 01/30/04 07/30/03
IBM IBM System p 550 Express Model 8204-E8A 629,159 2.49 US $ 04/20/08 03/20/08