Skip to main content

Compare UNIX Servers

POWER7 substantiation

1. POWER7® systems deliver up to three or four times the performance with less energy than POWER6® based systems.

Substantiation:

  • rPerf (Relative Performance) is an IBM estimate of commercial processing performance relative to other IBM UNIX® systems.
  • The comparison used in the claim is based on these comparisons:
    • 4-node Power™ 570 (POWER6+™) to a 1-node Power 780 (POWER7)
    • 4-node Power 570 (POWER6+) to a 3 node Power 780 (POWER7)
    • 4-node Power 570 (POWER6+) to a 3-node Power 770 (POWER7)
    • 2-node Power 560 Express (POWER6+) to Power 750 Express (POWER7)
  • Performance/WATT is calculated by dividing the performance (rPerf) from the tables below by the recommended maximum power usage for site planning. This defines the requirement for the power infrastructure. Actual power used by the systems will be less than this value for all of the systems. This information is available in the site planning guides available through www.ibm.com.
System name Nodes Processor
technology		 Processor
frequency		 Energy
(Watts)		 rPerf Factor*
(P7 over P6)
IBM Power 780 1 POWER7 3.8 GHz 1,600 195 1.38
IBM Power 570 4 POWER6+ 5.0 GHz 5,600 141 -
IBM Power 780 3 POWER7 3.8 GHz 4,800 523 3.7
IBM Power 570 4 POWER6+ 5.0 GHz 5,600 141 -
IBM Power 770 3 POWER7 3.1 GHz 4,800 443 3.1
IBM Power 570 4 POWER6+ 5.0 GHz 5,600 141 -
IBM Power 750 Express 1 POWER7 3.55 GHz 1,950 331 3.3
IBM Power 560 Express 2 POWER6+ 3.6 GHz 2,400 100 -

* Factor = Performance increase factor of POWER7 system over POWER6 system for less energy

2. POWER7 systems deliver up to three or four times the energy efficiency of POWER6 based systems.

Substantiation:

  • Benchmark results stated reflect results published on www.spec.org (link resides outside ibm.com) as of February 8, 2010. The comparison used in the claim is based on a consolidation of the best high-end POWER6 result (Power 595) to the Power 780, the best mid-range POWER6 result (Power 570) to the Power 770, and the best four-socket and above POWER6 Express results with the Power 750 Express. For the latest SPEC® benchmark results, visit http://www.spec.org (link resides outside ibm.com).
  • Performance/WATT is calculated by dividing the performance from the tables below by the recommended maximum power usage for site planning. This defines the requirement for the power infrastructure. Actual power used by the systems will be less than this value for all of the systems. This information is available in the site planning guides available through www.ibm.com.
SPECint_rate2006 results as of February 18, 2010
System name Cores Chips Cores/
chip		 Threads/
core		 Peak WATTs Peak /
WATT
IBM Power 780 64 8 8 4 2,530 6,400 0.39
IBM Power 595 64 32 2 2 2,160 28,300 0.07
IBM Power 770 64 8 8 4 2,013 6,400 0.31
IBM Power 570 16 8 2 2 542 5,600 0.09
IBM Power 750 Express 32 4 8 4 1,060 1,950 0.54
IBM Power 560 Express 16 8 2 2 363 2,400 0.15
IBM Power 550 Express 8 4 2 2 263 1,400 0.18

3. 24 Sun SPARC Enterprise T2000 systems can be consolidated to a single Power 710 1 socket system, saving 95% of the cores for software licensing, 95% on floor space, and 94% on energy.


Power 710 Express Sun SPARC Enterprise T2000 Delta
Power (Watts) 650 450  
Systems 1.0 24.0  
Total Watts 650 10,800 94.0%
SPECjbb2005 607,514 74,356 8.17
Cores 8 192 95.8%
Performance 8.17    
Virtualization (3X) 3.0    
T2000 systems 24.5    
Space (rack units) 2 2  
Total Space 2 48 95.80%

Substantiation:

  • Competitive benchmark results stated reflect results published on www.spec.org as of August 17, 2010. The comparison presented below is based on a consolidation of a legacy 8-core Sun SPARC Enterprise T2000 UltraSPARC T1 servers into a 8 core IBM Power 730. For the latest SPEC benchmark results, visit http://www.spec.org (link resides outside ibm.com).
  • SPECjbb2005 results are:
    • POWER7: IBM Power 710 Express with 1 chips, and 8 cores and four threads per core with a result of 607,514 bops and 75,939 bops/jvm submitted to SPEC on August 17, 2010.
    • SPARC: Sun Microsystems Sun SPARC Enterprise T2000 with 1 chips, 8 cores and 4 threads per core with a result of 74,356 bops and 18,591 bops/jvm.

The virtualized system count and energy savings were derived from several factors:

  • A performance ratio factor of 24.7X was applied to the virtualization scenario based on SPECjbb2005. Power 710 (8-core, 1 chips, 8 cores per chip, 3.55 GHz) 607,514 bops, submitted on 8/17/2010; Sun SPARC Enterprise T2000 (8-core, 1 chips, 8 cores per chip) 1.4 GHz, SPECjbb2005 74,356 bops. The performance factor is simply the SPECjbb2005 result of the Power 710 Express divided by the result of the competitive Sun SPARC Enterprise T2000 server.
  • A virtualization factor of 3X was applied to the virtualization scenario using utilization assumptions derived from an Alinean white paper on server consolidation. The tool assumes 19% utilization of existing servers and 60% utilization of new servers. Source - www.ibm.com/services/us/cio/optimize/opt_wp_ibm_systemp.pdf.

Calculation summary: the Power 710 to the Sun T2000 performance ratio is 8.22 Multiply by 3 for the virtualization factor. Hence, 8.22 * 3 = 24 servers T2000 servers can be consolidated into one 730 server.

The Sun T2000 is 2U in height and 21 can fit into a 42U rack. The 710 is 2U in height. One 710 server is 8 cores per system. A Sun T2000 has 8 cores per system. 24 systems multiplied by 8 cores is 192 cores. The Power 710 Express has 95% less cores.

Power consumption figures of 1100W for the IBM Power 710 and 450W for the Sun T2000 were based on the maximum rates published by IBM and Sun Microsystems, respectively. This information for the 730 is in "Model 8231-E2B server specifications" available at http://www.ibm.com/common/ssi/fcgi-bin/ssialias?infotype=dd&subtype=sm&appname=pseries&htmlfid=897/ENUS8231-_h01. Sun T2000 Maximum AC power consumption of 450 WATTs was sourced from Sun SPAC Enterprise T2000 Servers site planning guide at http://docs.sun.com/app/docs/doc/819-2545-11 (link resides outside ibm.com) as of 8/17/2010.

4. Consolidate 47 HP ProLiant DL380 G5 servers to a single Power 730 2 socket system to reduce energy costs by 98%, floor space by 97% and use 83% fewer cores.


  Power 730 Express HP ProLiant DL380 G5 Delta
Power (Watts) 1,100 1,193  
Systems 1.0 47.0  
Total Watts 1,100 56,071 98.0%
SPECint_rate 575 36.2 15.88
Cores 16 94 83.0%
Racks 0,095238095 2.2 96.0%
Performance 15.88    
Virtualization (3X) 3.0    
HP DL380 G5 systems 47.7    
Space (racks units) 2 2  
Total Space 2 94 97.9%

Substantation:

  • Competitive benchmark results stated reflect results published on www.spec.org as of August 17, 2010. The comparison presented below is based on a consolidation of a legacy 8-core Sun SPARC Enterprise T2000 UltraSPARC T1 servers into a 8 core IBM Power 730. For the latest SPEC benchmark results, visit http://www.spec.org (link resides outside ibm.com).
  • SPECint_rate2006 results are:
    • POWER7: IBM Power 730 Express with 2 chips, and 16 cores and four threads per core with a result of 575 to SPEC on August 17, 2010.
    • x86: HP Proliant DL380G5 with 1 chip, 2 cores and 1 threads per core with a result of 36.2.

The virtualized system count and energy savings were derived from several factors:

  • A performance ratio factor of 15.8X was applied to the virtualization scenario based on SPECint_rate2006. Power 730 (16-core, 2 chips, 16 cores per chip, 3.55 GHz) 575, submitted on 8/17/2010; HP DL380 G5 (2-core, 1 chip, 2 cores per chip) 3.0 GHz, 36.2. The performance factor is simply the SPECint_rate2006 result of the Power 730 Express divided by the result of the competitive HP Proliant DL380 G5 server.
  • A virtualization factor of 3X was applied to the virtualization scenario using utilization assumptions derived from an Alinean white paper on server consolidation. The tool assumes 19% utilization of existing servers and 60% utilization of new servers. Source - www.ibm.com/services/us/cio/optimize/opt_wp_ibm_systemp.pdf.

Calculation summary: the Power 730 to the HP DL380 G5 performance ratio is 15.86 Multiply by 3 for the virtualization factor. Hence, 15.86 * 3 = 47 servers DL380 servers can be consolidated into one 730 server.

The HP DL380 is 2U in height and 21 can fit into a 42U rack. The 730 is 2U in height.

One 730 server is 16 cores per system. A HP DL380 G5 has 2 cores per system. 47 systems multiplied by 2 cores is 94 cores. The Power 730 Express has 83% less cores.

Power consumption figures of 1100W for the IBM Power 730 and 1193W for the HP DL380 G5 were based on the maximum rates published by IBM and Sun Microsystems, respectively. This information for the 730 is in "Model 8231-E2B server specifications" available at http://www.ibm.com/common/ssi/fcgi-bin/ssialias?infotype=dd&subtype=sm&appname=pseries&htmlfid=897/ENUS8231-_h01. HP DL 380 G5 Maximum AC power consumption of 1193 WATTs was sourced from HP Proliant DL380 G5 Servers at http://h18004.www1.hp.com/products/quickspecs/12477_na/12477_na.html# Power Specifications as of 8/17/2010.

5. The Power 730 Express system and the PS702 are the highest performing 2-socket systems and blades in the industry for Java or integer based workloads.

SPECjbb2005 results – All results are the best result for the system indicated as posted at http://www.spec.org (link resides outside ibm.com) on August 16, 2010 except for the results for the IBM Power 795 and the IBM Power 730 Express which were submitted to SPEC as of August 17, 2010. For the latest SPEC benchmark results, visit http://www.spec.org (link resides outside ibm.com).

SPECjbb2005 results as of August 16, 2010
Company and system (2-socket) BOPS BOPS per JVM JVM instances Cores Chips Cores per chip Published
IBM Power 780 (3.86 GHz, 16 core) 1,331,641 83,228 16 16 2 8 May-10
IBM Power 730 (3.55 GHz, 16 core) 1,216,983 76,061 16 16 2 8 Aug-10
IBM BladeCenter PS702 Express (3.0 GHz, 16 core) 1,119,946 69,997 16 16 2 8 May-10
IBM BladeCenter PS702 Express (3.0 GHz, 16 core, SLES) 1,103,231 68,952 16 16 2 8 Jun-10
IBM System x3690 X5 1,015,260 126,908 8 16 2 8 Jul-10
Dell PowerEdge R810 (Intel Xeon X7560, 2.26 GHz) 1,011,147 126,393 8 16 2 8 Apr-10
IBM Power 750 Express (3.30 GHz, 16 core, IBM i) 976,223 61,014 16 16 2 8 Feb-10
Sun Blade X6270 M2 server module 931,637 465,819 2 12 2 6 Jul-10
PowerEdge R715 (AMD Opteron 6176 SE, 2.30 GHz) 931,368 232,842 4 24 2 12 Jul-10
Cisco UCS B200 M2 931,076 155,179 6 12 2 6 Apr-10
PRIMERGY BX924 S2, Intel Xeon X5680, 3.33 GHz 929,050 154,842 6 12 2 6 Jul-10
PRIMERGY RX300 S6, Intel Xeon X5680, 3.33 GHz 928,393 154,732 6 12 2 6 Apr-10
PRIMERGY TX300 S6, Intel Xeon X5680, 3.33 GHz 928,393 154,732 6 12 2 6 Apr-10
PRIMERGY BX922 S2, Intel Xeon X5680, 3.33 GHz 927,872 154,645 6 12 2 6 Apr-10
PRIMERGY RX200 S6, Intel Xeon X5680, 3.33 GHz 927,067 154,511 6 12 2 6 Apr-10
PowerEdge R815 (AMD Opteron 6176 SE, 2.30 GHz) 926,093 231,523 4 24 2 12 Jun-10
ProLiant DL385 G7 918,378 229,595 4 24 2 12 Apr-10
IBM System x3500 M3 916,251 152,709 6 12 2 6 Apr-10
IBM BladeCenter HS22V 915,323 152,554 6 12 2 6 Jul-10
IBM System x3650 M3 915,103 152,517 6 12 2 6 Apr-10
IBM System x3550 M3 913,505 152,251 6 12 2 6 Apr-10
IBM BladeCenter HS22 913,456 152,243 6 12 2 6 Jul-10
HP ProLiant SL165z G7 901,488 225,372 4 24 2 12 Apr-10
PRIMERGY TX200 S6, Intel Xeon X5670, 2.93 GHz 888,823 148,137 6 12 2 6 Aug-10
PRIMERGY BX920 S2, Intel Xeon X5670, 2.93 GHz 882,267 147,045 6 12 2 6 May-10
System x3620 M3 880,876 146,813 6 12 2 6 Jul-10
ProLiant DL180 G6 880,048 146,675 6 12 2 6 May-10
iDataPlex Server dx360 M3 879,266 146,544 6 12 2 6 Apr-10
ProLiant ML350 G6 877,364 146,227 6 12 2 6 Apr-10
IBM System x3400 M3 877,231 146,205 6 12 2 6 Apr-10
HP ProLiant DL360 G7 875,975 145,996 6 12 2 6 Apr-10

6. The Power 730 Express system and the PS702 are the highest performing 2-socket systems and blades in the industry for Java or integer based workloads.

SPECint_rate2006 results – All results are the best result for the system indicated as posted at http://www.spec.org (link resides outside ibm.com) on August 16, 2010 except the results for the IBM Power 795 and Power 730 Express which were submitted to SPEC as of August 17, 2010. For the latest SPEC benchmark results, visit http://www.spec.org (link resides outside ibm.com).

SPECint_rate2006 results as of August 16, 2010
Company and system (2 sockets) Result Baseline Cores Chips Cores per chip Published
IBM Power 780 (3.86 GHz, 16 core) 652 586 16 2 8 Mar-10
IBM Power 730 (3.55 GHz, 16 core) 578   16 2 8 Aug-10
IBM BladeCenter PS702 Express (3.0 GHz, 16 core) 520 456 16 2 8 Apr-10
PowerEdge R715 (AMD Opteron 6176 SE, 2.30 GHz) 402 312 24 2 12 Jul-10
PowerEdge R815 (AMD Opteron 6176 SE, 2.30 GHz) 401 314 24 2 12 Jun-10
ProLiant DL385 G7 (2.3 GHz AMD Opteron 6176 SE) 398 309 24 2 12 Mar-10
ProLiant DL585 G7 (2.3 GHz AMD Opteron 6176 SE) 398 308 24 2 12 Jun-10
IBM System x3690 X5 (Intel Xeon X7560) 390 364 16 2 8 Jul-10
ProLiant DL165 G7 (2.2 GHz AMD Opteron 6174) 386 302 24 2 12 Mar-10
PowerEdge R810 (Intel Xeon X7560, 2.27 GHz) 386 360 16 2 8 Apr-10
ProLiant SL165z G7 (2.2 GHz AMD Opteron 6174) 386 301 24 2 12 Mar-10
PowerEdge M910 (Intel Xeon X7560, 2.26 GHz) 385 360 16 2 8 Jun-10
ProLiant BL465c G7 (2.2 GHz AMD Opteron 6174) 385 299 24 2 12 Jun-10
NovaScale R480 F2 (Intel Xeon X7560, 2.27 GHz) 385 359 16 2 8 Jul-10
PowerEdge R910 (Intel Xeon X7560, 2.27 GHz) 385 359 16 2 8 Jul-10
ProLiant BL685c G7 (2.2 GHz AMD Opteron 6174) 384 302 24 2 12 Jun-10
PRIMERGY RX600 S5, Intel Xeon X7560, 2.26 GHz 383 358 16 2 8 Jul-10
ProLiant DL580 G7 (2.27 GHz, Intel Xeon X7560) 382 355 16 2 8 Aug-10
ASUS RS700-E6 server system (Intel Xeon X5680) 382 356 12 2 6 Jun-10
PRIMERGY BX922 S2, Intel Xeon X5680, 3.33 GHz 381 354 12 2 6 May-10
PowerEdge M710 (Intel Xeon X5680, 3.33 GHz) 380 355 12 2 6 Mar-10
Cisco UCS B200 M2 (Intel Xeon X5680, 3.33 GHz) 380 355 12 2 6 Apr-10
ProLiant DL380 G7 (3.33 GHz, Intel Xeon X5680) 376 352 12 2 6 May-10

7. The IBM Power 750 Express is the highest performing 4-socket system on the planet. In addition it outperforms all other non-IBM 8 and 16-socket systems.


System name Cores Chips Cores/ chip Threads/ core Peak*
IBM Power 750 32 4 8 4 1,060
HP ProLiant DL585 G6 (2.8 GHz AMD Opteron 8439 SE) 24 4 6 1 416
HP Integrity rx6600 (1.6 GHz/24MB Dual-core Intel Itanium 2) 8 4 2 1 102
HP ProLiant DL580 G5 (2.66 GHz, Intel Xeon X7460) 24 4 6 1 291
Sun SPARC Enterprise T5440 32 4 8 8 360
Sun SPARC Enterprise M4000 16 4 4 2 152
HP ProLiant DL 785 G6 (2.8 GHz AMD Opteron 8439 SE) 48 8 6 1 800
Unisys ES7000 Model 7600R, Intel Xeon X7460, 2.66 GHz 48 8 6 1 527
Sun SPARC Enterprise M5000 32 8 4 2 296
HP Integrity rx8640 (1.6 GHz/24MB Dual-core Intel Itanium 2) 16 8 2 1 209
Unisys ES7000 Model 7600R, Intel Xeon X7460, 2.66 GHz 96 16 6 1 1,049
Sun SPARC Enterprise M8000 64 16 4 2 753

* Peak = SPECint_rate2006 (Peak)

Substantiation:

  • Competitive benchmark results reflect results published as of February 3, 2010. The results are the best results for four-socket single (non-clustered) systems using POWER®, Intel® x86, AMD Opteron™ x86, SPARC and Intel Itanium® processors. IBM Power 750 result submitted on February 8, 2010.
  • For the latest SPEC benchmark results, visit http://www.spec.org (link resides outside ibm.com).

8. The IBM Power 750 Express is the most energy efficient 4-socket system on the planet.


System name Cores Chips Cores/ chip Threads/ core Peak* WATTs Peak / WATT
IBM Power 750 32 4 8 4 1,060 1,950 0.54
HP ProLiant DL585 G6 (2.8 GHz AMD Opteron 8439 SE) 24 4 6 1 416 1,548 0.26
HP Integrity rx6600 (1.6 GHz/24MB Dual-core Intel Itanium 2) 8 4 2 1 102 1,600 0.06
HP ProLiant DL580 G5 (2.66 GHz, Intel Xeon X7460) 24 4 6 1 291 1,412 0.20
Sun SPARC Enterprise T5440 32 4 8 8 360 2,629 0.13
Sun SPARC Enterprise M4000 16 4 4 2 152 2,016 0.07

* Peak = SPECint_rate2006 (Peak)

Substantiation:

  • Competitive benchmark results reflect results published as of February 3, 2010. The results are the best results for four-socket single (non-clustered) systems using POWER, Intel x86, Opteron x86, SPARC and Itanium processors. IBM Power 750 result submitted on February 8, 2010.
  • For the latest SPEC benchmark results, visit http://www.spec.org (link resides outside ibm.com).
  • Performance/WATT is calculated by dividing the performance from the tables below by the recommended maximum power usage for site planning. This defines the requirement for the power infrastructure. Actual power used by the systems will be less than this value for all of the systems. For Power Systems™ servers, this information is available in the site planning guides available through www.ibm.com. For HP systems, this information is contained in the QuickSpecs for each system available through www.hp.com (link resides outside ibm.com).
  • For Sun systems, this information is available through the Site Planning Guides and product datasheets available through www.oracle.com (link resides outside ibm.com).

9. The IBM Power 750 Express has more SAP performance than any 8-socket system in the industry – and is even comparable to a 128-core, 32-socket Sun M9000.

Substantiation: All results are 2-tier, SAP EHP 4 for SAP ERP 6.0 (Unicode). IBM results valid as of 2/8/2010. Competitive results valid as of 2/3/2010.

  • IBM Power 750 Express: 4p / 32–c / 128 – t, POWER7, 3.55 GHz, 256 GB memory, 15,600 SD users, dialog resp.: 0.98s, line items/hour: 1,704,330, Dialog steps/hour: 5,113,000, SAPS: 85,220, DB time (dialog/ update):0.015s / 0.028s, CPU utilization: 99%, AIX® 6.1, DB2® 9.7, cert# 2010004.
  • Sun SPARC Enterprise T5540: 4p / 32-c / 256 –t, UltraSPARC T2 plus OC, 1.6 GHz, 256 GB memory, 4720 SD users, dialog resp: 0.97s, line items/hour: 516,670, dialog steps/hour: 1,550,000, SAPS: 25,830, Solaris 10, Oracle 10g , cert# 2009026.
  • HP DL585 G6: 4p / 24-c / 24-t, AMD Opteron 8439 SE, 2.8 GHz, 64 GB memory, 4665 SD users, dialog resp: 0.96s, line items/hour: 510,670, dialog steps/hour: 1,532,000, SAPS: 25,530, Windows Server 2008 EE, , SQL Server 2008, cert#: 2009025.
  • HP DL785 G6: 8p / 48-c / 48-t, AMD Opteron 8439 SE, 2.8 GHz, 128 GB memory, 8280 SD users, dialog resp: 0.96s, line items/hour: 907,000, dialog steps/hour: 2,721,000, SAPS: 45,350, Windows Server 2008 EE, , SQL Server 2008, cert#: 2009035.
  • Sun Fire x4640: 8p / 48-c / 48–t, Six-core AMD Opteron 8435, 2.6 GHz, 256 GB memory, 10,000 SD users, dialog resp: 0.9s, line items/hour: 1,101,330, dialog steps/hour: 3,304,000, SAPS: 55,070, Solaris 10, Oracle 10g, cert# 2009049.
  • HP DL380 G6: 2p / 8-c / 16-t, Intel Xeon® X5570, 2.93 GHz, 48 GB memory, 3171 SD users, dialog resp: 0.94s, line items/hour: 347,670, dialog steps/hour: 1,043,000, SAPS: 17,380, SUSE Linux® Enterprise Server 10, MaxDB 7.8, cert#: 2009006.
  • Sun SPARC Enterprise M9000: 32p / 128-c / 256–t, Six-core AMD Opteron 8435, 2.6 GHz, 1 TB memory, 17,430 SD users, dialog resp: 0.95s, line items/hour: 1,909,670, dialog steps/hour: 5,729,000, SAPS: 95,480, Solaris 10, Oracle 10g, cert# 2009038.

10. Consolidation onto POWER7 can deliver significant savings. One hundred Sun SPARC Enterprise T2000 servers can be consolidated into a single IBM Power 750 Express system, saving 95% of the cores for software licensing, 97% of the floor space, and 95% of the maximum energy requirement.

Calculation summary: the Power 750 has 33.334 better SPECjbb2005 performance than the Sun T2000. Assuming a 3x virtualization factor for greater consolidation - then 100 Sun Fire T2000 servers could be consolidated onto one Power 750 Express server (3.334 * 3 = 100.002 servers rounded to 100 T2000 servers).

Hardware system name JVM Instances Cores Processor chips Hardware threading bops bops/JVM
IBM Power 750 Express 32 32 4 Yes 2,478,929 77,467
Sun File T2000 4 8 1 Yes 74,365 18,591

System name SPECjbb 2005 Max Watts Rack space Cores Systems Total performance Total cores Total Watts Total rack space
IBM Power 750 Express 2,478,929 1,950 4 32 1 1,380,000 32 1,950 4
Sun File T2000 74,365 450 2 8 92 1,368,150 736 41,400 184
Savings with Power 750 Express             95.6% 95.2% 97.8%

Substantiation:

  • Competitive benchmark results stated reflect results published on http://www.spec.org (link resides outside ibm.com) as of February 08, 2010. The comparison presented below is based on a consolidation of a legacy 8-core Sun SPARC Enterprise T2000 UltraSPARC T1 server into a 32 core IBM Power 750. For the latest SPEC benchmark results, visit http://www.spec.org (link resides outside ibm.com).
  • SPECjbb2005 results are:
    • POWER7: IBM Power 750 Express with 4 chips, and 32 cores and four threads per core with a result of 2,478,929 bops and 77,467 bops/jvm submitted to SPEC on February 8, 2010.
    • SPARC: Sun Microsystems Sun SPARC Enterprise T2000 with 1 chip, 8 cores and 4 threads per core with a result of 74,356 bops and 18,591 bops/jvm.

The virtualized system count and energy savings were derived from several factors:

  • A performance ratio factor of 33.334X was applied to the virtualization scenario. The performance factor is the SPECjbb2005 result of the Power 750 Express divided by the result of the competitive Sun SPARC Enterprise T2000 server.
  • A virtualization factor of 3X was applied to the virtualization scenario using utilization assumptions derived from an Alinean white paper on server consolidation. The tool assumes 19% utilization of existing servers and 60% utilization of new servers. Source - www.ibm.com/services/us/cio/optimize/opt_wp_ibm_systemp.pdf.

Space calculation: The Sun T2000 is 2U in height and 21 can fit into a 42U rack. The 750 is 4U in height.

Power consumption figures of 1950W for the IBM Power 750 and 450W for the Sun T2000 were based on the maximum rates published by IBM and Sun Microsystems, respectively. This information for the Power 750 is in "Model 8233-E8B server specifications" available at http://www.ibm.com/common/ssi/fcgi-bin/ssialias?infotype=dd&subtype=sm&appname=pseries&htmlfid=897/ENUS8233-_h01. Sun T2000 Maximum AC power consumption of 450 WATTs was sourced from Sun SPAC Enterprise T2000 Servers site planning guide at http://docs.sun.com/app/docs/doc/819-2545-11 (link resides outside ibm.com) as of 2/9/2010.

11. 345 million kilowatt-hours are used yearly by the 91,920* Sun SPARC Enterprise T2000 servers shipped since 2005 above what would be used yearly if consolidated into 1,000 IBM Power 750 Express servers at the rate of 100 to 1.

That’s enough electricity to supply 34,500 homes for a year.**

Substantiation:

  • Maximum power for 1 IBM Power 750 Express server = 1,950 watts
  • Maximum power for 100 Sun Fire T2000 servers = 100 x 450 = 45,000 watts
  • Excess power per consolidation instance (100 Sun T2000’s into one Power 750) = 43,050 watts
  • Number of consolidations required = 914 (91,320 / 100)
  • Total excess kilowatt-hours per year =
    43,050 watts x 24 hrs/day x 365 days/yr x 1,000 = 345 million kilowatt-hours per year

* Source: 3Q09 IDC Server Tracker
** Source: Wikipedia estimate of average annual household energy use of 10,000 kilowatt-hours

12. The IBM Power 750 Express has 28% more performance than a 64-core HP Integrity Superdome and requires only 83% as much power to run – at a fraction of the price.

Substantiation:

  • HP Integrity Superdome benchmark results stated reflect results published on http://www.spec.org (link resides outside ibm.com) as of February 08, 2010. For the latest SPEC benchmark results, visit http://www.spec.org (link resides outside ibm.com).
  • SPECint_rate2006 Peak results are:
    • POWER7: IBM Power 750 Express with 4 chips, and 32 cores and four threads per core with a result of 1060 submitted to SPEC on February 8, 2010.
    • Itanium: Hewlett-Packard Integrity Superdome with 32 chips, 64 cores, and one thread per core with a result of 824.
    • The HP Integrity Superdome is a rack cabinet. The 750 is 4U in height.

Power consumption is derived from the recommended maximum power for site planning. Actual power used by the systems will be less than this value for all of the systems. This information for the Power 750 Express is available at http://www.ibm.com/common/ssi/fcgi-bin/ssialias?infotype=dd&subtype=sm&appname=pseries&htmlfid=897/ENUS8233-_h01. The maximum power requirement for the Power 750 is 1,950 Watts.

The information for the Integrity Superdome is in "QuickSpecs HP Integrity rx6600 Server" available at http://h18000.www1.hp.com/products/quickspecs/11717_div/11717_div.HTML (link resides outside ibm.com), which shows the maximum power requirement for the Integrity Superdome of 12,196 VA. Using the Power Factor of 0.95 shown at http://www.spectra.com/pdfs/superdome.pdf (PDF, 434KB) (link resides outside of ibm.com), the maximum input power is 11,586 Watts.

Price comparison based on IBM analysis:
HP Superdome price estimated at $2,117,000 for the configuration described in the SPECint_rate2006 benchmark.

IBM Power 750 Express U.S. list price = $275,420.

13. The IBM Power 780 delivers leadership performance and consolidation capability vs. HP and Sun high-end servers. For example, eight HP Integrity Superdome 64 core systems utilized at 30% can be consolidated into a single IBM Power 780 server utilized at 80%, thus saving 87% of the cores for software licensing, reducing floor space from 80 square feet to 7.6 square feet, and reducing energy costs by 92%.


SPECint rate2006 results
System name Cores Chips Cores / chips Threads / core Peak Published Wattage PPW* PPC**
IBM Power 780 64 8 8 4 2,530 February 2010 6,400 395.31 39.53
HP Integrity Superdome 64 32 2 1 824 October 2006 12,196 67.56 12.88
HP Integrity Superdome 128 64 2 1 1,648 September 2006 24,392 67.56 12.88
Sun SPARC Enterprise M9000 256 64 4 4 2,586 October 2009 44,800 57.72 10.10

* Performance per watt
** Performance per core

Substantiation: Performance per watt is calculated by dividing the performance in the table above by the recommended maximum power for site planning. Actual power used by the systems will be less than this value for all of the systems. The maximum power requirement for the Power 780 is 6,400 Watts and is available at http://www.ibm.com/common/ssi/fcgi-bin/ssialias?infotype=dd&subtype=sm&appname=pseries&htmlfid=897/ENUS9179-_h01.

Power consumption figures of 6400 W for the IBM Power 780, 12,196 W / 24,392 W for the HP Superdome and 44,800 W for the Sun SPARC Enterprise M9000 were based on the maximum rates published by IBM, HP and Sun Microsystems, respectively. The information for the HP Integrity Superdome is in “QuickSpecs HP Integrity Superdome Servers 16- processor, 32-processor, and 64- processor Systems” available at www.hp.com (link resides outside ibm.com). The information for the Sun SPARC Enterprise M9000 is in the "Sun SPARC Enterprise M9000 Servers Site Planning Guide" available at www.sun.com (link resides outside ibm.com).

The virtualized system count and energy savings were derived from several factors:

A performance ratio factor was applied to the virtualization scenario based on SPECint_rate2006. The performance factor is simply the SPECint_rate2006 result per core of the Power 780 divided by the per core result of the HP or Sun system.

  • Power 780 (64-core, 8 chips, 8 cores per chip, 3.8 GHz) SPECint_rate2006 2,530 peak as of 2/8/2010.
  • HP Superdome (64-core, 32 chips, 2 cores per chip) 1.6 GHz, SPECint_rate2006 824 peak published October 2006. Data valid as of 2/3/2010.
  • Sun SPARC Enterprise M9000 (256-core, 64 chips, 4 cores per chip) 2.88 GHz, SPECint_rate2006 2,586 peak published October 2009. Data valid as of 2/3/2010. SPEC results available at: http://www.spec.org (link resides outside ibm.com).

A virtualization factor of 3.157X was applied to the virtualization scenario using utilization assumptions derived from an Alinean white paper on server consolidation. The tool assumes 19% utilization of existing servers and 60% utilization of new servers. Source - www.ibm.com/services/us/cio/optimize/opt_wp_ibm_systemp.pdf.

Air conditioning power requirement estimated at 50% of system power requirement.
Energy cost of $.1031 per kWh is based on 2009 YTD US Average Retail price to commercial customers per US DOE at http://www.eia.doe.gov/cneaf/electricity/epm/table5_6_b.html (link resides outside ibm.com) as of 1/27/2010.

The reduction in floor space, power, cooling and software costs depends on the specific customer, environment, application requirements, and the consolidation potential. Actual numbers of virtualized systems supported will depend on workload levels for each replaced system.

System data for HP from the HP Superdome Datasheet and HP Integrity Superdome Server — specifications both available at www.hp.com (link resides outside ibm.com). System data for Sun from the Sun SPARC Enterprise M9000 Tech Specs available at www.sun.com (link resides outside ibm.com). Data is current as of January 27, 2010.

14. Overall, the Power 780 delivered more than four times the performance-per-core of the fastest HP Itanium or Sun SPARC system and over 1.8x the performance-per-core of the fastest Intel x86-based system in the TPC-C benchmark.

Substantiation: Based on the TPC-C performance benchmark.
Transaction performance based on tpmC results as on 4/9/2010. Source: Transaction Processing Performance Council, www.tpc.org (link resides outside of ibm.com) as of 4/9/10. IBM result submitted on 4/13/10.

TPC-C benchmark results
Company System tpmC Price/ tpmC System availability Database Operating system Chips Cores Threads tpmC per core
Sun/Oracle T5440 7,646,486 $2.36 3/19/2010 Oracle 11g EE RAC Solaris 10 48 384 3,072 19,913
HP Integrity Superdome 4,092,799 $2.93 8/6/2007 Oracle 10g HP-UX 11i v3 64 128 256 31,975
IBM Power 780 1,200,011 $0.76 10/13/2010 DB2 9.1 AIX 6.1 2 8 32 150,001
HP Proliant DL370 G6 661,475 $1.16 2/1/2010 MS SQL Server 2005 MS Windows Server 2008 2 8 16 82,684

15. For businesses that run SAP, Power 780 handled 37,000 users on 64 cores – 16% more users than a 256-core Sun Enterprise M9000 and 130% more users than a 64-core Fujitsu system running Intel Xeon X7560 chips.

Substantiation: Based on the SAP Sales & Distribution 2-Tier benchmark.
IBM Power System 780, 8p / 64–c / 256–t, POWER7, 3.8 GHz, 1024 GB memory, 37,000 SD users, dialog resp.: 0.98s, line items/hour: 4,043,670, Dialog steps/hour: 12,131,000, SAPS: 202,180, DB time (dialog/ update):0.013s / 0.031s, CPU utilization: 99%, OS: AIX 6.1, DB2 9.7, cert# 2010013; SUN M9000, 64p / 256-c / 512–t, 1156 GB memory, 32,000 SD users, SPARC64 VII, 2.88 GHz, Solaris 10, Oracle 10g , cert# 2009046; Fujitsu 1800E, 8p / 64-c / 128-t, 512 GB memory, 16,000 SD users, Intel Xeon X7560, 2.26 GHz, Windows Server 2008 R2 DE, SQL Server 2008, cert#: 2010010. All results are 2-tier, SAP EHP 4 for SAP ERP 6.0 (Unicode) and valid as of 4/1/2010. Source: http://www.sap.com/solutions/benchmark/sd2tier.epx (link resides outside ibm.com).

16. The Power 780 also demonstrated the ability to deliver leadership, workload-optimized performance by setting new performance records across the three major industry standard processor benchmarks for Java, integer and high performance computing workloads achieving between 1.8 and 3.0 times the performance of all other competitive published 8-socket results.

Substantiation: Based on the SPEC benchmarks (SPECjbb2005, SPECint_rate2006, SPECfp_rate2006).
IBM Power 780 64-core (3.86 GHz, 8 chips, 8 cores/chip,4 threads/core) SPECint_rate2006 result of 2,530is best in class 8-socket system; IBM Power 780 64-core (3.86 GHz, 8 chips, 8 cores/chip,4 threads/core) SPECfp2006 result of 2,240 is best in class 8-socket system; IBM SPECjbb2005 result of 5,210,501 bops (81,414 bops/JVM) on a 64-core (8 chips, 128 threads) 3.86 GHz IBM Power 780 System running AIX V6.1 is best in class 8-socket system.
The tables below show the best IBM Power 780 result for the benchmarks and the best non-IBM result for each benchmark.

SPECjbb2005
Company System BOPS BOPS per JVM JVM JVM instances Cores Chips Cores per chip Published
IBM Corporation IBM Power 780 5,210,501 81,414   64 64 8 8 Apr-2010
Hewlett-Packard Company HP DL785 G6 1,984,616 248,077 "IBM J9 VM (build 2.4, J2RE 1.6.0 IBM J9 2.4 Windows Server 2008 amd64-64 jvmwa64 60sr5-20090519_35743 (JIT enabled, AOT enabled)" 8 48 8 6 Sep-2009

SPECint2006 rates (8-sockets or chips)
Hardware vendor System Result Baseline Cores Chips Cores per chip Published
IBM Corporation IBM Power 780 (3.86 GHz, 64 core) 2,526 2,299 64 8 8 Mar-10
Fujitsu PRIMEQUEST 1800E (Intel Xeon X7560) 1,339 1,254 64 8 8 Mar-10

SPECFP2006 rates (8-sockets or chips)
Hardware vendor System Result Baseline Cores Chips Cores per chip Published
IBM Corporation IBM Power 780 (3.86 GHz, 64 core) 2,240 2,033 64 8 8 Mar-10
SGI SGI Altix ICE 8200EX (Intel Xeon X5570, 2.93 GHz) 742 723 32 8 4 May-09

17. The Power 780 also delivers unprecedented price for performance on transaction processing workloads.

The first server to deliver over 1.2 million transactions per minute for less than $0.75 per transaction, the Power 780 delivers the performance and scalability of large systems for the cost of a small system. The 1.2 million transactions per minute establishes a new record in performance per core – 4.6 times an HP Superdome and 7.5 times a Sun SPARC Enterprise T5440 cluster running Oracle RAC.

Substantiation for the Power 780 being the first server to deliver over 1.2 million transactions per minute for less than $.75 / transaction:
Top Ten TPC-C by Price/Performance Version 5 Results As of 11-Apr-2010 7:35 PM with the Power 780 result added as the last row of the table [GMT].

System tpmC Price/tpmC System availability Database Operating system TP monitor Date submitted Cluster
Dell PowerEdge T710 239,392 .50 USD 11/18/2009 Oracle Database 11g Standard Edition One Microsoft Windows Server 2003 Enterprise x64 Edition Microsoft COM+ 11/18/2009 N
HP ProLiant ML350 G6 232,002 .54 USD 5/21/2009 Oracle Database 11g Standard Edition One Oracle Enterprise Linux Microsoft COM+ 5/21/2009 N
Dell PowerEdge 2900 104,492 .60 USD 2/20/2009 Oracle Database 11g Standard Edition One Microsoft Windows Server 2003 Standard Ed. x64 Microsoft COM+ 2/20/2009 N
HP ProLiant DL385G7 705,652 .60 USD 9/1/2010 Microsoft SQL Server 2005 Enterprise x64 Edition SP3 Microsoft Windows Server 2008 R2 Enterprise Edition Microsoft COM+ 4/8/2010 N
Dell PowerEdge 2900 97,083 .68 USD 6/16/2008 Oracle Database 11g Standard Edition One Microsoft Windows Server 2003 Standard Ed. x64 Microsoft COM+ 6/16/2008 N
HP ProLiant ML350G5 102,454 .73 USD 12/31/2007 Oracle Database 11g Standard Edition One Microsoft Windows Standard x64 Etd. SP1 R2 Microsoft COM+ 9/12/2007 N
HP ProLiant ML350G5 100,926 .74 USD 6/8/2007 Oracle Database 10g Standard Edition One Oracle Enterprise Linux Microsoft COM+ 6/8/2007 N
HP ProLiant ML350G5 82,774 .84 USD 3/27/2007 Microsoft SQL Server 2005 x64 Enterprise Edt. SP1 Microsoft Windows 2003 x64 Server Std. Ed. Microsoft COM+ 3/27/2007 N
Dell PowerEdge 2950 III 20,705 .85 USD 8/5/2008 Sybase SQL Anywhere 11.0 Microsoft Windows 2003 x64 Standard R2 SP2 Microsoft COM+ 7/29/2008 N
"PowerEdge 2900/1/2.33GHz/2x4M" 69,564 .91 USD 3/9/2007 Microsoft SQL Server 2005 Standard Ed. Microsoft Windows 2003 Server Std Edt SP1 Microsoft COM+ 3/9/2007 N

System tpmC Price/tpmC System availability Database Operating system TP monitor Date submitted Cluster
Dell PowerEdge T710 239,392 .50 USD 11/18/2009 Oracle Database 11g Standard Edition One Microsoft Windows Server 2003 Enterprise x64 Edition Microsoft COM+ 11/18/2009 N

System tpmC Price/tpmC System availability Database Operating system TP monitor Date submitted Cluster
HP ProLiant ML350 G6 232,002 .54 USD 5/21/2009 Oracle Database 11g Standard Edition One Oracle Enterprise Linux Microsoft COM+ 5/21/2009 N
Dell PowerEdge 2900 104,492 .60 USD 2/20/2009 Oracle Database 11g Standard Edition One Microsoft Windows Server 2003 Standard Ed. x64 Microsoft COM+ 2/20/2009 N
HP ProLiant DL385G7 705,652 .60 USD 9/1/2010 Microsoft SQL Server 2005 Enterprise x64 Edition SP3 Microsoft Windows Server 2008 R2 Enterprise Edition Microsoft COM+ 4/8/2010 N
Dell PowerEdge 2900 97,083 .68 USD 6/16/2008 Oracle Database 11g Standard Edition One Microsoft Windows Server 2003 Standard Ed. x64 Microsoft COM+ 6/16/2008 N
HP ProLiant ML350G5 102,454 .73 USD 12/31/2007 Oracle Database 11g Standard Edition One Microsoft Windows Standard x64 Etd. SP1 R2 Microsoft COM+ 9/12/2007 N
HP ProLiant ML350G5 100,926 .74 USD 6/8/2007 Oracle Database 10g Standard Edition One Oracle Enterprise Linux Microsoft COM+ 6/8/2007 N
HP ProLiant ML350G5 82,774 .84 USD 3/27/2007 Microsoft SQL Server 2005 x64 Enterprise Edt. SP1 Microsoft Windows 2003 x64 Server Std. Ed. Microsoft COM+ 3/27/2007 N
Dell PowerEdge 2950 III 20,705 .85 USD 8/5/2008 Sybase SQL Anywhere 11.0 Microsoft Windows 2003 x64 Standard R2 SP2 Microsoft COM+ 7/29/2008 N
"PowerEdge 2900/1/2.33GHz/2x4M" 69,564 .91 USD 3/9/2007 Microsoft SQL Server 2005 Standard Ed. Microsoft Windows 2003 Server Std Edt SP1 Microsoft COM+ 3/9/2007 N
Power 780 1,200,011 $0.76 10/13/2010 DB2 9.1 AIX 6.1   Submitted as of April 13, 2010 N

Substantiation: the 150,001 tpmC/core sets a new record is based on the complete list of TPC-C results found at www.tpc.org (link resides outside ibm.com) on April 11, 2010.

Substantiation for the claim that the Power 780 performance per core is 4.6 times an HP Superdome and 7.5 times a Sun SPARC Enterprise T5440 cluster running Oracle RAC:

TPC-C benchmark results
Company System tpmC Price/tpmC System availability Database Operating system Chips Cores Threads tpmC per core
Sun/Oracle T5440 7,646,486 $2.36 3/19/2010 Oracle 11g EE RAC Solaris 10 48 384 3,072 19,913
HP Integrity Superdome 4,092,799 $2.93 8/6/2007 Oracle 10g HP-UX 11i v3 64 128 256 31,975
IBM Power 780 1,200,011 $0.76 10/13/2010 DB2 9.1 AIX 6.1 2 8 32 150,001
HP Proliant DL370 G6 661,475 $1.16 2/1/2010 MS SQL Server 2005 MS Windows Server 2008 2 8 16 82,684

18. The Power 780 uses up to 87% fewer cores than a Sun SPARC Enterprise Cluster to deliver over one million transactions per minute enabling clients to slash database licensing and maintenance costs by 80%. In addition the Power 780 is 3.4 times more energy efficiency.

Substantiation: Based on analysis of system performance and performance per core. The Power 780 has demonstrated 150k transaction per minute on the TPC benchmark (see above). Therefore 8-cores of performance can deliver over 1 million OLTP transactions per minute. The Sun SPARC Enterprise T5440 cluster has demonstrated a performance of 19.9k transaction per minute per core on the TPC benchmark. Based on this 2 Sun SPARC Enterprise T5440 systems running Oracle RAC would be required to deliver over 1 million OLTP transactions per minute. These systems have 32-cores each for a total of 64-cores. IBM Power can deliver over 1 million OLTP with 87% fewer cores (8 vs. 64) then a Sun SPARC Enterprise Cluster.

All database prices based on License plus 5 years of Maintenance. Pricing on IBM Power 780 is based on DB2 V9.7. Pricing on Sun SPARC Enterprise T5440 cluster are based on Oracle Enterprise Edition plus Oracle Real Application Clusters plus Oracle Partitioning which were all used to achieve the 19.9 thousand tpmC per core result in the TPC-C benchmark.

Oracle pricing and processor factors found at www.oracle.com (link resides outside of ibm.com) http://www.oracle.com/corporate/pricing/technology-price-list.pdf (PDF, 163KB) (link resides outside of ibm.com).

Oracle pricing
Oracle EE Pricing $ 47,500
Oracle EE Maintenance $ 10,450
Oracle RAC $ 23,000
Oracle RAC Maintenance $ 5,080
Oracle Partitioning $ 11,500
Oracle Partitioning Maintenance $ 2,530

DB2 Pricing $ / PVU 120
DB2 with 1 Yr S&S $ 405 $ 48,600
DB2 Maintenance $ 81 $ 9,720

Database price DB2 on Power 780 Oracle / Sun T5440 Cluster
Total Cores 8 64
Database License Price + 1 yr S&S $466,560 $3,201,920
Annual Maintenance Cost * $77,760 $577,920
5 year Database TCA (List Price) $777,600 $5,513,600

* Prices are list prices in USD as April 13, 2010. Prices from resellers may vary. All prices are subject to change without notice.

TPC benchmark information used to identify the configurations to be compared. This does not imply a projection of actual TPC benchmark performance of the Oracle configuration. TPC benchmark results are found at www.tpc.org (link resides outside ibm.com).

19. More than four times that of the Sun SPARC Enterprise M9000 and three times that of the HP SuperDome, the Power 795 is the ideal for large scale data center consolidation.


System Chip / Core / Thread Date SPECint_rate 2006 Per core Energy requirement (WATTs)
IBM Power 795 (4 GHz POWER7) 8 / 256 / 1024 August 17, 2010* 11,200 43.75 28,529
IBM Power 595 (5 GHz POWER6) 32 / 64 / 128 April 2008 2,083 32.5 27,500
Sun SPARC Enterprise M9000 64 /256 / 512 October 2009 2,586 10.1 38,180
HP Integrity Superdome (1.6 GHz Itanium 2) 64 / 128 / 128 September 2006 1,648 12.875 24,392

*IBM results submitted on August 17, 2010. All other results as of 08/05/10. Not all results listed.

Source: http://www.spec.org (link resides outside ibm.com)

Performance per KWatt is calculated by dividing the performance by the recommended maximum power usage for site planning. This defines the requirement for the power infrastructure. Actual power used by the systems will be less than this value for all of the systems. For HP systems, this information is contained in the QuickSpecs available through www.hp.com (link resides outside ibm.com). For Sun systems, this information is available through the respective Site Planning Guides available through www.sun.com (link resides outside of ibm.com).

20. The IBM Power 795 delivers leadership performance and consolidation capability vs. HP and Sun high-end servers. For example, ten 128 core HP Superdome systems running at 50% utilization can be consolidated on one Power 795 utilized at 80%, dramatically reducing the number of cores to be licensed for software, and energy consumption.

Substantiation:

Power consumption figure of the IBM Power 795 is the max published energy usage in the Power 795 announcement letter + max energy for 1 powered expansion unit. The power consumption figure of 24,392 W for the HP Superdome was based on the maximum rates published by HP. The information for the HP Integrity Superdome is in “QuickSpecs HP Integrity Superdome Servers 16- processor, 32-processor, 64- processor Systems” available at www.hp.com (link resides outside ibm.com). The virtualized system count and energy savings were derived from several factors:

A performance ratio factor was applied to the virtualization scenario based on SPECint_rate2006. The performance factor is simply the SPECint_rate2006 result per core of the Power 795 divided by the per core result of the HP or Sun system.

Power 795 (256-core, 32 chips, 8 cores per chip, 4.0 GHz) SPECint_rate2006 11,200 peak submitted on Aug 17, 2010. HP Superdome (128-core, 64 chips, 2 cores per chip) 1.6 GHz, SPECint_rate2006 1648 peak published September 2006. Data valid as 08/11/2010. SPEC results available at: www.spec.org (link resides outside ibm.com).

The reduction in floor space, power, cooling and software costs depends on the specific customer, environment, application requirements, and the consolidation potential. Actual numbers of virtualized systems supported will depend on workload levels for each replaced system.

System data for HP from the HP Integrity Superdome Server Quickspecs — specifications available at www.hp.com (link resides outside ibm.com). Data is current as of August 5, 2010.

SPECint_rate2006 results
System name Systems Cores/ system Chips/ system Cores / chip Threads/ core Peak result Date Utilization / system Effective Performance (result x systems x Util %) Energy (watts)
IBM Power 795 1 256 32 8 4 11,200 Aug 2010* 80% 8,960* 41,129 System + 1 Expansion
HP Superdome 10 128 64 2 1 1,648 Sep 2006 50% 8,240 24,392*10 = 243,920

* Power 795 (256-core, 32 chips, 8 cores per chip, 4.0 GHz) SPECint_rate2006 11,200 peak submitted on Aug 17, 2010.

21. Achieve 39% lower total cost of acquisition with a full BladeCenter H chassis with 7 two socket (16-core) PS702 blades instead of a full HP C7000 Blade chassis with 16 two socket (12-core) HP BL460c G6 blades, leveraging the higher utilization and virtualization efficiencies of Power blades.

  • Performance based on IBM analysis of benchmarks on 2.93GHz X5600 processors and benchmarks on the PS702.
  • Prices from www.hp.com (link resides outside of ibm.com).
  • Assumption is that the IBM BladeCenter PS702 will be 60% utilized with PowerVM as the virtualization technology and the HP ProLiant BL460c G6 will be 40% utilized with VMware as the virtualization technology.

22. POWER7 blades are better than Oracle Sun’s T6340 blades in every important performance category: relative performance, performance density, and performance per watt.

Substantiation:

  • For the latest SPEC benchmark results, visit http://www.spec.org. All results are the best result posted at www.spec.org (link resides outside ibm.com) as of April 5, 2010 for the system indicated except for the IBM BladeCenter result which was submitted to SPEC on April 12, 2010.
  • HP prices from Aquila Technology Website at http://www.aquilatech.co.nz/ (link resides outside ibm.com).
  • Sun T6340 energy usage and performance density based on use in Sun Blade 6048 chassis. Maximum energy requirement and space requirement from the Sun Blade 6048 Chassis Datasheet available through www.oracle.com (link resides outside ibm.com). All data current as of April 5, 2010.
SPECint_rate2006 results as of April 5, 2010
System Name Cores Chips Cores / chip Threads / core Peak Peak / core
IBM Bladecenter PS702 16 2 8 4 520 32.5
HP Integrity BL860c i2 8 2 4 2 134 16.7
Sun Blade T6340 16 2 8 8 160 10.0

SPECfp_rate2006 results as of April 5, 2010
System Name Enabled cores Enabled chips Cores / chip Threads / core Peak result Peak / core
IBM BladeCenter PS702 16 2 8 4 431 26.9
HP Integrity BL860c i2 8 2 4 2 136 17.0
Sun Blade T6340 Server 16 2 8 8 121 7.5

SPECjbb2005 results as of April 5, 2010
HW System Name JVM instances Cores Chips Hardware threading bops bops/JVM bops / core
IBM BladeCenter PS702   16 2 Yes 1,119,946   7,496
Sun Blade T6340 16 16 2 Yes 69,997 24,279 24,278

Additional information about these benchmarks and associated trademarks

Get Adobe® Reader®

Content navigation

Related links