UPDATED November 2011!
On the 2U Dell PowerEdge R710, the Dell performance team has produced an industry-leading 2-U, 2-socket SPECpower_ssj2008 single-server score of 3,214 overall ssj_ops/watt.This result has been published by SPEC. The top 2U, 2-socket single-node results using Intel Xeon X5675 processors are summarized in the graph below; as you can see, the PowerEdge R710 has surpassed similar results, including the HP ProLiant DL380 G7 and IBM x3650 M3 (as of 11/21/2011):
Dell PowerEdge M915: The Most Energy-Efficient 4-socket on the Market!
Update November 2011: With the announcement of AMD’s Opteron 6200-series processors, the M915 maintains its world record position, improving the full-chassis (8-blade) score by nearly 8%. Thanks to our recently submitted world record score of2,919 SPECpower_ssj2008 overall ssj_ops/watt  on the PowerEdge M915, Dell servers maintain the highest SPECpower_ssj2008 score of all 4-socket servers!
In August 2011, the SPECpower_ssj2008 benchmark was used to compare the power efficiency of full-chassis blade solutions using 4-socket blades from Dell and HP. The results showed the Dell blade solution with PowerEdge M915 blades delivered 87% higher raw performance than the HP solution with ProLiant BL680c G7 blades. The Dell solution also has 92% higher power efficiency, and even with 8 blades (compared to 4 blades for the HP solution) consumes 35% less power at idle.
The full whitepaper study can be found here.The PowerEdge M915 also has a published result on SPEC.org: the 8-node M915 achieved 14,793,524 ssj_ops@100% target load and a SPECpower_ssj2008 result of 2,716 overall ssj_ops/Watt. As of August 2011, this is the highest overall ssj_ops/Watt of any 4-socket server on the market!
Benchmark results based on results published at www.spec.org as of August 2011. For the latest SPECpower_ssj2008 benchmark results, visit http://www.spec.org/power_ssj2008/results/power_ssj2008.html.
In June 2011, the Dell Performance Analysis team used the industry-standard SPECpower_ssj2008 benchmark to compare the performance, power draw, and performance/watt of AMD Opteron 6000 series-based four-socket rack servers from Dell and HP. The results showed the Dell PowerEdge R815 costs less and consumed less power in all tested configurations, while consuming half the rack space, and delivering better raw performance in all comparisons when both systems used AMD Opteron Model 6176 processors.Here are some of the advantages uncovered during this study:
The whitepaper study, which details the methodology and findings can be found here.
In August 2010, the Dell Performance Analysis team used the industry-standard SPECpower_ssj2008 benchmark to compare the power draw and performance/watt of current Intel Xeon 5600-series based blade solutions from Dell, HP, and IBM. The purpose of the testing was to examine on a level playing field the true power efficiency of the top three global blades server providers and compare and contrast the results. Each blade configuration included an enclosure configured as similarly as possible and was fully populated with each company’s best-selling blade servers using identical processor, RAM and hard drive selections.The results were clear: in like-for-like comparisons, a Dell PowerEdge M1000e enclosure fully populated with 16 PowerEdge M610 blade servers consistently demonstrated significant advantages over both the IBM BladeCenter H enclosure, fully populated with 14 HS22 blade servers, and the Hewlett-Packard C7000 enclosure, fully populated with 16 ProLiant BL460c G6 blade servers, in both power draw and performance-per-watt, regardless of utilization levels:
For more details about the methodology, and full disclosure of the results, download and read the technical whitepaper.UPDATE (April 2011): Since the original study, HP and IBM have released new Platinum-level efficiency power supplies and updated fans. These configurations were updated accordingly and have been re-tested to compare with the Dell results collected in 2010. The Dell result was not re-tested because the power supplies previously used were already Platinum-level efficiency, and all other components have remained performance and power neutral.At the time of the new test, Dell also measured the HP BL460c G7 blade servers and compared this data with the G6 server data gathered from the previous test. All of the results fell within the same performance and power level as the previous BL460c G6 results. Thus, it was concluded that the new G7-generation BL460c blade servers do not appreciably impact the performance or power characteristics of the HP blade solution as compared to the G6 results.Here is an updated whitepaper and graph, which shows Dell still maintains overall performance-per-watt leadership:
On the brand-new AMD Opteron-based Dell PowerEdge R815, the Dell performance team has produced the industry-leading 4-socket score of 2,696 overall ssj_ops/watt (disclosure on www.spec.org).All 4-socket single-node results are summarized in the graph below (as of 7/8/10):
Dell has also produced a 2-socket score of 2,357 overall ssj_ops/watt which beats a comparable HP ProLiant DL385 G7 result -- and the R815 is expandable to 4 processors in a 2U form factor!
SPEC and the benchmark name SPECpower_ssj are trademarks of the Standard Performance Evaluation Corporation. Benchmark results stated above reflect results published on http://www.spec.org as of July 8, 2010. For the latest SPECpower_ssj2008 benchmark results, visit http://www.spec.org/power_ssj2008/results/power_ssj2008.html.
SPECpower_ssj2008 is the first industry-standard SPEC benchmark that evaluates the power and performance characteristics of volume server class and multi-node class computers. With SPECpower_ssj2008, SPEC is defining server power measurement standards in the same way we have done for performance.The drive to create the power and performance benchmark comes from the recognition that the IT industry, computer manufacturers, and governments are increasingly concerned with the energy use of servers. Currently, many vendors report some energy efficiency figures, but these are often not directly comparable due to differences in workload, configuration, test environment, etc. Development of this benchmark provides a means to measure power (at the AC input) in conjunction with a performance metric. This should help IT managers to consider power characteristics along with other selection criteria to increase the efficiency of data centers.The initial benchmark addresses only one subset of server workloads: the performance of server side Java. It exercises the CPUs, caches, memory hierarchy and the scalability of shared memory processors (SMPs) as well as the implementations of the JVM (Java Virtual Machine), JIT (Just-In-Time) compiler, garbage collection, threads and some aspects of the operating system. The benchmark runs on a wide variety of operating systems and hardware architectures and should not require extensive client or storage infrastructure. Additional workloads are planned.(source: http://www.spec.org/power_ssj2008/)Dell is an active member company of the SPEC Power committee and was a significant contributor to the code development, testing, documentation, and methodology used in SPECpower_ssj2008.