Intel and AMD Servers Compared for Power-Efficiency

White paper with detail test results: [Word Format] [PDF Format]

Laboratory Home
Test Results
This Site
Email List
Neal Nelson
& Assoc. Home

Chicago, IL, August 3, 2007 -- Intel's recent claim of power efficiency superiority over AMD is being questioned by Neal Nelson & Associates, an independent Chicago area computer testing firm.

During a July 18, 2007 analyst conference call, Intel (NASDAQ: INTC) CEO Paul Otellini referred to an Intel lead over AMD (NYSE: AMD) in power efficiency. In an attempt to verify the accuracy of this statement, Neal Nelson & Associates has been conducting a series of power efficiency tests on Intel Xeon and AMD Opteron based servers. Initial test results indicate that there are some cases where servers manufactured by Intel trail AMD based servers in power efficiency, and sometimes by wide margins.

"It appears that when Intel chips are installed in Intel motherboards and sold as Intel servers the Intel claim of superior power efficiency is not supported by the empirical data," observed Neal Nelson, owner of the independent testing lab.

Recently conducted tests at various transaction processing load levels indicate that while an Intel Xeon based server was 1.4 to 5.1 percent more power efficient when both machines were configured with 4 gigabytes of main memory, an AMD Opteron based server was 6.1 to 12.7 percent more power efficient when both machines were configured with 8 gigabytes of memory.

In addition, when the systems were idle and waiting for more transactions to process, the AMD server was 33.3 percent more power efficient at the 4 gigabyte, and 43.3 percent more power efficient at the 8 gigabyte memory size.

This information about power consumption while the servers are idle is particularly significant since many servers spend most of their time waiting for work. In a November 16, 2006 press release(1), IBM quotes a report by the Robert Frances Group(2) which states the average utilization of most processors in a datacenter is between 15 and 20 percent. The IBM document continues by saying "Therefore, managing a server at its least productive state becomes critical to managing the issue of energy consumption in the datacenter".

The Nelson test results are consistent with other data that has recently appeared in the technical press. A July 17, 2007 article at AnandTech(3) reported on the issue of Intel versus AMD server power consumption and also identified cases where AMD had an advantage.

Neal Nelson's test results were generated by his second generation Server Power-Efficiency Benchmark. The test is a client server benchmark where world wide web transactions are processed against a server configured with SUSE Linux Enterprise Server from Novell, the Apache2 web server software and the MySQL relational database.

The test subjects a server to various user load conditions, reports the power consumed at each load level and allows quick and meaningful comparisons of server electrical power consumption.

Nelson's firm has a long history of data processing consulting to some of the world's largest computer customers including the U.S. Army, U.S. Navy, the Internal Revenue Service, McDonalds, WalMart and Federal Express. The Neal Nelson benchmarking laboratory is available to commercial and government users for independent computer performance tests.

A white paper with more detailed information about this test is available by clicking the links toward the top of this page.

(2) "The Rise to Power of Power: Dealing with the New Data Center Constraint," Jerald Murphy, Robert Frances Group. (2006)

Copyright © 2006-2008
Neal Nelson & Associates
Trademarks which may be mentioned on this site are the property of their owners.