Servers

AMD beats Intel in power efficiency


Some of you TechRepublic members might remember a statement made by Intel CEO Paul Otellini in July, when he claimed that Intel was the leader in power efficiency.

Well, that contention apparently did not sit too well with certain folks. In fact, computer performance firm Neal Nelson & Associates decided to come out with a benchmark report, pitting an AMD Opteron-based server up against an Intel Xeon-based server.

Based on the results, the firm claims that AMD-based servers beat Intel in 36 of the 57 power efficiency tests it conducted. A home-cooked benchmark was used, measuring Web transactions processed under Novell's SuSE Linux Enterprise Server, Apache2, and MySQL. Tests were performed on servers configured with various amounts of memory and load levels.

Excerpt from The Register:

The results show that under certain configurations and load levels, the Intel server was 2.4 to 11.7 percent more power efficient. But in a majority of cases, the AMD server was 9.2 to 23.1 percent more efficient.

Perhaps more significantly, when the systems were idle and waiting for transactions to process, the AMD server was 30.4 to 53.1 more power efficient. If accurate, it's a noteworthy figure, considering many servers spend most of their time waiting for work.

Basically, the tests show that Intel's power efficiency decreases as memory size goes up. AMD's power efficiency also apparently increases as the memory is upped.

Intel disputes the results. According to Intel rep Nick Knupffer in an e-mail:

The report ignores performance, in that you'd use less Intel servers to get the same job done, meaning less electricity is needed.

You can download the white paper with the benchmark results.

Do you factor in power efficiency when purchasing a new server?

About

Paul Mah is a writer and blogger who lives in Singapore, where he has worked for a number of years in various capacities within the IT industry. Paul enjoys tinkering with tech gadgets, smartphones, and networking devices.

8 comments
retro77
retro77

Another point for AMD. I have been an AMD fan/user for many years now. I have enjoyed the great performance and lower price of my AMD processors.

paulmah
paulmah

Do you factor in the power consumption when purchasing a new server?

Wrathlon
Wrathlon

It would be the right thing to do in this day and age but as we all know in the true corporate state of being it is not given enough consideration , or at least that is my experience. As the business world becomes more and more competitive, including the IT or I should say especially the IT aspect, then perhaps as it should this will become more of a consideration when purchasing equipment not only for the reason of performance but also its power consumption. It seems that the manufacturers are getting the message that one day these huge corporations no matter how large and successful they are can no longer ignore the necessity of paying closer attention to power consumption.

jdclyde
jdclyde

When someone says it will cost X dollars a day to run server A and Y dollars a day to run server B, no one in purchasing will care. And the savings will have to be REAL numbers. Saving 30 cents will not be worth the time to research it.

Absolutely
Absolutely

Why do you say that it "would be the right thing to do in this day and age" to make power consumption a higher priority in server purchase decisions? I'm all for a clean environment, but the idea that "it should ... become more of a consideration when purchasing equipment not only for the reason of performance but also its power consumption" is an example of sequencing criteria incorrectly. It's more important for the environment, "in this day and age" [i]of centralized production and distribution of power[/i], to support research into and development of means of converting energy to electricity cleanly, because the capacity of the power grid will always depend on maximum load, not average load. Just like you need your PSUs and UPSes (AD/DC converters & battery backups) to be rated to withstand a FWHM (sustained load, not peak) equal to the maximum power to be drawn on it, the bottom line for the environment is the load under extreme weather due to air conditioners and heaters. That maximum load [i]determines[/i] the average output of the grid. The impact on the environment of corporate computers' power consumption, therefore, isn't really significant because the coal and gas are going to be burned regardless of how many of the generated electrons get used. Companies that can afford to do so may as well leave the lights on for all the impact it will have on the environment.

Absolutely
Absolutely

"I take power comsumption into a very heavy consideration on what I buy. It dosent matter for home or for work. After all the less you use the less demand the less demand the less coal being burned in the long run." I just explained why that isn't so. It's a nice idea, but not how things really work. "When a corporate bill goes from $15000 to $38000 wouldn't you think twice. Yes it may seem a small bit that you save but every little bit counts." For the corporation or its agents to consider the monetary cost would be the consistent with my position, not contrary to it.

TonytheTiger
TonytheTiger

that due to "Green efforts", the server will be down from 8:00 AM to 4:00 PM :)

rmathis
rmathis

That is true "absolutely" but take into consideration one thing. This is also true on a corp scale. When the price of your power bill goes from 120 to 315 in one month becuase of the demand that is being placed down you and try to minamize what you use in your house? When a corp bill goes from $15000 to $38000 wouldnt you think twice. Yes it may seem a small bit that you save but every little bit counts. I take power comsumption into a very heavy consideration on what I buy. It dosent matter for home or for work. After all the less you use the less demand the less demand the less coal being burned in the long run. 1 person wont make a differance but a lot of people will if they do the same.

Editor's Picks