SeaMicro Unveils its Low Power Server

10 comments

The new SM-10000 server from startup SeaMicro, which can pack 2,048 CPUs into a rack.

The new SM-10000 server from startup SeaMicro, which can pack 2,048 CPUs into a rack.

Startup server maker SeaMicro today unveiled a new low-power server that promises to slash power costs for companies running large Internet services and cloud computing platforms. SeaMicro’s multi-core x86 server runs on Intel’s low-power Atom chips, whose energy efficiency has made them the processor of choice for many mobile phones and laptops.

The power profile of the SM10000 allows SeaMicro to pack 512 Atom CPUs into a 10U server, providing the option to fit as many as 2,048 CPUs into a single seven foot, 40U rack. The company says each of its servers use less than 2 kilowatts of power, suggesting that a rack filled with SeaMicro servers can have a power load of under eight kilowatts – a manageable power load for most data centers optimized for high-density computing.

‘Revolutioning’ Data Center Power? 
The server’s tantalizing power claims align with SeaMicro’s aim to “revolutionize the data center landscape” by slashing the power used in IT operations. Last year the Santa Clara company got a $9.3 million grant from the U.S. Department of Energy to further its development of technology to make data centers more energy efficient.

Atom chips aren’t the only key to SeaMicro’s power profile. The company has developed i/o virtualization technology that allows it to remove many of the components seen on a traditional server. I/O virtualization allows companies to save money by using fewer cables and newtork interface cards (NICs) to connect to networks and storage.

Interconnection Fabric and Load Balancing 
As we noted in January, SeaMicro has also designed an interconnection and switching fabric that links the 512 “mini-motherboards” in the SM10000. SeaMicro says its network fabric can support Ethernet, fibre channel and data center Ethernet.

It also includes a load-balancing feature called Dynamic Compute Allocation Technology (DCAT) that provisions traffic and workloads within the SM1000. This also provides power management advantages, allowing the server to focus workloads on a set of CPUs within the server while placing other CPUs in sleep mode to preserve power.  “A utilization threshold for a pool of compute can be set, and if met, CPUs can be dynamically added or removed from the pool,” SeaMicro says in its system overview (PDF).  

The basic building block is a credit card-sized compute block, comprised of an Intel Atom CPU and its chipset, DRAM, and a custom SeaMicro ASIC (application specific integrated circuit). Eight of these compute blocks are aggregated on a 5×11 inch motherboard, and 64 of the motherboards fit into a 10U SeaMicro server.

Optimized for Large Platforms
SeaMicro’s stripped-down server will be of immediate interest to companies running large-scale server farms that harness armadas of CPUs to specific tasks.   

“Historically, servers were designed to quickly solve a relatively small number of very hard problems,” SeaMicro says in a press release. “The Internet, however, changed this. In the Internet data center, the challenge is to handle millions of relatively small, independent tasks like those needed for searching, social networking, viewing web pages, and checking email. Volume servers failed to adapt to this fundamental change.

“This mismatch between volume servers and the now dominant Internet workload is the primary cause of the rapid increase in server power consumption and is responsible for the multi-billion dollar power problem in the data center. A completely new server design optimized for today’s data center was necessary.”

SeaMicro was started in July 2007 and got funded in December 2007. The Santa Clara, Calif. company is backed by well-known names in the Silicon Valley venture capital community, including Khosla Ventures, Draper Fisher Jurvetson and Crosslink Capital.

About the Author

Rich Miller is the founder and editor at large of Data Center Knowledge, and has been reporting on the data center sector since 2000. He has tracked the growing impact of high-density computing on the power and cooling of data centers, and the resulting push for improved energy efficiency in these facilities.

Add Your Comments

  • (will not be published)

10 Comments

  1. Interesting they set a rack standard at 8Kw, vs 4.8 (total draw, 4 Kw IT load) which is standard. Is that 8Kw of critical IT load or is it really 9.6 of draw?

  2. Mark: That's my math based on the SeaMicro docs stating less than 2 kilowatts of power usage for the 10U unit. SeaMicro is clearly saying you can four SM10000s in a rack to replace 40 1U servers, with a total 2,048 CPUs for the SeaMicro units. It could be that I'm missing some overhead in my math. If you have a better reference on the power draw, please share here. They indicate that the server has an upper temperature range of 95 degrees, which is interesting given the ongoing discussion of running gear at higher temperatures.

  3. anon

    hardware accelerator cards for web servers and related applications would have a broader market - and therefore more of an impact on the power problem if we are selling using this angle - not many folk are going to be buying these 10U monsters.

  4. Rich, I will take a stab at the math and see what I can find out. As far as the temperature range - it is interesting, but from my sperspective it begets the question - does the whole computer room need to run high temp gear? Seems to me that if these are running 20+degrees hotter than 'normal' then saying you don't need to provision as much cooling means for these ONLY and not the surrounding racks/room. I can see where this may cause an issue with surrounding gear, or take more cooling if they run hot and you are trying to keep an entire room at a level temp.

  5. luke-warm

    the intel atom 330 is dual core and has hyper threading. nVidia makes better CHIPSETS then intel.