Posts tagged with Energy Efficiency
Energy Efficiency Globally Must Start Locally
Posted by Guest Blogger in 2:02 pm
(3 votes, average: 4.33 out of 5)
Loading ...New York State is currently facing some difficult challenges including rising energy prices, an aging electricity delivery infrastructure, an imbalanced electricity generation portfolio and climate change. According to a 2007 EPA study conducted by Lawrence Berkeley National Laboratory and sponsored by AMD, New York’s data centers – home to second largest concentration of data centers in the country – consume an average 4.5 billion kilowatt hours a year. This is the equivalent of nearly 700,000 single family homes with a year’s supply of electricity — at a cost of roughly $594 million. To meet these challenges, energy efficiency must play a central role in reducing consumption and improving reliability.
With that in mind, NYSERDA, AMD, New York State, HP and GLOBALFOUNDRIES have all come together to address these issues head on, discussing them at the latest NY State Performance Computing Seminar on October 28.
The New York State Energy Research and Development Authority (NYSERDA), in collaboration with AMD, have been instrumental in driving sustainable computing and business practices in New York. Improvements in operating practices, coupled with installation of energy efficient systems, can enable significant energy savings and help reduce the strain information technology and data centers place on the electric grid while helping to ensure a reliable and affordable supply of electricity. In addition, by improving the energy efficiency of data centers and working in synergy with NYSERDA and AMD, New York State and its IT businesses and data centers can make considerable strides toward achieving their respective energy and environmental goals, while supporting economic development in this growing industry. More importantly, investments in energy efficient systems can help improve a data centers’ bottom line.
NYSERDA’s Industrial and Process Efficiency program plans to invest more than $100 million over the next three years in new and existing manufacturing and data center facilities that help reduce energy consumption. These funds can encourage sustainable load growth and help to significantly reduce the use of electricity and natural gas.
For more information, visit www.nyserda.org
Sal Graven is a Technical Information Associate at the New York State Energy Research and Development Authority (NYSERDA)
His postings are his own opinions and may not represent AMD’s positions, strategies or opinions. Links to third party sites are provided for convenience and unless explicitly stated, AMD is not responsible for the contents of such linked sites and no endorsement is implied.
The Scorecard
Posted by John Fruehe in 9:43 am
There’s an old saying in baseball - “you can’t tell the players without a scorecard.” In this age of always-connected information, that scorecard is the internet. It’s amazing how accurate that information can be. And not be. All at the same time.
I am constantly asked about our different products, both existing and future. Clarification of what the AMD Opteron TM processor products are and where they are targeted is a common request, so I thought it might be a good idea to put it all down in one place; a “cheat sheet” for the IT professional. When it comes to the future products, you might see a lack of details. This is on purpose, as there is some information that we don’t release until we launch the products. (When we discuss the target markets, we are speaking in general terms, because, depending on applications, actual processor choices could vary. That is why we recommend talking to your OEM or system integrator to choose the best solution.)
The Current lineup:
Quad-Core AMD OpteronTM processor (formerly codenamed “Shanghai”) – This is a 45nm quad-core processor with a 6MB level 3 cache. It fits into all of the existing Socket F (1207) systems and is targeted at current workloads like web services, network infrastructure, departmental applications, technical workloads, and those applications that favor clock frequency over thread count. It is productized as the AMD Opteron 2000 Series processors (2P) and AMD Opteron 8000 Series (4P and 8P) processors.
Six-Core AMD OpteronTM processor (formerly codenamed “Istanbul”) – This is also a 45nm design that is based on the same core as the Quad-Core AMD Opteron processor, but the design includes 6 cores, not 4, teamed up with the 6MB L3 cache, and plugs into the Socket F (1207) systems. Because of the higher number of cores (12 cores in a 2P system and 24/48 cores in a 4P/8P system), customers typically use these processors for workloads like cloud computing, virtualization, database and HPC where workloads can be very threaded.
Quad-Core AMD OpteronTM processor (formerly codenamed “Suzuka”) – This is the single socket version of the “Shanghai” die, focused on 1P servers that are typically utilized for web serving, remote locations or running small businesses. You’ll see all of the same features of the “Shanghai” processor, with the exception that it is available only in the standard power band (which is by far the most popular choice for AMD Opteron 1000 Series processors.)
The Future Lineup:
In Q1 2010 we plan to introduce the “Maranello” platform, featuring the processor variant currently codenamed “Magny-Cours.” This is a new socket (G34) and the processor is expected to merge both the top end of the 2P market with the 4P/8P market, all conveniently in a single processor, the AMD Opteron 6000 Series processor. Core choices are expected to be 8 and 12 cores, with massive memory scalability through the 4 channels of DDR-3 memory per processor. By utilizing the same processor for both 2P and 4P designs, the AMD Opteron 6000 Series processor should enable several very interesting and flexible platforms with scalability of 16 cores through 48 cores. Clearly this processor is targeted at virtualization, HPC, database and business applications.
In Q2, we plan to introduce a new platform for web, cloud and infrastructure applications – the “San Marino” platform, featuring the 4-core and 6-core processor variants currently codenamed “Lisbon” in the C32 socket. With low core counts, these processors are expected to be a favored choice for applications that scale well up to 8-12 threads. The platform choices around “San Marino” are expected to help OEMs optimize their C32 systems for low power consumption and low cost. We believe these AMD Opteron 4000 Series processors will have the potential to help users achieve new levels of price/performance and performance/watt. When you walk through a data center and see rack after rack of servers, it’s clear that reducing the cost and power footprint of the “workhorse” servers can have a huge impact on the bottom line. Two channels of DDR-3 memory help provide the right level of scalability for these 1P/2P designs while contributing to low power consumption and cost.
Best of all, while the C32 and G34 sockets are physically different, the chipsets, cores and main BIOS core functions are expected to be common across both of these platforms, helping enable OEMs to develop platforms around AMD offerings, and contributing to easier deployment and management by end users.
The consolidation of 1000/2000/8000 to the 4000/6000 product line is expected to reduce the number of overlapping platforms, increase commonality and flexibility for customers, and reduce the overall SKU count for OEMs – contributing to more flexibility and cleaner scalability from 4 cores to 48 cores. Truly a re-definition of the server market that focuses on how industry partners take products to market and customers deploy instead of how processor manufacturers see the world. Customer-centric innovation.
Power Bands:
Customers have a variety of needs, and it would be foolish to think that one processor can solve all of your processing challenges. So, just as we have different models (1000/2000/8000 today and 4000/6000 in the future) we have different power bands to meet specific power needs.
By far the most popular model is the “standard power” with a 75W ACP (average CPU power), which doesn’t even have a designator. This is “Opteron classic” if you are filling in your score card, the choice for price/performance. In addition to this model, there are 3 specialty power bands:
SE – for those that want relatively higher raw performance. By driving to a 105W ACP, we can increase the clock speed for customers running frequency-dependent applications.
HE – Delivering a lower ACP (55W), the HE processors focus on delivering great price/performance/watt for environments where power may be constrained/more expensive or where density is an issue (like with blades)
EE – This is the specialty processor that delivers absolutely the lowest power consumption of any AMD Opteron processor, with a 40W ACP. Customers, like cloud/web 2.0, look for processors like the EE to help reduce the total power per rack because they are in extremely dense environments.
The “Maranello” platform is expected to support SE, Standard and HE power bands, and the “San Marino” platform is planned to support Standard, HE and EE power bands.
So, there you are, a full lineup of heavy hitters – it should be a great game. With this score card you’ll be able to tell who is at bat and who is on deck, so sit back and enjoy the game.
John Fruehe is the Director of Business Development for Server/Workstation products at AMD. His postings are his own opinions and may not represent AMD’s positions, strategies or opinions. Links to third party sites are provided for convenience and unless explicitly stated, AMD is not responsible for the contents of such linked sites and no endorsement is implied.
Efficiency, Flexibility, and VMworld 2009
Posted by Andy Parma in 10:19 pm
Efficiency and flexibility are two of the themes for VMworld 2009, so it seems like an appropriate time to discuss the efficiency and flexibility of AMD’s Direct Connect Architecture. As Tim Mueting and I discussed on YouTube , this architecture and AMD VirtualizationTM (AMD-VTM) technology enable servers using AMD OpteronTM 8400 Series processors to consolidate more virtual machines than servers using a competing architecture1. Direct Connect Architecture also lets AMD deliver socket-compatible processor solutions ranging from our highest performance AMD OpteronTM 8400 SE Series processors to our new energy efficient AMD OpteronTM 2400 EE Series processors.
When Quad-Core AMD OpteronTM processor Model 2384 was introduced less than a year ago, servers using this processor achieved a number of performance records (here and here) and a key reviewer concluded that “Right now, it is clear that the latest AMD Opteron is in the lead.” If you look at the performance of Six-Core AMD OpteronTM 2419 EE processor-based servers, you see that servers using this new low-power processor are outperforming servers using the Quad-Core AMD OpteronTM processor Model 2384. That’s pretty impressive.
And the energy savings from using low-power Six-Core AMD OpteronTM EE processors (compared to 75W ACP Six-Core AMD OpteronTM processors) are significant. When we replaced the 75W ACP Six-Core AMD OpteronTM processors in a ZT Systems server with 40W ACP Six-Core AMD OpteronTM EE processors, server power consumption at 100% load dropped by 124W (40%).
Surpassing the performance of Quad-Core AMD OpteronTM processor Model 2384-based servers using extremely energy efficient processors is quite a feat. Achieving higher performance, while consuming less power, is even more impressive. To top it off, the Six-Core AMD OpteronTM processor Model 2419 EE is being offered at the same price that the Quad-Core AMD OpteronTM processor Model 2384 was sold for last year2.
Higher performance. Lower server power consumption. Same introductory processor price. Wow!
As excited as I am about our current products, I can’t resist the temptation to mention the Six-Core AMD OpteronTM EE processors (codenamed “Lisbon”) that we’re planning to introduce next year. These six-core processors are planned to have a rated power consumption of less than 40W – that’s lower than the rated power consumption of most of today’s quad-core mobile processors. A processor that combines the registered memory and RAS (reliability, availability, and serviceability) features of a server processor with the power consumption of a mobile processor?
I think that I’ll be typing “Wow!” again next year.
Whether you think of “flexibility” as the ability to host more virtual machines using a Six-Core AMD OpteronTM 8400 Series processor-based server, or you view “efficiency” as the capability to achieve higher performance while consuming less server power using Six-Core AMD OpteronTM 2419 EE processors, it’s clear that servers using AMD’s Direct Connect Architecture are ideal for the next generation of computing.
To find out more about AMD VirtualizationTM (AMD-VTM) technology and AMD OpteronTM processors, visit us at booth 1408 at VMworld or visit www.amd.com/virtualization.
Andy Parma is a Product Marketing Manager for Server/Workstation products at AMD. His postings are his own opinions and may not represent AMD’s positions, strategies or opinions. Links to third party sites are provided for convenience and unless explicitly stated, AMD is not responsible for the contents of such linked sites and no endorsement is implied.
1Based on 30 tiles x 6 VMs for 48-core HP ProLiant DL785 G6 server, as tested using the VMmark benchmark (http://www.vmware.com/products/vmmark/results.html).
2Pricing for Quad-Core AMD OpteronTM processor Model 2384 reflects 1kU tray pricing on www.amd.com as of November 2008. Pricing for Six-Core AMD OpteronTM processor Model 2419 EE reflects 1kU tray pricing on www.amd.com as of August 2009.
How Low Can You Go?
Posted by John Fruehe in 10:19 pm
Back in April, when we introduced the Quad-Core AMD OpteronTM EE processors, Gordon Haff of Illuminata had this to say in his blog:
“Opteron EE is therefore not just your basic low-end-of-the-frequency-scale parts. Rather, they’re explicitly targeted for cloud computing and Web 2.0-in other words, the type of uses and customers who explicitly value power efficiency.”
Gordon points out that these processors are “not simply the fall-outs at the low end of the frequency range as lower power processors have often been historically.” We are specifically targeting low power consumption because we understand the power needs that customers have. We understand the environments and are targeting processors to meet those needs.
Those quad-core processors that were launched in April have an ACP of 40 watts, so if you do the very simple “watts per core” math (ACP / cores) you have ~10W per core. Of course that is not a scientific measurement (that would require much more complex testing because the simple math does not comprehend that there are components besides the cores in the processor).
Now, today, we introduce new Six-Core AMD Opteron EE processors that have the same 40W ACP. Again, the simple math says 40W ACP / 6 cores = ~6.67W per core. Does anyone remember the world before AMD introduced the first AMD Opteron processor?
Server state-of-the-art, B.O. (Before AMD Opteron), was Prestonia, a single core processor with configurations of up to 58W max TDP just one core! A mere six years later, the new Six-Core AMD Opteron EE processors have six times the number of cores and dramatically lower power per core.
We really have come a long way, with as much as 58 watts of power for a single core (2.0GHz), to today’s new standard of single digit ACP per core. With the shorter pipelines and better efficiency of today’s AMD Opteron processors, I am guessing that a single 2GHz Opteron core is going to be much more efficient than the legacy NetBurst cores in Prestonia, and the fact that the power is so much lower means that we are really heading in the right direction.
Next year, we plan to introduce the “San Marino” platform, featuring our “Lisbon” (C32) processor. We expect to have a platform that is specifically tuned to the needs of these very low power environments. It may not be a stretch to say that with next year’s optimized platforms, we may be able to provide even lower total power consumption than we see today with these AMD Opteron EE processor-based platforms.
Now, the question is “how low can you go?” Well, only the future will know. Well, that and our design engineers.
If you’re interested in a deeper dive on our Six-Core AMD Opteron EE processor, check out the presentation below.
John Fruehe is the Director of Business Development for Server/Workstation products at AMD. His postings are his own opinions and may not represent AMD’s positions, strategies or opinions. Links to third party sites are provided for convenience and unless explicitly stated, AMD is not responsible for the contents of such linked sites and no endorsement is implied.
Master of My Domain
Posted by John Fruehe in 4:08 pm
Recently I blogged about AMD OpteronTM processors in the world’s largest supercomputers. But what about the other end of the spectrum?
Like many of you, I have a second job as the network administrator for a small network – at home. The pay is non-existent but the working conditions are flexible. We all come home from our jobs, and as the de facto technical person in the house, it is my responsibility to keep it all running.
Over time I have managed to assign 16 different IP addresses on my home network. Consolidating print services on to a NAS allowed me to retire one print server this weekend, simplicity reigns supreme. When everything runs fine, it is a well oiled machine; but it generally always chooses the day I have just left for Asia to come tumbling down.
Here is what I grapple with when I get home at night:
The complexity of the network has grown over the years. The first server was NetWare 3.11, followed by Windows NT Server. Eventually it simply became Windows Vista on the server, again, in the need for simplicity and commonality with the other OS’s (and not needing an enterprise-class OS at home.)
What I have noticed about the network is that everything is purpose-driven. There isn’t anything that has just been added for the heck of it. I scrutinize IT purchases just like you, because they add complexity to my life.
Commonality is really important. The 2 NAS devices (one for data, one as a mirror backup) are identical models with the same drive models. Disaster recovery is simply changing the IP from the primary to the secondary.
I shoot for commonality on the motherboards so that when I have to update drivers, I can take care of all of those chores at one time. It’s funny that I talk to customers all the time who talk about the importance of commonality in their data centers, and even on a personal level, on a home network, it makes sense. They love the commonality of the platforms based on AMD OpteronTM processors and I can see why. Being able to count on the same driver to update different generations of AMD-based servers is a huge reduction in the amount of time spent managing the update process.
This past weekend I decided to build up another system and load Windows Home Server to see how the experience was. Deploying a new server, especially one with a new technology is always a challenge.
I am adding this server because I want to be able to allow my wife to have a universal file storage – with remote file editing (without having to deal with “upload/download.”) If I can figure out how to enable this functionality on the current system, then I would probably want to consolidate some of the functions with an HP MediaSmart Server because I really love that compact form factor.
Working with the WHS software presented an interesting challenge when it came to power consumption. The software is based on Windows Server 2003 but it was not very clear which drivers you need for power savings. I have the whole system around 45W in idle (where is sits most of the day), which is probably about $.10 a day in power (it consumes roughly a kilowatt hour and we pay ~$.10/KwH here in Austin). While that might not seem like a lot, consolidating down to the HP system would hopefully drop the consumption even more.
Walking around the house with a power meter, and doing some quick math, it looks like the network is drawing ~$10/month in power (based on that rough estimate of $.10/KwH). Sleep mode helps cut that number down a bit, but don’t let it fool you, even when sleeping, devices are pulling power.
The lessons I learned this weekend probably sound very similar to what you deal with:
1. Commonality is good
2. Consolidation reduces your management tasks
3. Software is never as easy as it appears
4. Power efficiency is very critical
Hopefully, by spending some time with WHS I can build the application that I need and then can consolidate some of the functions down to 1 box. Ah, the life of a network administrator.
John Fruehe is the Director of Business Development for Server/Workstation products at AMD. His postings are his own opinions and may not represent AMD’s positions, strategies or opinions. Links to third party sites are provided for convenience and unless explicitly stated, AMD is not responsible for the contents of such linked sites and no endorsement is implied.
Cash for Clunkers
Posted by John Fruehe in 11:34 am
One of the hardest parts about managing a data center is not necessarily the task of “keeping the lights on”, but instead, it’s “changing the bulbs.”
This week I am hearing a lot about the “cash for clunkers” program here in the US, where older cars can be traded in for newer more fuel efficient vehicles, with the government kicking in some rebate dollars to make it all happen if you are buying a more fuel efficient vehicle. I’m not going to even pretend to think about this from a political sense (there is a reason I chose product marketing and not politics), but, from a business sense, this sort of activity is happening in data centers all over the world. Even without the government assistance.
Are people changing out servers just to drive more efficiency? Hardly. But in a tough economy, they think about optimization, they think about efficiency and they think about virtualization when they have to make a change.
As customers look to pull the plug on older servers that are past their useful life -- either through warranty or lease expiration, or potentially due to a hardware failure that is too costly to repair -- they are looking to more power efficient servers as well as virtualization to solve their future computing challenges.
The migration to a new platform is costly, time consuming and full of interesting challenges. One way to overcome some of these challenges is to use virtualization to encapsulate the server, creating a virtual machine. With the system image (software) now physically abstracted from the hardware, it can be moved anywhere around the data center, or around the world.
This new level of abstraction is great, but one of the real challenges is that a virtual machine has to take the “lowest common denominator” in order to move between two systems. The flexibility to move virtual machines around, at will, is dependent on the underlying hardware.
AMD has designed an amazing consistency in our architectural design, allowing a server to have backwards and forwards consistency in moving virtual machines -- a feat that is much more challenging on other’s platforms unless you really scale the virtual machine back to a very simple configuration.
Here, we actually demonstrate taking a virtual machine from an old dual core, to a newer six-core platform, and then move that virtual machine to our future AMD OpteronTM 6000 processor-based platform.
As you look to optimize your data center, utilizing AMD Opteron processors can help keep your environment humming along, with the ability to load balance and drive incredible efficiency through virtualization. The fact that we can deliver such flexibility across the board shows that AMD understands the challenges that you face, and we tailor our products to meet those needs.
Not only can we help you drive better efficiency by moving from older AMD platforms to AMD’s newer, more power efficient platforms, but our common socket strategy allows many of the AMD OpteronTM based-platforms to be easily upgraded from dual core to quad core, or even six core, virtually without having to change out any other hardware. It’s an instant upgrade in performance, all within the same general power and thermal ranges.
These are just two more ways that AMD is helping you drive more efficiency in the data center. Maybe we can’t hand back “cash for clunkers”, but we can help get you back on the road to recovery, and that’s where we all want to be these days.
John Fruehe is the Director of Business Development for Server/Workstation products at AMD. His postings are his own opinions and may not represent AMD’s positions, strategies or opinions. Links to third party sites are provided for convenience and unless explicitly stated, AMD is not responsible for the contents of such linked sites and no endorsement is implied.
Energy Efficient Four-Socket Servers – REALLY?
Posted by Andy Parma in 9:27 pm
As VMworld 2009 approaches, I’ve been thinking about virtualization and energy efficiency and how they apply to four-socket servers. I know what you’re thinking – isn’t an energy efficient four-socket server an oxymoron like a hybrid SUV?
First, let’s take a closer look at how IT managers are actually using four-socket servers. It’s very rare for four-socket servers to be used strictly as a way to increase compute density. Appro and SGI have twin servers and half-depth servers to serve this purpose. In most cases, four-socket servers are used for applications that require a single server to have access to large amounts of memory. These applications typically consist of database, virtualization/consolidation, and some high performance computing applications. With the ability to support 32 memory DIMMs per server, optimized virtualization features such as AMD VirtualizationTM (AMD-VTM) technology, and unique energy efficiency features (AMD-P), Six-Core AMD OpteronTM 8400 series processors are ideal for these applications.
When comparing the feature sets of Six-Core AMD OpteronTM 8400 series processors and Intel Xeon 7400 series (”Dunnington”) processors, the feature set consistency of the AMD OpteronTM 8400 series processors is in stark contrast to the Intel Xeon 7400 series offerings. Four items about the Xeon 7400 series stand out:
- The feature set differs dramatically between each of the seven models in the Xeon 7400 series
- No Xeon 7400 series processor includes RVI or Tagged TLB virtualization features
- The low-power Xeon L7445 and L7455 do NOT include the same energy efficiency features as the highest performance Xeon X7460
- All Xeon 7400 series processors use Fully Buffered DIMM (FB-DIMM) memory
The feature set consistency of the AMD OpteronTM 8400 series processors gives customers a more predictable approach to data center planning and helps capacity planning, software image development, and validation efforts.
What about the performance and value of servers using these processors? If servers using Intel processors are clearly superior, then the differing feature sets won’t matter, right?
While the 46% VMmark performance advantage that a four-socket AMD OpteronTM 8400 series processor-based server has over the top-performing four-socket Intel Xeon 7400 series processor-based server is impressive, I think that the performance of energy efficient Six-Core AMD OpteronTM 8425 HE processor-based servers is even more impressive. This comparison shows that, depending on benchmark, a server using low-power Six-Core AMD OpteronTM processors can provide nearly double the server performance of a server using low-power Hex-Core Intel Xeon L7455 processors at nearly half the processor price. Servers using low-power Six-Core AMD OpteronTM 8425 HE processors even significantly outperform servers using the highest performance, highest power Hex Core Intel Xeon X7460 processors, again at nearly half the processor price.
When combined, all of these advantages (superior performance, energy efficiency, virtualization features, and pricing) make the Six-Core AMD Opteron 8400 series processors the ideal solution for four-socket servers.
Now, let’s go back to my original analogy about the hybrid SUV. In theory, the reason that people buy an SUV is because they need to move a lot of stuff, whether that’s people, groceries, or furniture. It’s the same way with four-socket servers – people buy a four-socket server because they need lots of memory to process lots of stuff, whether that’s database applications, virtualization/consolidation, or processing large datasets in high performance computing applications. Why shouldn’t you able to process lots of “stuff” and get energy efficiency too?
Andy Parma is a Product Marketing Manager for Server/Workstation products at AMD. His postings are his own opinions and may not represent AMD’s positions, strategies or opinions. Links to third party sites are provided for convenience and unless explicitly stated, AMD is not responsible for the contents of such linked sites and no endorsement is implied.
Hot, Hot, Hot
Posted by John Fruehe in 8:07 am
I managed to dodge the Austin weather for the past 2 weeks, and boy did I pick a good time to be out of town. The temperatures were over 100 degrees Fahrenheit almost every day that I was gone. For those of you on the metric system, I believe that 100 degrees roughly translates into the temperature of the surface of the sun. A cold beer can’t cool you off if it immediately vaporizes.
Truthfully, 100F translates to ~37C or so, which is very close to the 35C spec that is so common in the computer industry. We spec our processors to run at a maximum temperature of 35C, which means that there is a lot of heat being generated inside those boxes. Blades, twins and 1U servers aren’t helping the situation because their form factors are driving greater heat density.
Many are starting to investigate things like liquid cooling racks or other proprietary schemes to drive lower temperatures. The reason for this is that the chillers used to keep the server room at a lower temperature are very expensive.
Everyone is trying to avoid “the million dollar server.” That is the last server that breaks the camel’s back and forces you to upgrade the AC units. Power is generally a continuum – add another server and your cost go up marginally (as long as you don’t need to pull another circuit to the rack.)
But with cooling, it is not quite as linear. There is both a variable cost for each server, as well as a large step-function jump when you have to add new equipment.
The two big challenges that are hitting the data center when it comes to cooling the data center with proprietary methods are cost and risk.
Cost is always a driver with any type of proprietary system. It will be a driver until the system hits critical mass. Even then, if there are complicated manufacturing methods, even widespread market adoption might not drive the cost curve down fast enough. If some proprietary intellectual property is included in the system, expect royalties to continue to take their bite.
The other factor is risk, but it is not risk in the classic sense. It is the risk of making the wrong decision. If you choose the new technology too early, you might miss some innovation that helps drive the cost down somewhat. And if you guess wrong, well, we don’t need to say how that story ends, we’ve all see the movie.
Cost and risk are driving a lot of customers to investigate more non-standard cooling technologies, but that process is not happening fast enough for some companies. Air cooling is not only the traditional method, but even with new technology, it remains the predominant method.
I’m very interested in your cooling strategies. Post a comment with what you are doing to combat “global warming” in your own environment.
John Fruehe is the Director of Business Development for Server/Workstation products at AMD. His postings are his own opinions and may not represent AMD’s positions, strategies or opinions. Links to third party sites are provided for convenience and unless explicitly stated, AMD is not responsible for the contents of such linked sites and no endorsement is implied.
The Longest Day
Posted by John Fruehe in 8:02 am
My iPod is playing “Everyday is Like Sunday” by Morrissey. I am currently somewhere between Russia and Alaska on a plane home from 2 weeks of vacation in Japan. Woke up this morning about 6, and at this rate, I won’t sleep for about 36 hours. Every day may be like Sunday, but this Sunday is like part of Monday too.
With lots of time to kill, I have been thinking about designing for power efficiency. The latest generation of server processors are definitely designed to focus on the energy efficiency needs of the data center, but how do companies go about doing this? Is it all the same?
I took a look at a recent article at Tech Report.com that highlights, amongst other things, some power efficiency calculations they performed on both our latest six-core AMD OpteronTM processor and our competitor’s latest release. These are Tech Report’s test results – I haven’t had the time or the opportunity to recreate the tests in my own lab, but I trust their methodology, and believe these results are reasonable and accurate.
One thing to note is that in all of the numbers below, we are looking at our six cores versus our competitor’s four cores, so from a core count, when it comes to power, we have a “hand tied behind our backs” with 50% more cores drawing power. Plus, they configured the competitor’s system with 6 DIMMs vs. our 8 DIMMs.
One of the most striking points, according to Tech Report, is that our six core platform and their four core platform both idle at approximately the same level – our 154W to their 153W. 2 extra cores for only 1 watt, what a great deal.
However, the platforms diverge when you add the slightest load. With only a 2% load, we moved from 154W to 158W, not a large delta at all. Our competitor, however, jumped from 153W to 186W. That is a pretty dramatic jump in power consumption. If you consider that throughout the course of a day (even a long one like today), the processors will jump back and forth between idle and light loads quite often. Having a low idle power on a server is like having low gas consumption on a car in idle. Eventually you are going to want to get somewhere; people don’t buy cars to sit on their driveways.
Based on my interactions with customers, I’ve found that most IT managers believe that their average CPU utilization is ~15-20%. This means the processors are probably spending a lot of time over the course of the day going in and out of idle. So these small jumps in utilization could mean big jumps in power consumption, all depending on whether you make the right processor choice of course.
Peak power is another story. Our peak power in the Tech Report article was 278W, quite respectable for a system with 12 total cores. Our competitor’s Nehalem-based system pulled 330W at the wall when in peak load. 52 watts is a lot of power, and if you are running virtualization or an HPC cluster, where high utilization is the profile, then you are really limiting yourself by not choosing these AMD Opteron processors.
Data center floor space is expensive, probably the most expensive area in all of your facilities. Those extra 52 watts add up quickly. With a 1200KVA power budget per rack you could fit 42 1U AMD Opteron processor-based platforms, or only 36 competing 1U servers. That means potentially wasting almost 15% of the space in each rack. Meaning more racks, and more valuable floor space consumed.
It’s not enough to say that you design processors for energy efficiency. You have to optimize across the whole range of utilization levels. Only then can you be sure that whether you are dealing with a low level of utilization or a highly utilized server that you are getting the most for your money. No matter how long your day is.
John Fruehe is the Director of Business Development for Server/Workstation products at AMD. His postings are his own opinions and may not represent AMD’s positions, strategies or opinions. Links to third party sites are provided for convenience and unless explicitly stated, AMD is not responsible for the contents of such linked sites and no endorsement is implied.
Still a Distant (Server) Memory
Posted by John Fruehe in 4:40 pm
While the prices for DDR-3 memory on the desktop side are finally starting to fall into reasonable price levels, the server world is still not there yet when it comes to the transition to DDR-3 memory.
Just this past week, we saw that there is still a lot of volatility in the memory market, and memory is a commodity that has a huge impact on the price of a server.
For instance, taking a quick look at Crucial (http://www.crucial.com) shows that in comparing the prices of 2GB and 4GB DIMMs, there is still a pretty substantial premium on DDR-3.
| 2GB | 4GB | |
| DDR-2 | $54 | $110 |
| DDR-3 | $80 | $160 |
| Premium | 48% | 45% |
(pricing for single DIMM SKUs, as of 7/09/09 at http://www.crucial.com)
For customers buying servers, memory is often the most expensive part of the system. With four and eight DIMMs per socket, you can easily see the price differential rocket upward as the system requirements grow.
Where desktops generally use 2GB of memory or 4GB for more power users (Windows Vista 32-bit only sees about 3GB), in the server space we are talking about 16GB as the minimal point of entry and 32GB as a common configuration. Couple this with the more expensive registered, ECC memory that servers demand, and you can start to feel the pain in your budget.
By utilizing the energy-efficient DDR-2 memory, AMD OpteronTM processor-based platforms can take advantage of the economies of scale as they grow the memory capacity on their servers to meet the needs of the business.
Next year, when the DDR-3 memory prices for server memory have likely started to drop out of the stratosphere and down into a reasonable range, we’re planning for our Q1 2010 platform, codenamed “Maranello”, to go to market with support for DDR-3 memory. And, more importantly, low voltage DDR-3 memory. The low voltage offerings, along with these anticipated lower prices, should make DDR-3 the smart choice for server customers.
So, while the cost of migrating a desktop from DDR-2 to DDR-3 is becoming reasonable for many customers, we still have a long way to go before DDR-3 has similar economic advantages for the server market.
John Fruehe is the Director of Business Development for Server/Workstation products at AMD. His postings are his own opinions and may not represent AMD’s positions, strategies or opinions. Links to third party sites are provided for convenience and unless explicitly stated, AMD is not responsible for the contents of such linked sites and no endorsement is implied.