Posts tagged with AMD Opteron
“Istanbul” – The HPCWire Reader’s Choice
Posted by John Fruehe in 3:39 pm
Loading ...This week, at the SuperComputing 2009 event, AMD received the HPCWire Reader’s Choice Award for our Six Core AMD Opteron™ Processor (the one you know as “Istanbul.”)
When you consider that two of the largest High Performance Computing systems in the world (Oak Ridge’s “Jaguar” and the University of Tennessee’s “Kraken”) were recently and simply upgraded from their quad-core processors to six core processors and got a sizeable boost in performance, to put it mildly, it probably wouldn’t be a shock to hear that we were recognized by the HPC community.
What makes this award great is not that we are being recognized, but the fact that this is a reader’s choice award. When we get a press award, it’s great; but when the readers decide that we deserve recognition, that’s another story.
In the HPC world, flash and advertising won’t cut it. These are the men and women who slave over huge systems, trying to squeeze every ounce of performance out of them. They appreciate performance, but they also really appreciate consistency and simplicity – those help them focus on the task at hand – designing alternative energy sources, predicting weather and climate change, developing the next breakthrough drug, or analyzing crash data to make our transportation systems safer.
Knowing that these folks find our product outstanding means the world to AMD, because we build the AMD Opteron processor for them.
So, Margaret, be sure to bring that award home, we’ll try to clear some room in the trophy case. While we are in there, maybe we need to expand it a bit, because with 12-core processors with 4 channels of memory on the horizon, we might need some extra space.
John Fruehe is the Director of Product Marketing 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.
AMD Opteron Processor – Delivering True Value, Not Just Benchmarks
Posted by John Fruehe in 4:39 pm
(10 votes, average: 4.80 out of 5)I left the office yesterday at 5:00, headed to the trail for my weekly ride. Because of the recent rain, we shifted to City Park, one of the most technical mountain bike trails in Austin. If you just looked at us and our bikes in the parking lot, you’d notice that the older guy with the borrowed bike (because I finally broke my frame after 4 years) would probably not fare well against better riders, younger riders or newer bikes.
But looks can be deceiving. While I didn’t have the flashy skills or the newest ride, I did have a dedication and a lack of technical problems that actually allowed me to finish the ride ahead of others. Sometimes the outcome isn’t always as obvious.
Too many people get caught up in the benchmarks and think that is the end of the story. Benchmarks can be important as an indicator, but they don’t tell the whole story. As a matter of fact, they really only tell a portion of PART of the story. In my opinion, making a decision based on benchmark alone is like buying a car because you like the color. That is great if all you care about is a red car, but so few of us really think that way in the real world.
One of the things that I have been telling customers is that they need to step out of the “socket” world and start thinking more about the work that needs to get done when assessing the platforms. Sometimes the answer is not obvious.
The non-obvious thread for today started this morning when I went to check out the latest review on AnandTech.com of the AMD Opteron processor vs. our competitor’s processor (which my colleague Margaret Lewis also discusses here). We sent them a review system recently and during their testing, they came up with some interesting comments.
Comparing a 2 socket Intel server to a 4 socket AMD server, you start to see that if you step outside of the raw performance area, there are plenty of other places to be looking, like threading and efficiency, not just clock speed. Here is what AnandTech had to say when comparing our six-core 2435 to a higher clock speed quad core:
“If your application scales well, two 2.6GHz Opteron 2435 will offer 15% better (and sometimes more) performance than a 2.9GHz Opteron 2389 with the same power consumption.”
More performance, at the same power envelope and lower clock speed. Not always the most obvious answer, but clearly a great choice for threaded applications. (And don’t think that applications will be less threaded in the future.) Memory is another area less than obvious. Here is what AnandTech had to say:
“Using relatively ‘old’ technology such as DDR2, the hex-core Opteron based servers are very affordable, especially if you compare them with similar Xeon servers.”
Yeah, I was the old guy on the trail, but that didn’t mean I was the last. I can think of a couple of younger guys finishing after me – so sometimes “newer” doesn’t me “better”. Again, not the obvious answer, but until the prices of DDR-3 come in line with DDR-2, this is our secret weapon. And in a tough economy, who doesn’t want to be smart with their IT budgets.
We have tremendous consistency, but people expect that. And they expect that all processors in the family will have the same features. That is obvious, right? Here’s how AnandTech saw it.
“If you chose the Xeon platform, you should be aware of the fact that Intel’s low end is much less interesting: the best Xeon 55xx CPUs have a clock speed between 2.26 and 2.93GHz. The low end models, the 5504 and 5506 are pretty crippled, with no Hyper-Threading, no Turbo Boost, and only half as much L3 cache (4MB). These crippled CPUs can keep up with the quad-core Opterons at about 2.5GHz, but they are the worst Xeons when you look at idle and full load power. The performance per Watt of the Xeon EE550x is pretty bad compared to the more expensive parts.”
You’ve heard me talk a lot about “no compromises” in our products, and this is what we are talking about. Just because you can’t afford to buy the most expensive processor in the stack shouldn’t mean that you have to compromise so much. Cutting the cache in half? That is not obvious to the typical customer.
When you step back and take a look at all of this in full, you see what was obvious before – 2P servers are the best value for most workloads – is becoming somewhat less obvious. And as we get into 2010, the value that AMD will bring to market will blur the lines even more. It’s time to stop looking at the world of sockets and start looking at the workload, the power consumption and the cost. Only then, does everything become obvious.
Otherwise you’ll just be like that young guy tonight, standing on the trail as the old guy blew past him. Sometimes the obvious answer isn’t always the only answer.
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.
AMD Opteron Processors + AMD Chipsets = the Whole is Greater than the Sum of the Parts
Posted by John Fruehe in 10:00 pm
It’s a cloudy afternoon here in London, which is great because yesterday I was soaked while trying to see customers. Clouds trump rain any day in my book. A little known fact (outside of this country) is that the city of London is actually 2 cities, the City of London and the City of Westminster.
For London, “the the whole is greater than the sum of the parts” is very true, especially when morning hits and the tube traffic swells the population of the city.
For AMD, our new server platform, the combination of AMD OpteronTM processors and the new AMD chipsets show that the whole is larger than the sum of the individual parts as well.
AMD introduced 3 new chipsets this week, the SR5690, SR5670 and SR5650. And just like the different neighborhoods here in London, each has their own unique characteristics, yet each shares a lot in common with each other. They are all built around the same silicon, with the same socket, same programming interface and same drivers. Sound familiar? But they have different numbers of PCI Express® lanes, different power consumption and different pricing, allowing our partners to customize their platforms around these different offerings.
And just like the swelling tube traffic this morning, these new chipsets have 2 features to help the system handle large amounts of traffic. They feature HyperTransportTM 3 Technology allowing for high speed interconnection between the processors and the I/O devices. In addition, once you get to those devices, you’ll find that our support for the new PCI Express® Gen 2 technology allows for high speed peripherals, like Quad Data Rate Infiniband, to be included in the servers. If you are in an HPC environment or other clustering environment, having a high performance and low latency interconnect can really swing performance in the right direction.
And if you are into virtualization, these chipset provide the hardware support for I/O Virtualization, an emerging virtualization technology for increased performance and security. If you want to get a peak at these chipsets in action – check out the AMD Virtualization blog “AMD Showcases Virtualization Innovation at VMworld Conference” for recap of the demos we did at VMworld.
These new platforms will also include a new platform specification from AMD called “Kroner.” Kroner is a nod to those in the cloud computing space who are looking for ultra-high levels of power efficiency. Optimized around best practices in power efficiency, the first Kroner platform, the Tyan S8208 is a single 1U system with two “twin” Kroner boards, able to deliver that impressive performance while only “sipping” the power at the wall. After hearing a customer tell me that they have a 4KW limit for their racks and can only put 10 servers in each, a platform like Kroner can go a long way towards helping optimize space in the data center.
With these new introductions, AMD is moving much more towards a platform company and not just a supplier of microprocessors – this was a key driver for the acquisition of ATI. We’ve seen the fruits of that on the client side, and now we are seeing it on the server side. Customers live the idea of an integrated platform with processor and chipset from the same company because that allows for much tighter integration of technology. Just like London and Westminster.
The whole becomes greater than the sum of the parts.
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.
Under the Influence
Posted by John Fruehe in 12:31 pm
NetworkWorld recently released its list of the 11 Most Influential Microprocessors of All Time. I have to agree that the AMD OpteronTM 240 Series processor is one of the most influential processors because it is the singular reason that I left my last job of almost ten years – I came to AMD specifically to work on the AMD Opteron processor because I realized the game-changing potential of this technology.
The list is a relative “who’s who” of the microprocessor world. Every processor clearly had a huge impact on the market, and it is an honor to be included in the list. Most importantly, we are the only processor in the list that was released after 1993. AND we are the only one that is still in the market today in any reasonable volume (somehow, somewhere, somebody is probably utilizing the others because they were such strong designs).
Back in 2002, a year before the launch of the AMD Opteron processor AMD and Microsoft announced a cooperative effort to develop 64-bit version of windows for AMD 64-bit processors. We were the first mainstream processor to enable 64-bit, and our shared vision was that 64-bit computing would drive the data center of the future. I guess you could say that we helped influence the move to 64-bit.
In 2003, there was no such thing as a “low powered server processor.” Enterprise-class processors were big, hot and power hungry. Now look at where we are today. The new AMD Opteron EE processor delivers six cores in a single processor, with a 40W ACP. Power is one of the biggest concerns in the data center today, and we influenced the market with the introduction of power efficiency for servers back before people realized how important power would be.
What about price? If you can remember the pre-AMD processor Opteron days (or the “dark ages” as I like to call them), there was little competition in x86 server processors. Price/performance was a very limited concept, your price/performance came at the platform level, not really at the processor level. Prices were significantly higher than they are today. AMD influence? Check.
Remember the world of single core computing? It may be hazy for most of you because for the past 4 years, almost every server processor sold was probably a multi-core. And who had the first multi-core X86 processor? Yep, the influencer.
So what does the server processor of 2009 look like? Well, let’s take a look at the profile:
- Multi-core
- 64-bit x86
- Integrated memory controller
- Low power options
- High performance per watt
- High price performance
- Hardware-based virtualization capabilities
When you put the list together, it is clear that AMD has had a significant impact on the server processor market. Regardless of which processors you are deploying, what you enjoy today in your X86 data centers are either AMD Opteron processors, or processors that have been heavily influenced by our technology decisions. Some would say that “imitation is the sincerest form of flattery.” I like to call it influence.
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.
Rise of the Big Machines
Posted by John Fruehe in 10:00 am
I remember the first rack-based servers at Compaq. The world was ruled by tower servers. We wanted to introduce racks to the x86 world, but we knew that customers were risk averse. We had this great marketing campaign about “the same servers you know and love, now in rack format.” Brilliant, right?
When we went out to test market the idea with focus groups, they all saw the pictures of a full rack and said “No, these are like mainframes. They are much more powerful than servers – they are supercomputers.” And we actually tried to convince them that these weren’t supercomputers, they were the same servers they were used to buying – we’re talking x86, after all.
Then we realized the smart person should play up the idea of being more powerful, because that is what the customer expects. “x86 Supercomputers” -a phrase that we all joked about back in 1994. Fast forward 15 years and what we see now are x86 supercomputers – and the AMD OpteronTM processor has had a large hand in leading the change.
But no company has explored and delivered on the idea of x86 supercomputing more so than Cray, the company that is synonymous with supercomputers.
If you take a look at recent IDC data on the $3M+ segment of the supercomputer market, what you see is that Cray has the leadership position, with 35% of the market. Their penetration into defense, academia and research is legendary, but Cray is increasingly present in the corporate world as well. With the XT3, XT4 and now XT5 systems, Cray has been able to build highly scalable, highly parallel supercomputers by innovating around industry-standard components like AMD Opteron processors.
Through their high speed bus architecture and meticulous engineering, they have managed to achieve key wins on the worldwide Top 500 supercomputer list year after year., with one third of the Top 15, all based on Cray systems featuring AMD Technology:
- #2 Oak Ridge National Laboratory
- #6 National Institute for Computational Sciences/University of Tennessee
- #11 NERSC/LBNL
- #12 Oak Ridge National Laboratory
- #13 NNSA/Sandia National Laboratories
Of course, the folks at Cray will tell you (and they’d be right) that the most important aspect of these systems – and all the others they have deployed through the years around the globe – is the real-world work that’s achieved when they’re in action.
This week Cray announced two major pieces of news. First, the #2 supercomputer in the world, the “Jaguar” system at ORNL will be upgraded to a 2 Petaflop system by replacing the existing quad-core processors with AMD’s latest six-core processor. A total of around 225,000 processor cores if you are doing the math. In addition, NERSC, the holder of the current #11 system, will also be adding a new Cray XT5 system with 1 Petaflop of performance.
Congratulations to the team at Cray, their execution in the world of supercomputing is second to none.
When I look back on those Compaq customers that were telling me that the shiny new racks of x86 servers were really supercomputers, I have to admit that they were 100% right. They just didn’t know how visionary they were back then.
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.
Playing 20 Questions (Part 2)
Posted by John Fruehe in 12:18 pm
Last week, I kicked off the first round of “20 Questions” by answering questions around our upcoming server platforms. This week I’ll focus on questions I’ve received related to virtualization and cloud computing.
When AMD pushed further into “Cloud\VM” computing there was talk of enabling GPUs as APUs in “future” sockets. Where is that initiative and where is it going?
Accelerated Processing Units (APUs) are essential to AMD’s product strategy and roadmaps. We still continue to see the potential for GPGPUs handling certain types of server workloads. However, this technology faces some challenges before becoming “the next big thing” for cloud clusters. The first is the programming model. One way we plan to address this is through the work that our ATI Stream team is doing around OpenCL, (ATI Stream technology is key in merging the GPU and CPU to form an APU)a framework for writing applications that execute across heterogeneous platforms consisting of modern multi-core CPUs, GPUs, and other processors
The second challenge is the power and cooling aspect of the solution. Customers are moving towards lower power solutions, especially in cloud, and CPUs are now approaching levels of performance/watt that still make them the leading choice for mainstream applications. The final area is the prioritization aspect. GPGPU is an emerging technology for servers. Integration into client processors has a much greater benefit for both customers and AMD so you will see much more emphasis on client integration first.
How will AMD address Green Computing?
We plan to continue to deliver low power processors, such as our HE and EE power bands. A new planned platform definition for Socket F, codenamed “Kroner” is focused as a “best practice for power management and design.” We expect to see a future “Kroner” follow-on platform that continues that thread. The C32 processors are really targeted at customers who want power efficiency, but we also plan low power options for the G34 processors as well. We are also planning enhancement to our AMD-P feature set with new technologies that are expected to help increase the power efficiency at the processor level, the platform level, and even at the data center level.
AMD has several initiatives targeting “cloud computing” and virtualization. Doesn’t this reduce the market for AMD products, since customers would buy only one server where they would have bought several?
Back when we are all in a single core world, everyone expected that dual-core would reduce the number of servers. It didn’t happen. Neither did the transition to quad-core. As virtualization and cloud computing help make IT more efficient, this can free up dollars for more investment in other areas. In any IT shop you’re likely to find that for every project that they are working on, there are dozens of other that they can’t fit into their budget. Add to this fact that world Internet usage is probably less than 25% today. As use of the Internet as a vehicle to deliver applications, data, and services continues to increase throughout the world, there is an obvious continued need for servers to power web and cloud clusters.
While AMD has made progress in the area of energy efficiency (e.g. AMD OpteronTM EE processors, Cool’n'QuietTM technology), does AMD feel there is more room for improvement in this area?
There is always room for more improvement. Power efficiency will continue to be an important focus for us.
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.
Not Lost in Translation
Posted by John Fruehe in 8:03 am
Spend any time in a foreign country and you get a new respect for the bilingual folks in the world. Especially those that can seamlessly move between two very different languages. After 3 years of high school German, I can stumble my way through German as long as the conversation is simple (especially if I am talking to children). I’ve made my way through Spain, France and Italy with the help of dictionaries and patient people.
But even with a dictionary, I fell all over myself in Japan. Having dinner with a friend last week who has been in Japan for several years, it was amazing to see him move between a conversation with me in English and Japanese conversations with the waiters without even hesitating. It was second nature to him. Seamlessly moving between the languages was far more efficient than trying to look up a sentence in the dictionary one word at a time.
That drove home the concept behind our Rapid Virtualization Indexing (RVI) feature that is integrated into all AMD OpteronTM four and six core processors.
Much of the overhead of virtualization is from having to translate memory requests between the individual virtual machine and the physical hardware. In the old world, these translations were handled in the software – an infinitely slower process.
Today, that translation is done in silicon, speeding up the process greatly. Increasing the speed of memory translation is important because this is one of the biggest bottlenecks for virtualized servers. And as the memory footprint of the average server increases, this becomes even more critical.
Virtualization is one of those IT directions that is not a trendy “flavor of the week.” Virtualization was how we managed computing resources back in the “big iron age” when mainframes, costing millions of dollars were subdivided to provide virtual computing resources. Every department couldn’t afford their own mainframe, so there wasn’t a question about the value of virtualization. As servers became far less expensive, throwing hardware at the problem was a quick fix. Until we all started running out of space.
With the clear trend back to virtualization, there are hardware implementations of virtualization that can be done to help improve performance and processing efficiency. RVI is one of them, and in the second half of this year, you should expect to see new systems based on our next generation server platform. This new AMD platform is designed for “Shanghai” or “Istanbul” processors coupled with the new AMD SR56×0/SP5100 chipset.
This chipset will feature IOMMU, which is another hardware-based virtualization feature. IOMMU allows you to virtualize I/O traffic, enabling greater performance, scalability and efficiency as well as increased security by providing independent virtual I/O channels.
Through IOMMU, again, the I/O communications are translated at the hardware level vs. at the software level, providing greater performance and efficiency.
So as I stumble around like an old server, trying to manually translate the menu and Ramen Jinro, my friend will be able to do it on the fly without even thinking about it. Nothing is lost in translation and it happens so much faster when it is done at the hardware level. Progress is great on this front, and one day I hope to catch up with the rest of the world. Hopefully my noodles won’t be cold by then.
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.