Low Power Server CPU Redux: Quad-Core Comes to Play
by Jason Clark & Ross Whitehead on September 13, 2007 6:05 AM EST- Posted in
- IT Computing
A couple months ago, we took a look at the low voltage (LV) server CPU market. At the time, we focused on four-way solutions using two dual-core processors, since those represent the largest slice of the server pie. Our conclusion was that while the power savings brought about by using low voltage CPUs were real, processor choice was only one part of the equation. AMD came out ahead overall in performance/watt, not because they were faster or because their CPUs used less power, but rather because their platform as a whole offered competitive performance while using less power.
We discussed previously exactly what's involved in a low voltage part, but of course the picture is far bigger than just talking about power requirements. Take for example Intel's low-voltage Woodcrest parts; they are rated at 40W compared to the regular Woodcrest parts that are rated at 80W. The price premium for upgrading to a low-voltage part varies; in the case of AMD it's typically anywhere from $100 to $300 per CPU, while on the Intel side some low-voltage parts cost more, the same, or even less than the regular parts (i.e., the Xeon 5140 currently sells for about $450 while the low voltage Xeon 5148 only costs $400). Regardless of price, it's difficult to justify low-voltage processors in terms of power bill savings.
An extra 40W of power in a device running 24/7 for an entire year works out to around $35 per year, so at the low-end of the equation you would need a minimum of three years to recoup the investment (at which point it's probably time to upgrade the server). Other factors are usually the driving consideration.
Saving 40W per CPU socket may not save you money directly in terms of power bills, but generally speaking these chips are going into servers that sit in a datacenter. Air conditioning for the datacenter typically has costs directly related to the amount of power being consumed, so every 40W of power you can save could end up saving another 20W-40W of power in air conditioning requirements. That's still not even the primary concern for a lot of companies, though.
Datacenters often run dozens or even hundreds of servers within a single large room, and the real problem is making sure that there's enough power available to run all of the equipment. The cost of building a datacenter is anything but cheap, and if you can pack more processing power into the same amount of space, that is where low-voltage parts can really become useful. Blade servers were specifically created to address this requirement, and if you can reduce the total power use of the servers by 20% that means some companies could choose to run 20% more servers.
Of course, that doesn't mean that every company out there is interested in running a datacenter with hundreds of computers, so individually businesses need to look at what sort of server setup will best fit their needs. After determining that, then they need to look at low-voltage CPUs and decide whether or not they would actually be helpful. Assuming low-voltage parts are desired, the good news is that it's extremely easy to get them in most modern servers. Dell, HP, and other large server vendors usually include low-voltage parts as an easy upgrade for a small price premium. And that brings us to our low-voltage CPU update.
Intel Quad G-Stepping
Intel doesn't seem to sit still these days, pushing the power and performance envelope further and further. Recently, Intel announced two new G-stepping quad-core parts. The new parts run at the extreme ends of the power consumption spectrum. The first is a 2.0GHz 1333FSB part that runs at 50W while the second is a 3.0GHz 1333FSB part that runs at 120W. There are two main changes to the G-stepping parts, the first of which is power consumption: G-stepping introduces optimizations for idle state power. The second change involves enhancements to the Virtualization Extensions (VT), which mainly improve interrupt handling in the virtualization of Microsoft Windows 32-bit operating systems.
Of course, we would be remiss if we didn't mention AMD's recently launched Barcelona processor here. AMD expects their new quad-core processor to run within the same power envelope as the previous dual-core Opterons, which means twice as many CPU cores potentially without increasing power requirements, resulting in a potential doubling of performance/watt on the socket level. Low-voltage (HE) Barcelona parts will still be available, but even the regular chips include many new enhancements to help with power requirements. We are doing our best to get some additional Barcelona servers in-house in order to test this aspect of the performance/power equation and we hope to follow up in the near future.
One final item worth mentioning is that Intel's 45nm Harpertown refresh of Clovertown is due out in the very near future, which is one more item we can to look forward to testing. Unlike the desktop world, however, acquiring and testing server products often requires a lot more time and effort. Even with the appropriate hardware, the sort of benchmarks we run on servers can often take many hours just to complete a single test, and there are many parameters that can be tuned to improve performance. Since there aren't a lot of early adopters in the server market, though, we should be able to provide you with results before any of the IT departments out there are ready to upgrade. Now let's get on to the testing.
We discussed previously exactly what's involved in a low voltage part, but of course the picture is far bigger than just talking about power requirements. Take for example Intel's low-voltage Woodcrest parts; they are rated at 40W compared to the regular Woodcrest parts that are rated at 80W. The price premium for upgrading to a low-voltage part varies; in the case of AMD it's typically anywhere from $100 to $300 per CPU, while on the Intel side some low-voltage parts cost more, the same, or even less than the regular parts (i.e., the Xeon 5140 currently sells for about $450 while the low voltage Xeon 5148 only costs $400). Regardless of price, it's difficult to justify low-voltage processors in terms of power bill savings.
An extra 40W of power in a device running 24/7 for an entire year works out to around $35 per year, so at the low-end of the equation you would need a minimum of three years to recoup the investment (at which point it's probably time to upgrade the server). Other factors are usually the driving consideration.
Saving 40W per CPU socket may not save you money directly in terms of power bills, but generally speaking these chips are going into servers that sit in a datacenter. Air conditioning for the datacenter typically has costs directly related to the amount of power being consumed, so every 40W of power you can save could end up saving another 20W-40W of power in air conditioning requirements. That's still not even the primary concern for a lot of companies, though.
Datacenters often run dozens or even hundreds of servers within a single large room, and the real problem is making sure that there's enough power available to run all of the equipment. The cost of building a datacenter is anything but cheap, and if you can pack more processing power into the same amount of space, that is where low-voltage parts can really become useful. Blade servers were specifically created to address this requirement, and if you can reduce the total power use of the servers by 20% that means some companies could choose to run 20% more servers.
Of course, that doesn't mean that every company out there is interested in running a datacenter with hundreds of computers, so individually businesses need to look at what sort of server setup will best fit their needs. After determining that, then they need to look at low-voltage CPUs and decide whether or not they would actually be helpful. Assuming low-voltage parts are desired, the good news is that it's extremely easy to get them in most modern servers. Dell, HP, and other large server vendors usually include low-voltage parts as an easy upgrade for a small price premium. And that brings us to our low-voltage CPU update.
Intel Quad G-Stepping
Intel doesn't seem to sit still these days, pushing the power and performance envelope further and further. Recently, Intel announced two new G-stepping quad-core parts. The new parts run at the extreme ends of the power consumption spectrum. The first is a 2.0GHz 1333FSB part that runs at 50W while the second is a 3.0GHz 1333FSB part that runs at 120W. There are two main changes to the G-stepping parts, the first of which is power consumption: G-stepping introduces optimizations for idle state power. The second change involves enhancements to the Virtualization Extensions (VT), which mainly improve interrupt handling in the virtualization of Microsoft Windows 32-bit operating systems.
Of course, we would be remiss if we didn't mention AMD's recently launched Barcelona processor here. AMD expects their new quad-core processor to run within the same power envelope as the previous dual-core Opterons, which means twice as many CPU cores potentially without increasing power requirements, resulting in a potential doubling of performance/watt on the socket level. Low-voltage (HE) Barcelona parts will still be available, but even the regular chips include many new enhancements to help with power requirements. We are doing our best to get some additional Barcelona servers in-house in order to test this aspect of the performance/power equation and we hope to follow up in the near future.
One final item worth mentioning is that Intel's 45nm Harpertown refresh of Clovertown is due out in the very near future, which is one more item we can to look forward to testing. Unlike the desktop world, however, acquiring and testing server products often requires a lot more time and effort. Even with the appropriate hardware, the sort of benchmarks we run on servers can often take many hours just to complete a single test, and there are many parameters that can be tuned to improve performance. Since there aren't a lot of early adopters in the server market, though, we should be able to provide you with results before any of the IT departments out there are ready to upgrade. Now let's get on to the testing.
Facebook
Twitter
RSS
36 Comments
View All Comments
Justin Case - Saturday, September 15, 2007 - link
I would rather have an article called "quad core coming to play" that actually tests quad-core CPUs from both major manufacturers, and not quad core from one versus dual-core from the other. This article would have made perfect sense one week after Intel's quad-core Xeons were launched. Not one week after Barcelona was launched.As other have pointe out, this article is obsolete on arrival. People choosing between dual-core Opterons and quad-core Xeons have made their decision a long time ago. Anyone who still hasn't is obviously waiting for quad-core Opteron numbers.
If three days (actually seven, but let's pretend your Barcelona systems were busy 24/7 during the first four) weren't enough to run a single "performance-per-watt" server benchmark on both systems (or on a dual-core vs. quad-core Opteron), and present a comparison, then maybe it would have made more sense to wait another day or two, and release a relevant article instead of one that looks more like a (bad) Intel PR exercise, timed just to coincide with the lanuch of faster CPUs from the competition.
Griswold - Friday, September 14, 2007 - link
Funny link to the "Intel Resource Center" at the end of the article. Could also just call it the "AnandTech Intel Filter sponsored by Intel" for it only presents articles related to Intel.No, I'm not complaining. Just pointing out something mildly amusing, because I havent seen such sponsoring yet (nor did I really expect AT to accept it).
Goty - Thursday, September 13, 2007 - link
Since it appears that AMD made good on its work to deliver Barcelona chips that don't consume any more power than their dual-core Opteron counterparts, I'm thinking that Intel is either going to have to make some significant speed improvements with Harpertown or lower the power consumed by the platform somewhat to beat AMD in the performance/watt category. The move to 45nm will certainly help in that area (barring another Prescott SNAFU), but it's still going to be interesting.cmdrdredd - Thursday, September 13, 2007 - link
Performance per watt? Why do people keep saying that...to most people that really matters little in the grand picture. They want the best performance at a certain pricepoint. Who counts wattage when making a choice in CPU or any type of hardware not including a PSU. If one is faster at a certain price you're ready to pay then that's the one you'll get. Performance per watt is something people say when their choice is slower.smitty3268 - Friday, September 14, 2007 - link
You're right that people with a couple servers running don't really care too much, unless it's extremely bad like Prescott was. However, there is a reason that both Intel and AMD have been hyping efficiency and coming out with special low-power products. It's because the data centers that are trying to cram 200 computers into a single room without overloading the buildings power system can get more computing power given their limited resources. For this market price almost doesn't matter, as it costs much more to get a new building then it does to pay more to get lower power efficient cpus.Goty - Thursday, September 13, 2007 - link
Everything you're saying is pretty much true... until you actually start talking about the market sectors these parts are aimed at.You have to get out of your tiny little mindset that all new product releases are aimed at enthusiasts and consider that there is more than one market out there.
In the server and HPC spaces, performance per watt is very important because lower power consumption allows for greater density (in most cases). Who cares if one architecture is 15% faster per-core-per-clock when I can fit 25% more of the slower architecture into the same power and space envelope?
mattsaccount - Thursday, September 13, 2007 - link
"What, no Barcelona?"Frags - Saturday, September 15, 2007 - link
Barcelona is being shipped to OEMs and is not available to customers regardless of what AMD says. Are you suprised that there is no Barcelona Benchmarks? There hasn't been any for months and even weeks before the launch. AMD is the only one to blame for no 3rd party benchmarks as they don't offer press kits ahead of time for 3rd party testing. If they have, it's a rarity. For now it's back to the AMD slideshows.Proteusza - Friday, September 14, 2007 - link
Their decision was sound from a business point of view. Native quad core is difficult, it costs a lot. AMD was months late to market with their quad core, most likely in part due to it.
Yet, with AMD's native quad core CPU already out, I fail to see why you say that Intel decision was the right one. Had you posted this article a month ago, yes I would have agreed that Intel's quad cores were running rings around AMD, and despite their higher power efficiency in some cases, Intel was winning over all, especially in higher loads. But now, the situation is different - AMD has their own native quad core, which in all likelihood is more energy efficient than the Intel. Thus, your conclusion isnt valid - Intel's choice was right at the time, because AMD could not compete, but now they can, and now its quad core vs quad core, and the fact that AMD's quad core is a single die gives it a small but significant advantage.
Why did you not benchmark the Barcelona chip? I'm aware its not really available in volume right now, but...
coolme - Monday, September 17, 2007 - link
Intel having quad-core before AMD did give them the upper hand, because native quad-core as it turns out, only gave a small advantage to AMD. Although it doesn't seem like AMD is winning this race with barcelona, they need to stop their stubborness and accept the fact that MXM design is the way to go and start pumping out octa-cores before intel does. It's the only way amd can regain their crown. I mean the fact that native quad-core means low yields is already costing them millions, they need to accept the fact that MXM is a more efficient solution for making processors.