1 877 422 8729

For the release of the new Intel® Xeon® X3-1200 V3 “Haswell” processors, we’ve brought in Emily Hutson from Intel’s Channel Server Product Marketing department. Emily joins Alexey and Mike on this latest episode of “Server Talk” to discuss the new features and advantages of the third generation of Xeon E3 processors.

Topics Covered:

  • Introduction to Haswell architecture
  • Haswell vs. previous generation Xeon® E3 “Ivy Bridge”
  • Four areas of focus for Haswell: small-business servers, workstations, server graphics, microservers
  • Energy efficiency and graphics performance
  • Haswell hardware-based security features
  • Where the new technology fits in
  • Cost of ownership questions

Get started with ICC Haswell server solutions today!

Full Podcast Transcription:

Mike: Welcome to another exciting edition of Server Talk. Today’s a very special edition because I’m not only here with Alexey. I’m here with Alexey and Emily from Intel®. So, the reason we have our special guest on today is because Intel® announced its release of the new Haswell Architecture. Now, I don’t know too much about it and that’s why we invited Emily to discuss it a little further.

Alexey: So, Emily, can you tell us what you do at Intel®?

Emily: Sure. I work in our data center group and specifically, I do marketing for our Intel® Xeon® processors that are put into our small business servers.

Alexey: Okay, so that would fit right in there.

Emily: Absolutely. So, Haswell Microarchitecture is coming out June 2nd. We announced the launch of it. There are three different versions. There’s one for desktop, and one for mobile and then the version that I’m gonna talk about today which is going to be used for servers – for entry servers platforms and workstation platforms.

Mike: Okay, sounds good. What is the significance of this new architecture?

Emily: So, it’s a follow on to our Ivy Bridge Architecture. It’s also on a 22-nanometer process but it’s going to be using a different socket. So, the major changes or features of Haswell versus the previous generation is that we’re going to be seeing more power savings and performance benefits, and we’re also gonna be seeing greater graphics processor technologies. So, there’s an integrated GPU – Graphics Processor Unit – on the Haswell technology and it’s gonna be having twice the performance of the previous generation Ivy Bridge. So, those are the big features that will be updated with Haswell. So, better performance, better performance per watts and then, better graphic performance.

Mike: Okay, what kind of servers do you think are best suited to take advantage? What are the purposes of these servers? So, you know, are we gonna be looking at databases or what?

Alexey: What application?

Emily: Well, it’s interesting. There are a number of good applications for this new Haswell Xeon® processor technology. So, the four main ones that we’re thinking that can be used in are:

  1. I mentioned small business servers, so entry-server pedestals.
  2. Also, workstations – entry workstations, they can take advantage of the integrated HD graphics performance.
  3. Then, we have a couple new… kind of… newer technology. So, one of them would be using them in a media server, so media or cloud gaming. Those are server graphics usages. You’d have a rack of servers that would be used for streaming media or doing cloud host gaming.
  4. Another one would be for microservers and so when we say “microserver”, we don’t mean a little server. What we mean is a number of small lightweight nodes so they amass in a larger server unit. So, you could have a server that has, say, any nodes and each node has its own processor in it and that would be a microserver. And so, those are really used for a lot of web hosting applications. So, if you’re a webhoster, you need to deliver up a lot of content. It doesn’t require a lot of processing power but it requires a lot of redundancy so a microserver would be great for that. And, with this new Haswell Xeon®, we have the lowest power Xeon® ever so it goes down to 13 watts. So, that’s great if you have a lot of processing to do and you want to save on your power consumption. It’s great for those data center usages.

Emily: But, I’d say for the most of your listeners, they’re gonna be seeing this Haswell Intel® Xeon® Processor E3 V3 and the servers for small business and for the workstations.

Mike: So, how does the power consumption compare against the old model? You mentioned that it’s 13 watts?

Emily: Yes, that’s the lowest power. You probably won’t be seeing a 13-watt processor in a small business server or workstation. You know, we are concerned about power consumption in Intel® but more on the database usage. So, say you have a server in your office and you just have one server. I mean, you don’t care that much about the power consumption. What, you’re saving $5 a year? But you are seeing better performance per watt, overall. So, that is the benefit but I think it’s a benefit that the data center usages will see more than, say, if you just have one server or one workstation in your office.

Mike: You mentioned that there’s a significant improvement in the graphics? Can you talk about that? Can you expound about it a little bit more?

Emily: Sure. So, Intel® HD graphics is the brand name of this integrated graphics processors that we build into our computer processors. So, the previous generation Ivy Bridge had HD Graphics 4000 which had 16 execution units. This next generation Haswell is coming out with Intel® HD Graphics 4600 which has 20 execution units. So, that’s definitely a proven in the performance of the graphics. When we say we have graphics in our CPUs, you know, we are comparing them performance-wise, to say, an entry-level external graphics card or discrete graphics card and we’re seeing very similar performance. So, if you are a price-sensitive workstation user, for example, you wouldn’t have to spend the extra $150 to $200 to buy an external graphics card. You could rely on the integrated HD graphics in your Intel Xeon® Haswell CPU and get a comparable performance. So, it’s definitely a benefit to the entry workstation market.

Alexey: So, Emily, one thing that we’re seeing a lot nowadays is in workstations, users want more than one monitor and traditionally on motherboards you would have a limited number of monitor support. With this more horsepower graphic card, do we see more monitor support?

Emily: You are absolutely right, we do. We see support for up to three displays. So, that’s something you’re absolutely right – that the workstations users need to use and we’re basically building the capability support up to three displays.

Alexey: Ok, and so, what about… does the memory still – the memory for the graphics card – is that used as part of the system memory?

Emily: Exactly. It relies on the system memory. So, for the E3 V3 Haswell, it would support up to 32GB of memory capacity. There are 2 channels of memory and it would support either DDR3 1600 or 1333 memory. It also now supports low voltage memory which is DDR3L 1600 if you’re concerned about the power usage of that.

Alexey: Ok, so I see a few trends in Haswell. One of them is that low power consumption targeted at data centers against specifically that microserver architecture where you have a lot of nodes in a small space, and we covered power consumption a little bit in a previous podcast. So, how do you… power consumption users just to lower the power consumption and improve graphics for workstations – you don’t need additional cards and you can see a lot of performance just straight off from the processor?

Emily: Absolutely right. Those are all very… You summed it up great. I also wanna highlight a couple of things for our small business customers. Some features that we have had and will continue to have on Haswell Xeon®: one is the support for ECC or Error Correcting Code Memory. And so, Error Correcting Code Memory is more advanced than the standard memory that you would get on your desktop or your laptop system. What it can do is basically, if there is a memory error that occurs, it will automatically correct that memory error and keep your system running. So, if you’ve ever had a laptop, what they call the “blue screen of death”, that’s often due to a memory error. And if you are running your data on a server or you’re using a workstation, you know, you would know that if you have a blue screen of death and you haven’t saved your work or if it’s a server and you’re running your whole business up that server, you don’t want your system shut down. And so, having Error Correcting Code Memory support on those platforms is really, really useful for those usages.

Alexey: Now, is ECC on the server and workstation based or is that all on Haswell?

Emily: It’s supported by the Haswell Xeon® platforms and so, not on the desktop or the mobile laptop.

Mike: So, Emily, what is the lifespan of this new architecture and where do you see it going? Where do you think Intel®’s gonna take it further?

Emily: Well, I can’t predict too much into the future you know but I think we’re gonna continue the focus future generations on… still continuing to lower that power consumption and also improve our graphics technologies. Another thing I wanted to mention for Haswell is that we have some new security features that are hardware-based security technologies that are built into the processor.

Mike: And what will those security features be used for?

Emily: So, they’re basically designed to help protect platforms or systems from being changed without authorization. So, one of them is kind of designed to protect the system BIOS, and another one is designed to protect the system operating system. And, they’re both known as Intel® Platform Protection Technology and so, under that Intel® Platform Protection Technology we have something called BIOS Guard… Let me say this again, so with the new Haswell, we have some new integrated hardware security features. They are bucketed under something called Intel® Platform Protection Technology. These includes BIOS guards which help protects the BIOS from being changed without authorization, and protects it during BIOS updates. We also have OS guard which protects the operating system from attacks. And so, both of these are hardware-based built-in security technologies that will be included with Haswell. We also have a couple data protection technologies. One of them is called Secure Key. What it does is provide random numbers – it’s a hardware-based random numbers generator. A lot of security applications will use random number generators to validate their applications. And so, this is a hardware-based random number generator that is super secure. We also have support for ASNI, which is the encryption standard. It uses hardware to accelerate data encryption and decryption. It’s great if you’re using SSD for example, and you want to speed up the amount of time it would take to encrypt your data on your hard drive.

Alexey: Will users use all of these in a server or workstation, or is it a mixture? Like, I feel encryption and drives can be used in workstations in case someone pulls it out, can it be used in a server environment as well?

Emily: It can be. I think you’re correct in that there’s more of a risk of someone removing the drive if it’s a workstation versus a server. I think that’s probably true and a lot of servers – server platform, server chasis – will actually have features like a physical lock on the drive base so that you can’t remove the drive but it’s a great feature to have if you do want to end up securing your data using encryption.

Alexey: Alright. Now, in terms of performance – just raw performance of the processor – compared to the previous generation Ivy Bridge, what would you say the percentage increase would be versus Ivy Bridge?

Emily: Sure. You know, one thing I would also say about performance is that we do compare generation to generation but a lot of times, when people are making a purchasing decision, they’re not on a year refresh cycle. So, they may be looking at a two or three-year old system that they wanna upgrade. So, we also like to look at the performance data that way too. But, for servers, we are seeing around 18% generation to generation performance improvement.

Alexey: Okay

Emily: And then, if we’re looking at versus 2-years old, we’re looking at about 40% generation to generation energy-efficient performance, I should say.

Alexey: Okay, so we’re definitely seeing a lot more performance in same SKU or the same BIN level, right?

Emily: Certainly, certainly. We’re definitely seeing more performance, a better energy efficient performance, better graphics performance and at similar price points to the existing generations. So, we’re making it really easy for our customers to upgrade.

Alexey: Okay. Now I know the SKU-ing (method) and how the SKUs for Haswell and Ivy Bridge works. For our listeners, can you tell us, if they’re choosing a processor, what are the different characters in the part number and what do they mean?

*For our listeners who may not be familiar with the term, SKU stands for ‘stock keeping unit’ and in this case refers to the processors’ unique ID

Emily: Great question. Well, we definitely don’t make it easy to understand things like that in Intel®. But, for the E3-1200 V3, that’s the official name of the Haswell processor – Xeon® processor family. So, we have five SKUs that are designated as server SKUs: the E3-1280 V3, 1270 V3, 1240 V3, 1230 V3 and 1220 V3. So, those are five of the SKUs in the family that we’ve said are for server use. We have three workstations skus – the 1275 V3, 1245 V3 and 1225 V3. Those all have the integrated graphics on them. So, the server SKUs, do not have the integrated graphics the workstations SKUs do. And the way you can tell is that if it ends with a 5, it has graphics.

Alexey: So, the previous generation was V2. So, this one’s gonna be V3 right?

Emily: Correct.

Alexey: So, it’s easy for users to identify “Hey, we’re using a 1230 v2; we just have to go to 1230 V3 again”. More performance for roughly the same price range.

Emily: Absolutely, yes.

Mike: You guys took it off the deep end with the technical stuff…

Emily: So, one thing that I’d like to… to maybe bring it back from the technical side of things… you know, one thing we see a lot is the trend about cloud services and I think the “cloud” is a big buzzword right now. People are thinking that everyone is just going to go to the cloud and maybe we don’t need to have a server in your office anymore. And so, we’ve looked at that and we kind of come to the conclusion that the cloud is great and a lot of small businesses, for example are already in the cloud. They may not even realize it – they may be using Gmail or Office 360 or doing their hosted email in the cloud. But, we still really see a strong value proposition for having a server on premise. So, for example, let’s say you want to store 4TB of data. And if you want to just store it in a cloud using the Amazon cloud service; it would cost you $400 a month to store 4 terabytes of data in the cloud. However, you could spend $1200 and you could get a server with 4 terabytes of storage, and that you could have at your office, and that would be an expense you can write off; and you can access that anytime. And so, in three months, it will basically be paying for itself. And certainly, depending on where you’re located there are benefits to having your data hosted locally versus backed up in the cloud. I’m sure everyone has had their internet go out once in awhile and if your whole business is in a cloud, you don’t have access to it that could be pretty frustrating. So, just something to think about as you’re looking at your small business customer and you’re looking at your IT infrastructure, you’re thinking “How do I wanna do this? Do I want to do everything in the cloud?” Maybe you’re not ready for that yet. Maybe you wanna put some things in the cloud but you also wanna have a server on site to host your data and to back it up. I think that’s a great option.

Mike: Gotcha

Alexey: So, one of the questions we still have a lot of times come up is what’s the cost of ownership of this new technology versus the older technology? Where does Intel® see the cost of ownership for Haswell?

Emily: Well, I think it’s not just in isolation right? You’re gonna be buying a platform that has Haswell on it. I think that we’re seeing around a 3 to 4-year lifespan for your server. I think one of the benefits of purchasing a Xeon®-based server is that the platform often has a lot of capability to expand so you can add hard drive and memory down the road as your business grows. They tend to be priced affordably so maybe not too much more than a high-end desktop. You could buy an Intel® Xeon®-based server onto your workstation and Haswell and then you can have the flexibility to add memory, add storage as you go.

Alexey: I think the question was meant in a little bit of a different way. So, say for example in your data center, you have five nodes and they’re consuming 70 watts per node. And then with this new technology, they’re gonna consume less power. You have to spend money upfront to purchase the system, roughly how long would you expect it to get paid off?

Emily: I don’t really have that kind of calculation off the top of my head in terms of break even on that. I’m sorry.

Alexey: That’s alright. Usually you guys have these little charts that say “Hey, in 6 to 7 months…” But, alright.

Emily: We do, in some platforms. Not so much for this one. I do that more for E5 where they do the TCO. I haven’t seen a TCO breakout for this one, really.

Alexey: And E5, that’s definitely the one that I’ve seen.

Mike: So, we’re gonna publish a blog post [published earlier] that goes along with this podcast so if that information becomes available to us then we could represent it graphically or in the blog post itself.

Emily: Sure

Mike: Emily, thank you so much for your time. We really appreciate you coming out today and introducing Haswell.

Emily: You’re very welcome. I hope this was useful for your listeners.

Mike: Thank you very much

Alexey: Thank you.