Stage 2: Datacenter

Intel’s Datacenter Group SVP, Navin Shenoy, also took to the stage at CES in order to discuss some new products in Intel’s portfolio, as well as to deliver updates on ones that were disclosed last year. Back in 2018, Intel held its Datacenter Summit in August, where it lifted the lid on Cascade Lake, Cooper Lake, and 10nm Ice Lake. Along with this, we saw new instruction support for AI and security as the top two areas of discussion.

Cascade Lake: Get Yours Today

Intel’s first generation of Xeon Scalable processors, Skylake-SP, was launched over 18 months ago. We’ve been hearing about the update to that family, Cascade Lake-SP, for a while now, along with its brother Cascade Lake-AP and how it will tackle the market. The announcement today from Intel is that the company is now shipping Cascade Lake for revenue.

This means, to be crystal clear, that select customers are now purchasing production-quality processors. What this doesn’t mean is retail availability. These select customers are part of Intel’s early sampling program, and have likely been working with engineering samples for several months. These customers are likely the big cloud providers, the AWS / Google / Azure / Baidus of the world.

It’s worth pointing out that at Intel’s Datacenter Summit, they said that half of all of its Xeons sold were ‘custom’ processor configurations that were not sold though its distributors – these parts are often described as ‘off roadmap’. It is likely that when Intel says Cascade Lake-SP is shipping for revenue to select customers that they are likely to be purchasing these off-roadmap processors. They might be running at a higher TDP than Intel expects for the commercial parts, or have different core/cache/frequency/memory configurations as and when they are needed.

One of the big draws for Cascade Lake is Intel’s Optane DC Persistent Memory support, which will enable several terabytes of memory per socket, but also the hardware security patches for Spectre v2. Businesses who want to be sure their hardware is patched can guarantee security if it's in the hardware, rather than relying on a firmware/software stack. So this might be part of why Intel’s demand for 14nm CPUs is at an all-time high and outstripping supply – if a company wants to be 100% sure it is protected, they need the hardware with baked-in security.

The full retail launch of Cascade Lake is expected in 2019. Based on what we saw at Supercomputing in November, given by a rolling slide deck at the booth of one of Intel’s OEM partners, that time frame looks to be somewhere from March to May.

Nervana for Inference: NNP-I coming in 2019

To date, when Intel has discussed the Nervana family of processors, we have only known about them in the context of large-scale neural network acceleration. The idea is that these big pieces of silicon are designed to accelerate the types of compute commonly found in neural network training, at performance and power efficiency levels above and beyond what CPUs and GPUs can do. It has been disclosed that Intel is working on that family of parts, NNP-L, for a while now, and we are still waiting on a formal launch. But in the meantime, Intel is announcing today that it is working on a part that's optimized for inference as well.

There are two parts to implementing machine learning with neural networks: making the network learn (training), and then using the trained network on new information to do its job (inference). The algorithms are often designed such that the more you can train a network, the more accurate it is and sometimes the less computationally intensive it is to apply it to an external problem. The more resources you put into training, the better. But the scale of compute between training and inference is several orders of magnitude: you need a big processor for training, but don’t need a big processor for inference. This is where Intel’s announcement comes in.

The NNP-I is set to be a smaller version of the NNP-L and built specifically for inference, with Intel stating that it will be coming in 2019. Exact details are not being disclosed at this time, so we don't have any information on the interface (likely PCIe), power consumption, die size, architecture, etc. However, we can draw some parallels from Intel’s competition. NVIDIA has big Tesla V100 GPUs with HBM2 for training that can draw 300-350W each, with up to eight of them in a system at once. However for inference it has the Tesla P4, which is a small chip below 75W, and we’ve seen systems designed to hold 20 of NVIDIA's various inference processors at once. It is likely that this new NNP-I design is along the same lines.

Snow Ridge on 10nm: An SoC for Networking and 5G (Next-Gen Xeon-D?)

The Data Center Group will be making two specific announcements around 10nm. The first is disclosing the Snow Ridge family of processors, focused on networking and specifically targeting the wide array of 5G deployments coming up over the next decade. The purpose of Snow Ridge is to enable wireless access base stations and deployments, as well as functions required at the edge of the network, such as compute, virtualization, and potentially things like artificial intelligence.

Intel gave no other details, however going back in my mind, I realise that we’ve heard this before with Intel. They already have processors on their roadmap focused specifically on networking, with 40 GbE support and features like QuickAssist Technology to accelerate networking cryptography: the Xeon-D line of processors. This makes me believe that Snow Ridge will be the name for the next generation of Xeon D, either the Xeon D-2500 or Xeon D-3100, depending on the power envelope Intel is going for.

Given this assumption, and the fact that Intel has said that this is a 10nm processor, I suspect we’re looking at a multi-core Sunny Cove enterprise design with integrated networking MACs and support for lots of storage and lots of ECC memory. There’s an outside chance that it might support Optane, allowing for bigger memory deployments, although I wouldn’t put money on it at this stage.

Ice Lake Xeon Scalable on 10nm

To finish up Intel’s announcements, Nevin also talked about Ice Lake Xeon Scalable. At Intel’s Architecture Day, a processor was shown at the event that was described as Ice Lake Xeon, so this is just Intel repeating the fact that they now have working silicon in the labs. There is still no word as to how Intel is progressing here, with question marks over the yields of the smaller dies, let alone the larger Xeon ones. Working silicon in this case is just a functional test to make sure it works – what comes now is the tuning for frequency, power, performance, and optimizing the silicon layout to get all three. I’m hoping that Intel keeps us apprised of its progress here.

 

 

What Happened at CES 2018, and why CES 2019 is Different

A memory that will stick in my mind is Intel’s CES 2018 announcements. At the heart of the show, we wanted to know about the state of Intel’s 10nm process, and details were not readily available. 10nm wasn’t mentioned in the keynote, and when I tried to ask then-CEO Brian Krzanich about it, another Intel employee hastily cut in to the conversation saying that nothing more would be said. In the end we got a single sentence from Gregory Bryant at an early morning presentation the day after the keynote, and that sentence was only after 10 minutes of saying how well Intel was executing. That single sentence was to say that Intel was shipping 10nm parts in 2017, although so far only two consumer products (in limited quantities, and region specific) have ever been seen.

This year, coming off the back of Intel’s Architecture Day last month, shows that Intel is becoming more open to discussing future products and roadmaps. A lot of us in the press and analyst community are actively encouraging this trend to continue, and the contrast between CES 2018 and CES 2019 is clear to see. Companies tend to hide or obfuscate details when product execution isn’t going to plan; now that Intel is starting to open up with details, the outlook is clearly returning to one with more optimism.

Stage 1: Consumer
POST A COMMENT

61 Comments

View All Comments

  • TheJian - Monday, January 07, 2019 - link

    "The fact that Intel has achieved HVM with *any* 10nm product" as quoted at wccftech.

    I call BS until I can buy a 10nm DESKTOP chip. This even, is only coming for xmas 2019? LOL. HVM is I can BUY TODAY, or you're lying. I think they can do ONE chip, this one, which I think is Ice-Lake U, so very small compared to desktops. Again, I call BS if they can't make a ~200-300mm^2 desktop. Not ANY chip IMHO if you choose a very small one to HVM and then only for xmas, so just starting today even for the small chips? Is it even HVM yet for this one? Xmas is 12 months away, you only need to pile up a launch for a few months, so should be in a box by sept, back to school etc. If it's xmas I think this is just trying to stop shares prices from dropping due to yet another 10nm delay speech. Either way, 10nm hopes are now 2020 I guess for us desktop people, or do like me and buy 7nm AMD this year ;) I really doubt Intel 10nm will be much better if at all, than TSMC 7nm and 5nm TSMC a year later...LOL. Intel better hurry with 7nm or they've lost the fab race IMHO (said they were losing it 5yrs ago many times, here anandtech, seekingalpha etc). Good for AMD, but bad for America as I don't really want china/arabs winning fab wars vs. america.

    I'm really hoping AMD can put out a great 12core (main desktop), and buy a few 8 core apu for HTPC's for xmas (will upgrade board/mem/apu at xmas in 2-3 htpcs). Get 7nm out the door amd, and you have 2-4 cpus sold to my family :) I'm not waiting for 10nm Intel with more mitigations slowing me down again and probably still not solved in hardware totally. AMD has issues too, but more of them require you to be AT the desk to do damage, where with Intel you can do a lot of damage remotely.

    "greater execution capability is now part of the architecture"
    This whole speech seems like no info. What the heck is this? Are you saying you build worse execution capability into chips on purpose before? Is this not the goal of ALL cpu/gpu enhancements? More execution ability? I'm confused...LOL. I've heard of TRUSTED execution, but what is "greater" execution ability built in? You google that you get nothing. I'm feeling like 2019 Intel CES speech was a repeat of 2018 (nothing burger). Until I see a 10nm desktop, whatever man.

    https://www.anandtech.com/show/13774/intels-keynot...
    As anandtech says, these are the smallest chips to balance yield/cost. So HVM with terrible yields then? So you have to go SMALL...Is it HVM for ALL if you can't do anything but PUNY chips or they are too expensive to ship? TSMC has HVM for A12x (122mm^2), which is probably in the realm of the size of this chip based on previous die sizes of U chips. Again, not impressed Intel. Wake me when you put out 10nm desktops above 200mm^2 at least and AMD's coming cards have chips in the mid 200's or so it seems again at 7nm TSMC.

    "Terms like ‘5G’ and ‘Artificial Intelligence’ are only going to become more ubiquitous through 2020 and beyond, so Intel is jumping on it today."
    You're a little late, as everyone else is ALREADY here. See Nvidia etc. Is Intel available today, or 2020?

    "With Project Athena, Intel is going to discuss with OEMs, with partners, customers, software developers, etc. what they need in order to enable these new terms to provide a good user experience."
    "As this is a new program based in aspirational discussion followed by execution"
    Yeah, like I said, nothing burger again. We're about to "discuss" what we might JUMP ON, at a later date when we "EXECUTE" what we "DISCUSS" with oems etc...LOL. How about EXECUTING on some 10nm DESKTOP chips? Never mind, I can just buy AMD 7nm 12 core this year and a gen I won't mind gaming on finally ;) I just hope AMD prices to them make some NET INCOME finally so they can R&D 5nm etc going forward. You can't do that on <100mil a quarter vs. NV (1B a Q) or Intel (never mind...LOL). Price LIKE YOUR WINNING AMD! Not like some discount 2nd rate Intel. IF you win nearly every benchmark at 7nm vs. Intel 14nm (the case for another year probably if they release shortly), PRICE ABOVE Intel. PERIOD. No point in discounting chips if you're WINNING watts/perf. Make hay while the sun shines! Sell to the RICH or continue to be POOR AMD. Ask NV/Intel what this means (HEDT, Titan, 2080ti etc).
    Reply
  • CajunArson - Monday, January 07, 2019 - link

    Wow you are clearly butthurt and scared.

    Anyway, instead of a wall of text here's a fact: if you dropped Intel's non-volatile memory division.. you know the division that makes the "failed" Optane into AMD, then it would be by far the largest and most profitable segment of the company.

    And at Intel they call that division a "side hustle".
    Reply
  • TheJian - Tuesday, January 08, 2019 - link

    Not even sure what you mean by butthurt or scared? Irrelevant post. How about attacking the data? Typical Democrat post ;) Attack data, not the person. Not sure what Optane has to do with AMD, but I do own micron stock (and AMD...LOL), and am UP double (16.77 avg price of my shares), so I'll be laughing when 3dxpoint crap hits next year anyway :) MU/NVDA both great pics (1/2 off in both vs. recent highs), and AMD though risky, not really until Intel gets it's act together in the next 12-18 months. IE, 7nm vs. Intel 14nm is a loser for Intel until then if AMD gets them out soon that is, and never mind what happens with 64core 7nm vs. Intel 14nm 48 core. That's a bloodbath for Intel IMHO, and don't think 10nm Intel will fix it. I think this brings you even maybe, but 5nm right after on TSMC anyway maybe even beating Intel again to 7nm so keeping the lead. Time will tell.

    https://en.wikipedia.org/wiki/File:Graham%27s_Hier...
    Please read the pic, then understand how to respond :) You don't know how to debate. You lose.
    Reply
  • SpartanJet - Tuesday, January 08, 2019 - link

    I think you meant typical republican post. Lie about everything and hide your head in the sand when science is involved. Reply
  • PeachNCream - Tuesday, January 08, 2019 - link

    Not sure Anandtech's article comments are an appropriate venue for something involving American politics. Reply
  • FunBunny2 - Tuesday, January 08, 2019 - link

    trade wars and IP fights and the like aren't??? of course it is. there's a reason Apple's China Syndrome has crashed and burned. much of that is due to politics: Apple decamped production to China, and other off-shore facilities, because politics encouraged it. politics matter. Reply
  • PeachNCream - Wednesday, January 09, 2019 - link

    There are other places for that kind of discussion. Reply
  • TheJian - Wednesday, January 09, 2019 - link

    You just proved my point. Blah, blah, blah, I hate you, you suck, you didn't say it right, you lie, blah blah, I have no data, but again, you suck and I'm right...LOL. That is all I'm hearing as you said nothing more.

    Please be more specific. What science is ignored? When did the topic change to science anyway (I’m talking process and failed chips)?

    OK, as a republican (er, only because I have to register with someone, to vote in a presidential primary), what in my post was a lie?

    How about you attack the data in MY post instead of shifting topics like all dems do when faced with an actual argument. I’m supposed to defend republican posts I’m not even aware of now? I pointed out exactly what he did and BTW he did nothing but change topics to irrelevant junk not remotely related to my post regarding Intel’s failures in chips/process. However, I addressed his irrelevant comment anyway…LOL.

    How'd we NOW get to science and lies (is that relevant to my post?)? The OP responded to my post with "butthurt and scared"...WHAT? Again, this is the point I made with my "typical dem post" reply. ;) The OP responded to me dismantling the article, with, but but but, if Optane (who brought that up?) was part of AMD it would be the largest and most profitable segment of that company. How did we go from “failure of Intel process, and defective chips” to “er uh, that memory tech Intel has (that they’ve failed to successfully launch too BTW), is awesome and AMD sucks because they don’t have it”? Again, he proved my point. I have no comment on what you ACTUALLY said, so here’s my shiny object…LOL. Comic, you just did the same thing. I can’t make an argument, so er, uh, you lie and don’t like science! LOL.

    Read my post you responded to, where I gave a link that tells you how to form an argument so we can have a reasonable debate about something, rather than crap like you posted which only serves to PROVE my point yet AGAIN. You have not provided an argument (rather an opinion backed by nothing), and thus have nothing for me to debate...

    Can you provide an actual argument on science or lies? I see nothing. Then again, that isn’t even addressing my post either (which the OP ignored too), but I’d like to see you actual attempt to make a REAL point…LOL. Please stick to content in MY post, as it's not my job to defend OTHERS not relevant to what I said anyway :)
    Reply
  • sorten - Monday, January 07, 2019 - link

    Too much shouting. Reply
  • TheJian - Tuesday, January 08, 2019 - link

    https://en.wikipedia.org/wiki/File:Graham%27s_Hier...
    Figure out how this comment crap works ;) Who cares how I say it, it's the data that counts. You remind me of crazy AOC comments like:
    “I think that there’s a lot of people more concerned about being precisely, factually, and semantically correct than about being morally right,”

    Uh, yeah, I don't give a rats behind about what people think about my MORALS (or shouting...LOL). I care about getting the DATA and FACTS right. You should too, regardless of how it's said, or how the text looks. Be thankful I didn't cap it all...ROFL.
    Reply

Log in

Don't have an account? Sign up now