The green flash

The green flash

Summary: When the prospect of doing a blog came across my screen, I turned to my most trusted advisor on such matters, one of my 19-year old twin boys...


When the prospect of doing a blog came across my screen, I turned to my most trusted advisor on such matters, one of my 19-year old twin boys. He’s a technology buff in his own right and a freshman in EECS at Oregon State University. I managed to get him to look up from his latest copy of Wired for a few moments to answer my question, “If I were to If I'm Intel's CTO, why isn't this blog on the Intel site? do a blog, what would you want me to write about?” His answer convinced me that I might have something to unique to say, “People always want to know what will be new and cool, not just next year or even the year after, but in five years or even ten years and that’s what you’re really good at explaining.” With my mission statement so quickly crafted, I agreed to accepted the blog offer, and this is installment one.

Unfortunately, you won’t find much of the long range stuff in this first entry. Before jumping in, I thought it best to get the start-up information out of the way and, hopefully, save you from asking me a bunch of obvious questions. That’s not to say I don’t look forward to answering your questions, because I do, but I’d rather keep them focused on the technical topics to be discussed in the upcoming entries rather than all the whys and wherefores of doing a blog for ZDNet.

First off, I hope to give you a unique vantage point to the future of information technology. Most of what you read in print and online is at least one extra hop from the people doing the work. Even then it is generally after the fact or at least after the prototype is working. My job will be to eliminate the middle man so to speak and get you closer to the leading or, as we often say, the bleeding edge. Since part of my responsibilities include directing a 1000-person research organization, a lot of what I hope to say will be real-time. We’ll go inside the labs and talk about the ideas and experiments now underway that will define the future of information and communication technology (ICT) in five to ten years.

Weaving technology and opinion

This leads to an obvious question: If I'm Intel’s CTO, why isn't this blog on the Intel site? The simple answer is that Intel’s corporate site is not a place for personal opinion. It’s the place you go for information about new products or standards efforts. If that’s what you’re looking for, then please go there. What I plan to do here will be technical -- but also rife with opinion and, hopefully, insight you’re not likely to get anywhere else.

Sure, when there is new and relevant Intel product technology, I'll talk about it. It would be foolish not to. But my real focus will be on the ideas and innovations that will underlie information and communication systems of tomorrow. Similarly, I won’t be discussing the competitive environment or engaging in a lively debate over who has the best game platform. There are plenty of other web sites and blogs for having those discussions, so I won’t deal with them here.

Connecting the dots

One of the things I will try to do is seek out the technological inflection points, a term Intel’s legendary ex-CEO, Andy Grove, made a household phrase more than a decade ago. It’s often easy to miss them when you just look at the individual data points. My job will be to put them into context to see if they point to an impending inflection point in technology. Perhaps an historical example would be useful.

The move to multi-core processors caught a lot of people by surprise. I was shocked when I heard an industry visionary say he thought multi-core was just a short term work-around until we got the semiconductor processes back on track. He was literally expecting us to invent a new transistor that would be super fast and eliminate all the problems with power. Most importantly, from his point of view, it would let us get back to building screaming fast single-threaded processors. When he saw our 5-year processor roadmap had only multi-core processors from 2006 onward, he finally realized that the industry had reached a true inflection point. If I’m doing my job here, you won’t be similarly surprised in five years (more or less) when something equally dramatic occurs.

Another thing worth noting is that many of the data points will come from outside of Intel. As Bill Joy, another industry legend, once said, “Most of the smart people work for some other company,” It follows that most of the good ideas come from elsewhere, too. Since talking to technology leaders across industry and academia is part of my job description, I’ll do my best, without breaking confidentiality, to bring you their insights and observations. I’ll also to try to give you the global perspective of the trends and ideas shaping information and communication technology. As Intel’s network of research labs reaches across the U.S. to Europe, the Middle East, Asia, and now Mexico, I’ll share the views of the technology and policy leaders from around the world.

Let the dialog begin

I welcome your questions and comments, as well as your ideas for future topics. Feel free to point me to key breakthroughs, inflections or scenarios that I’ve missed. I’ll address the most interesting ones, and I’ll try to do so in a timely fashion.

Enough housekeeping. Now that you’ve seen the green flash, we can start to look over the horizon.

Topics: Intel, Browser, CXO, Hardware, Processors, IT Employment

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • General purpose vs Special purpose computing elements

    Justin, thank you for sharing your thoughts on a public blog site.

    I can see the inflection point, as you say, with multi-core (MC)- truly exciting. It seems Moore's law not only created the heat dissipation challenge, but also provided the solution with MC!

    Please comment if you believe special purpose computing elements (acceleration chips) alongside general purpose computing elements (microprocessors) are ALSO a coming inflection point for our industry? Do you feel the era of MC, based on many general purpose computing elements will also soon reach it's practical value and yeild to something else - like heterogeneous elements? Kind regards from Dallas.
    • Question

      By "Heterogeneous elements" you mean things like the 8087, 487SX, DX2/OVERDRIVE?

      I vaguely remember those...
      • Heterogeneous

        Heterogeneous, meaning core elements that are not of the same kind.
        I assume Intel refers to multi-core as having many cores that all look the same and do the same thing, ie. woodcrest-2 cores and clovertown-4 cores, etc.. implying a homogeneous elements.

        Heterogeneous, would refer to having multiple cores, but say, 1/2 the cores are similar CPU's and the rest are not. A good example is a physics acceleration core and an XML acceleration core all in the same monolithic package.
        • Hetero

          what about any of the math coprocessors? At a certain level it is just where you draw the line. Most single core processors consist of a number of heterogeneous elements. Since they are on the same die, you could say this already exists. Integrating chipset elements on the same die would also be an example of this. All of that has been done before.
  • It's about time...

    someone in a position of technical knowledge and authority at Intel
    started a blog. Good to see it's you Justin. Look forward to
    reading this on a regular basis.

    Best wishes,

    Dave Budde
  • Welcome to the blog scene

    It?s great to have you blogging Mr. Rattner. I guess we are going to be doing multicore processors for the foreseeable future. Do you think there will ever be a time where a technological advancement will permit to the return to single threaded scaling? I mean if it?s possible to have one big gun that has the aggregate throughput of a multicore processor then that would always be desirable.

    The way it stands now, your typical application scales anywhere from 10% to 100% improvement in performance in a two way system and mostly towards the lower end of the scale unless you can efficiently break a task in to two simultaneous threads. Of course if there were two processor intensive tasks that needed to be processed simultaneously to begin with then that?s no problem.
  • Use the Force, Justin

    Justin - Dan Warne from APC Magazine in Australia here. Looking forward to reading your blog very much. When I met you at IDF SF recently, it struck me that Intel and you in particular have a much more interesting vision of the PC than often gets portrayed in the regular media. I hope you can use this blog to create some headlines that push system makers to be a bit more adventurous in system design (particularly in advanced features they could include like array microphones for VoIP.)
  • I need today's information, not tomorrow's

    Mr. Rattner -

    I think maybe you should have consulted some real IT professions about what they need to do their jobs, not a 19 year college student. Most It professionals are not given the budget for today's hardware, let alone tomorrow's hardware.

    More importantly, what is Intel, Sun, Microsoft, and other companies doing to improve the way coders write code, and the way the code runs?

    I posted a blog on TechRepublic just a few minutes ago about how I improved execution speed by [b]100 times[/b] by breaking "best practices" ( If I had followed standard thought, I would have told my users to just throw more hardware at the problem. This is what developers expect, and hardware companies thrive on.

    In the real world, we still have desktops running on Pentium III class Celeron processors; I know of a company with entire departments using Pentium I 90 mHz machines. This is the reality that IT professionals have to deal with. We deal with [b]yesterday[/b]. Our boss's do not sign off on high end hardware, and they give us grief when we want a mid-range system.

    Today, my boss and I looked at government surplus computers ($25 with keyboard, mouse, and monitor), and they were being snatched up as fast as the warehouse people could put them on the shelves. They were Pentium III, 600 mHz systems. I saw people buying them in bulk for businesses, and people buying them for their kids. These are machines that were being used a few weeks ago by government employees, and are now about to become the primary work machine for hundreds of people. When do you think the people using these computers are going to get their first hyper threaded or multi core processor? 2010? 2012?

    I write software for a living and perform systems and network administration. My customers, as part of their standard computer installation, rip off Windows XP, and put on Windows 2000 and Office 2000. Yes, they throw out a Windows XP license, just to ease their management pain. This is the reality that I, as an IT professional, need to deal with.

    Real IT people have limited amounts of time. I know I certainly do. I do not have the time for rumor mongering, vaporware, "what-ifs", "just-so-stories", and other nonsense. My reality fills my day. Thinking about five years from now is a complete waste of my time. If I am lucky, five years from now, the IT infrastructure I deal with will hopefully be up to snuff by today's standards. If I want fanatsy, I will read Gartner reports, or better yet, an adult magazine. If I want reality, I want to get that from ZDNet.

    There are plenty of other bloggers on this site doing future oriented stuff. I do not read them. Yes, you may be in a unique position to talk about the future of hardware with a great amount of accuracy, but this is not what I need. Do the community a service. Ignore your 19 year old son, and listen to real IT people. Give us real information about today's reality that we can bank on. And if you want to help me do my job better, give me some quality ROI information that I can bring to my boss to justify my hardware requests.

    Justin James
    • Re: I need today's information, not tomorrow's

      Well, Mr. James, if Mr. Rattner does not want to talk about what you want to hear, then you don't have to waste your time reading it. He is in charge of the future for Intel, not the past. That's his specialty. Surely you don't think that the manager of "Hydrogen-fueled internal combustion engine research" at GM would be the best person to help you with a carburetor problem on your 1967 Pontiac, do you?

      I'm looking forward to his blog. Sorry it doesn't work for you, obviously nobody can be "all things to all people."
      • Today's past was Yesterday's future

        Rattner has been around a while.

        Certainly you don't think that someone who doesn't know why yesterday's decisions were good or bad is going to be particularly effective at making tomorrow's decisions, do you?
  • MC moving forward


    With x86 it use to be the MHz race and now it is the MC race, while I/O Latency and the lack of multi-treaded OS and applications hold back the end user from really being able to utilize the processing power. Multi-processor systems have been available for along time, when will software caught up to hardware and start taking advanage of MC processor systems?

    Thank you
    • The answer to that is...

      ... most likely NEVER. Most software, outside of OS's, bug RDBMS's, etc. still are not written to take advantage of SMP, let alone MC. Why? Because people are writing in high level languages, and it is up to the compiler or managed code environment (Java, .Net, etc.) to do the work. Dev tool makers just do not seem to be keeping up, and for most projects, the developer simply does not have the time or resources (time, knowledge, QA team, hardware, etc.) to write code that really takes advantage of the new architechtures. Most of your software is still running on only one or two threads. Look at the new T1 Niagara SPARC CPU; the servers using it show basically zero performance boost on anything that is not multithreaded... and most of your code isn't multithreaded. The upshot? Single threaded apps, or apps that do most processing in a single thread do not get any advantage from the new CPU technologies, and show no performance gain. George Ou's previous comment to this blog is spot on. When are chip makers going to recognize reality, and design CPUs to enhance the performance of how code is actually written, as opposed to making CPUs that require a massive, across the board change in how code gets written?

      Justin James
  • remember the people

    great to see you talking about what we can do with technology 5 even 10 years from now. I would like to understand your thoughts on how to make computing truly become a supporter of people's work/play/desires, not the technology as an end state. It feels that we, as humans, have to alter ourselves to fit technology. I would like to see things like active knowledge support, inferencing, etc more engrained. Thoughts?
  • Welcome to the ranks of Intel Bloggers!

    Welcome to the ranks of those of us who work at Intel, and (sometimes) write about it on our blogs. I'm really glad to see someone at the CxO level from Intel blogging, and I hope more of your executive brethren join you!

    Intel employees are very active in the blogosphere, inside and outside the company, and we're happy to see you join us.

    I'd love to chat with you some time about how you feel about your blogging, etc. Drop me a line any time! :-)

    Josh Bancroft
  • $

    Mr. Rattner,

    Are you getting paid for this?
    Just wondering why ZDNET is a better choice for you than
    • Not speaking for Intel

      He said: "If I'm Intel?s CTO, why isn't this blog on the Intel site? The simple answer is that Intel?s corporate site is not a place for personal opinion."

      CxO level execs do sometimes speak officially in blogs and can make news doing it, as Sun's Jonathan Schwartz has shown. If Rattner isn't speaking for Intel, blogging at ZDNet makes sense. He's a senior Intel executive, but here he's just sharing his personal perspective.
  • The Pentium D

    I was very pleasantly surprised how freaking fast a 3 GHZ PentiumD was. I went through an upgrade at the office about 2 months ago. First time I've had a machine at work faster than what I run at home. I'm looking forward to the day when the software writers wake up and can deliver multi-threading-multi-processor apps.