Slashdot Mirror


POWER Processors, SMT and the True Origins of AI

Crow writes "IBM developerWorks has posted an interview with John McCalpin, one of the guys who works on the POWER line of processors. He discusses work on POWER5 (and how the design process works at IBM -- he's also involved in work on the POWER7) and defends the decision that IBM doesn't hand-tune their ICs (as has often been criticized on Ars Technica. Also covers some of the features in the POWER processors, like SMT, the Hypervisor and virtualization -- even addresses the question of whether AIX was designed by space aliens or not. The POWER5 just broke the 3 million transactions per minute barrier on the TPC-C benchmark."

34 comments

  1. AI by jbloggs · · Score: 3, Insightful

    Where's the bit on the true origins of AI?

    1. Re:AI by shadow303 · · Score: 3, Insightful

      I think that was supposed to be AIX instead of AI.

      --
      I've got a mind like a steel trap - it's got an animal's foot stuck in it.
    2. Re:AI by commodoresloat · · Score: 1

      Actually, AIX was the true origin of AI. That's why they needed help from space aliens.

  2. Aliens, IBM, and AIX oh my! by infinite_improbabili · · Score: 5, Funny

    dW: That is a shame. I have one more on the subject of OSes -- was AIX really designed by space aliens?
    McCalpin: I hadn't heard that one. It feels like it was designed by, [not space aliens, but] in some part by people with mainframe backgrounds..."

    Apparently the aliens have embedded mind control in AIX!

  3. O God by dretay · · Score: 0

    I stopped reading the article after I saw this:
    That's my classification, but we don't have any other kinds of titles here. On my office door the sign says, "Dr. Bandwidth" - is that a title?
    I wonder if ignorant people actually address him as such?

    1. Re:O God by Anonymous Coward · · Score: 2, Funny

      No, sometimes we call him the Bandwidth Bigot.

    2. Re:O God by putaro · · Score: 3, Funny

      I used to have cards that read "Speaker to Disk Drives" as my title when I was at Apple.

  4. Comp Eng at IBM by KingOfTheNerds · · Score: 4, Insightful

    I worked recently as a contractor for IBM where I helped contribute to the design decisions for the POWER7. This article is great because it describes how the design decisions happen at IBM and it is right on the money. It's interesting because the way IBM does it is different from most other companies.

    --
    Want to learn about anything sexual? Check out the sex wiki:
    1. Re:Comp Eng at IBM by pohl · · Score: 5, Informative

      Though your skepticism may be warranted, it appears that you have no understanding of what it means to work on a large team engineering effort. The article mentions in the first paragraph that there are hundreds of engineers working on the POWER processors. Wouldn't cooperation be an essential aspect of making that work? Do you think everybody stands around sucking their thumb while one hotshot at the top makes the design decisions? It's more likely that experimentation and lively debates lead to design decisions, don't you think?

      --

      The "cue the foo posts in 3, 2, 1..." posts will commence with no subsequent foo posts in 3, 2, 1...

    2. Re:Comp Eng at IBM by bursch-X · · Score: 1

      Well if he contributed anything himself, he wouldn't write he helped contribute.

      I guess he just let a friend who contributed something use his computer.

      --
      There are two rules for success:
      1. Never tell everything you know.
    3. Re:Comp Eng at IBM by Anonymous Coward · · Score: 0
      it appears that you have no understanding of what it means to work on a large team engineering effort.

      It appears that you have no understanding of what it means to contribute. If you "help" contribute, then you contribute. But anyone who feels the need to couch it with "help" didn't actually do jack shit.

    4. Re:Comp Eng at IBM by Anonymous Coward · · Score: 0

      It appears you have no understanding of what it means to post to slashdot. I'm guessing the guy is a little more careful with his language on his resume than he is while posting here, so cut him some slack if he typed "helped contribute" without carefully pondering the meaning of the term "contribute" and flagging the redundancy before hitting "submit."

    5. Re:Comp Eng at IBM by TheLink · · Score: 1

      Could you elaborate on the differences?

      I see:
      "We spend a gread deal of time in meetings"
      and
      "So, the issues as far as the design of the system are argued out in many meetings by the technical leadership, some of whom are first and second line managers and some of whom are STSM's and DE's [Distinguished Engineers --eds.] and Fellows. We eventually come to a level of consensus about which of the items that have been proposed are fundable and the remainder are not. "

      How is that different from the other companies you have experience with?

      --
    6. Re:Comp Eng at IBM by mkiwi · · Score: 1
      As a part of EECpE at Iowa State University, I know how valuable team and communication skills are to making a successful project. I 100% agree with the parent post, because there are so many different ways to design hardware and software. Just for the record, in an earlier post I ruffled someone's pantaloons by saying CMOS processors are limited to about 25 Mhz. TTL/CMOS goes up to about 125 Mhz, and ECL (Emitter Coupled Logic) comes from the TTL family. The theory behind it is that when a transistor is in partial saturation, it can get a logic HIGH (1) or LOW (0) switched faster than going through an entire cycle of "on's and off's."

      This is the same technology IBM uses to design their processors, and it's kind of like walking a tight rope between signals. I would be interested to see some of the IBM fabrication techniques for this unique method of high-preformance computing.

  5. Vector supercomputers vs clusters by CaptKilljoy · · Score: 4, Insightful

    A good quote to break out when people claim that cluster supercomputers are "better" than vector supercomputers (when they're really two different types of systems with differing strengths that can't be directly compared):

    >McCalpin: The majority of high-end systems are used for throughput workloads of one kind or another. The vector machines, both from NEC and Cray(TM), are very well-liked by the end users because vectorization is a relatively easy thing to understand, how to write code that will vectorize. And the machines -- with relatively little effort -- give you a good utilization, you'll get a good fraction of the peak theoretical performance without a whole lot of work. And customers find that comforting. You put the code on there, you get 35% of the theoretical peak performance and you say, "Well, that's pretty good and I don't need to mess with it anymore."

    On the machines that IBM sells and that HP sells and AMD(TM) and all of the others, the costs are much lower, but it's harder to get very high utilization on those machines, in part because they don't have so much expensive memory bandwidth. So there's an interesting discrepancy between the end users who love vector machines because they're easy to use and then the purchasing manager who doesn't like vector machines because they cost too much.

  6. POWER and desktop computing by user9918277462 · · Score: 4, Interesting

    The IBM POWER line (including PPC) is the one and only true threat to Intel's dominance of the hardware market. Remember that AMD pays royalties to Intel for every chip it sells, which pretty much condemns them to permanent second-place competitor.

    Once the PPC 970 (or some successor) starts shipping in commodity beige boxes, however, the entire marketplace will be turned upside down. Can you imagine buying the generic equivalent of a dual proc G5 for $600? Sweet.

    1. Re:POWER and desktop computing by 0racle · · Score: 1

      Um, ya. Go read some newsgroup posts from around '90 on the google groups archives. Apperently we were all supposed to be using Moterolas, SPARCS, MIPS or something like it by '95 and have all this Intel crap as just a memory. Hmmm, let me check my systems here, oh look what they have.

      Its going to take a hell of a lot more then just a beige PPC box to change the status quo.

      --
      "I use a Mac because I'm just better than you are."
    2. Re:POWER and desktop computing by zpok · · Score: 3, Insightful

      Suppose a 90% Windows marketshare - and be very nice and political and give Linux and the Mac each 5%.

      So what you mean is the whole Linux 5% would switch over to PPC beige boxes?

      Considering the fact they already can today (well, beige is a bit out of the question, but Linux on PPC is a definite - and pretty cool - option today, and there's always Darwin, BSD and related options) don't you think that even a 5% increase of PPC's is far-fetched in this scenario?

      Or is Windows supposed to support IBM chips suddenly?

      Seriously, am I missing something?

      --
      I think, therefore I am...I think.
    3. Re:POWER and desktop computing by user9918277462 · · Score: 2, Interesting

      In the early 90s Intel's dominance was just ramping up. Remember back then? The cheapest PC you could get cost $1500 and was a 33 MHz 486 with 4MB RAM.

      In 2004 it's going on 20 years. We all know that market leadership is temporary, it is inevitable that someone will knock Intel from the pedestal. The question is not if but when, and by whom.

    4. Re:POWER and desktop computing by Anonymous Coward · · Score: 1, Interesting
      Or is Windows supposed to support IBM chips suddenly? Seriously, am I missing something?
      Yes.
    5. Re:POWER and desktop computing by jshriverWVU · · Score: 1
      >Remember that AMD pays royalties to Intel for every >chip it sells, which pretty much condemns them to >permanent second-place competitor.

      When or why would/does AMD have to pay Intel royalties? Is that still in effect or did it end with the AMD64's? I thought x86 in general was an open architecture, hense the tons of IBM-clones that hit the market. Remember reading once, that Intel specifically left x86 open, because the money was in software and not so much in the architecture, but that was 5-10 years ago I might be wrong.

      Though I would like more info :)

    6. Re:POWER and desktop computing by LWATCDR · · Score: 2, Interesting

      Actually the new Xbox will use the PPC. What do you think .net is all about except to free microsoft from Intel. Why else would you write a vm that only runs on one ISA?

      --
      See my blog http://ilovecookes.blogspot.com/ for light hearted technical information.
  7. POWER Processors, SMT and the True Origins of AI by Scott7477 · · Score: 1

    This interview was absolutely fascinating. The comment I found most interesting was McAlpin's opinion that 128-bit processing won't be coming to PC's any time soon. Excellent choice...

    --
    "Lack of technical competence coupled with the arrogance of power, as usual, leads to no good end."
  8. Re:POWER Processors, SMT and the True Origins of A by thogard · · Score: 1

    Its just like 64 bit processing. What does it buy you? In 99.999% of the cases it means more crud to move around and less effecent use of memory. Going from 32 bits to 64 bits didn't buy as much as going from 16 to 32 did and going 64 to 128 is almost pointless. Now a 32 bit machine that can munch on 1024 bit data at a time in an effecent way would be useful but now they call that a "video card"

  9. Power9+ VLIW clean full 32 64bit-regs, WAR,WAW,RAW by Anonymous Coward · · Score: 0
    My dream is to have a Power9+-like (4-level 8K-pages, alpha-like, delayed slots, ..):

    vliw0: LD R0,[R31] ; LD R1,[R27] ; LD R2,[R26] ; LD R3,[R28]
    vliw1: SUB R31,R31,#0x0008 ; ADD R30,R4,R5 ; ADD R29,R6,R7 ; ST [R29],R8
    vliw2: ST [R0],R19 ; ST [R1],R18 ; ST [R2],R17 ; ST [R3],R16
    vliw3: jgtu #vliw0,R31,R25 ; ADD R20,R12,R13 ; ADD R21,R14,R15 ; ADD R22,R16,R17
    vliw4: ...

    open4free ©

  10. Re:POWER Processors, SMT and the True Origins of A by Anonymous Coward · · Score: 0

    You are right; I have thought about building a machine based around one of these ATI or Nvidia cards. It would be interesting to see if anyone has done this.

  11. I'm 'Dr. Throughput' using lower consumption core. by Anonymous Coward · · Score: 0
    IF ((cache||tlb||page)'s miss && next-present-thread) THEN switch-few-cycles-context-to-back-inmediately-next -thread-from-stack END
    ELSE IF ((cache||tlb||page)'s miss && next-present-process) THEN switch-few-more-cycles-context-to-back-inmediately -next-process-from-stack END
    Programming multi-threads aplications are more easy (and interactive) than programming multi-processes aplications.

    It's better & cheaper than the complex Itanium2's out-of-order-execution.

    open4free ©