AMD responds to ZDNet reader's inquiry

AMD responds to ZDNet reader's inquiry

Summary: In reponse to my ongoing coverage of AMD's Turion 64 processor (a 64-bit capable mobile chip for the "thin & light" notebook market), ZDNet reader Bradley Coleman wrote in to find out why notebook manufacturers were choosing to go with the more power hungry of AMD's Turions rather than something that consumes less power.  Wrote Coleman: I think [the Turion] is a great chip.

SHARE:
TOPICS: Laptops
8

In reponse to my ongoing coverage of AMD's Turion 64 processor (a 64-bit capable mobile chip for the "thin & light" notebook market), ZDNet reader Bradley Coleman wrote in to find out why notebook manufacturers were choosing to go with the more power hungry of AMD's Turions rather than something that consumes less power.  Wrote Coleman:

I think [the Turion] is a great chip.  Its more feature rich and advanced than the Pentium M.  But, It frustrates me that (as far as I know) every single laptop has been built with the ML version which uses 35W.  Meanwhile there is an MT version that uses only 25W.   See [the page on AMD's Web site that describes Turion model numbers].  This is a bit old, since there is an ML-40 now.  Why do you think [the notebook manufacturers] won't use the better chip?  I have a guess, which you may think is nuts and it may be.  We know that Intel is ruthless and that they defend their strong markets.  They have a lock on the thin and light market, and the MT is the thin and light chip.  Perhaps they are stopping box makers from using the MT?

If you're even reading this, than chances are that you know by now that AMD has filed an antitrust suit against Intel for using a monopoly to stifle AMD's success.  The complaint gives examples of where AMD feels as though its customers (systems manufacturers) were coerced into making product decisions that they may not have otherwise made had it not been for Intel's dominant and leveragable position in the market.  At least one of the examples refers to Acer's plans to support an AMD product rollout and alleges that Intel's then CEO Craig Barrrett "expressed to [Acer] Intel's concern and said Acer would suffer severe consequences if it publicly supported AMD’s launch." The launch was to involve an Acer desktop and notebook.  If Intel engaged in any such coercion (and I'm not saying it did), I have my doubts as to whether it would have been permissive of one of AMD's chip, but not another.  So, while I'm not saying it's impossible, the aforementioned MT conspiracy theory doesn't strike a chord with me.  That said, I asked AMD to comment on the inquiry and here is what AMD spokesperson Jo Albers had to say:

AMD has seen a great deal of interest in AMD Turion™ 64 mobile technology.  Yes, we have a number of customers using the MT series models as well as the ML series.  For example, pan-European supplier Fujitsu-Siemens Computers is offering the AMILO A7645 notebook PC powered by AMD Turion 64 mobile technology MT-32.

Kind of a boring comeback.  I know. More interesting to me, in the bigger picture of the ongoing Centrino "debate," would be lab tests of an AMD notebook (featuring AMD's most advanced Turion 64 processor in terms of power consumption and performance) with an Intel-based WiFi radio versus a Centrino notebook (Pentium M, 855/915 chipset, Intel Radio) ,versus a non-Centrino notebook with the most advanced Pentium M processor/chipset and the best of the non-Intel WiFi radios out there (my fellow blogger George Ou says he has a preference for Broadcom-based devices).  To make it fun, the test would be to see which does the best job of delivering on Centrino's four brand promises: 1) outstanding mobile performance, 2) great battery life, 3) thinner & lighter laptop designs, and 4) integrated wireless LAN capability.  Given the subjective nature of these criteria, the methodologies would have to be worked out.   But now that Intel has been attempting to set the agenda for how people should recognize a system that will serve them well in their road warriorship, the big question is "How well do non-Centrino systems live up to those criteria?" 

Tangentially, riddle me this.  When is a Centrino notebook not a Centrino notebook?

Suppose a new, thin or light Celeron Mobile-based notebook with an Intel radio in it (integrated wireless LAN capability) outperforms an older Centrino-notebook (under Intel's definition of Centrino, any notebook with a Celeron M in it doesn't qualify) and has better battery life?  In other words, it's purely an Intel-based system that lives up to the Centrino promise better than an older Centrino notebook does, but it's not a Centrino.  Hey, but a Centrino the Celeron M system is not.   Perhaps it's time for the Centrino police.  They'd canvas various Starbucks and hotel lobbies and when they spotted an older, not-quite-up-to-current-Centrino-snuff  notebook with a Centrino sticker on it, they'd unholster their sticker scrapers and scream "YOU.  OVER THERE.  WORKING ON THAT THINKPAD. YES. YOU.  SLOWLY, STEP AWAYYYYYY FROM THE NOTEBOOK!"

When asked about this twist, Intel spokesperson Barbara Grimes added a time element to the Centrino promise and said that the Centrino-brand  helps consumers to recognize those notebooks that best live up to the four Centrino promises at any point in time.  She acknowledged that a 2005-class non-Centrino notebook may do a better job of living up to the 2003 vintage Centrino promise than a 2003-class Centrino notebook, but that same rules apply to cars.  For example, a 2005 Volkswagen may have a better saftey rating than a 1970 Volvo.   But that doesn't necessarily mean it has a better safety rating than the 2005 Volvo.  So, much the same way the benchmark for safety to which cars are held is a moving target, so too apparently is the benchmark for the Centrino promise.

Topic: Laptops

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

Talkback

8 comments
Log in or register to join the discussion
  • Chipset preferences

    "my fellow blogger George Ou says he has a preference for Broadcom-based devices".

    In the past, I've preferred Broadcom or Atheros because they offered 802.11g or 802.11 a/b/g when Intel did not. Currently, the Intel wireless chipset has caught up in this regard with their own 802.11 a/b/g offering so I'm fine with the Intel chipset too. Within the next few months, I'm really looking forward to Airgo's True AG chipset with superior speed and range.
    george_ou
    • Turion flavors

      It's quite possible that laptop manufacturers use the ML because it's capable of higher clock rates. Alas, Intel's "GHz marketing" still dominates the business, even for laptops where it's patent nonsense.

      Beyond that, there are a couple of other possibilities. One is that the TDP [1] comparison between the ML and MT oversimplifies too much. AMD's TLP definition is based on flat-out maximum power, but that's not what really affects battery life -- in real use, the ML and MT may not be that different.

      Another is that those round numbers are suspicious. Few things in electronics round out quite so neatly, and the ML and MT TDPs are the same for each regardless of clock frequency. In practice it may turn out that AMD unwisely rounded the MT up by a larger margin. I've been trying to track down the detailed specs for the Turion, but AMD doesn't seem to have made them readily available.

      I did run across Intel's Pentium M power, by the way -- the TDP is 27 watts, but the peak power is actually 33.6 at 1.7 GHz and doesn't include the DRAM controller like the AMD. Since AMD has used peak power as TDP for their other processors, I'm inclined to suspect that the Turion is very close to the Pentium M on power until I see some real specs from AMD.

      [1] Thermal Design Power -- precise definitions vary, but it's supposed to be the power that the cooling system should be prepared to handle.
      Yagotta B. Kidding
      • Wrong button

        Above should have been posted at top level.
        Yagotta B. Kidding
      • Pentium M 1.4 GHz has 10 W spec

        The Pentium M 1.4 GHz low voltage processor has a 10 watt specification. I don't think AMD plays in this market of thin and light notebooks with 12 to 14 inch screens.

        Intel's ultra low voltage products are about 5 watts and are designed for embedded applications.
        george_ou
        • Depending

          [i]The Pentium M 1.4 GHz low voltage processor has a 10 watt specification.[/i]

          If you underclock the Pentium M low enough, you can get the core dissipation down to about 4.5 watts, plus whatever the FSB draws (1.25 W on average, when active.) Of course, that's at 600 MHz and is maximum power, not TDP.

          At 1.4 GHz, Intel's data sheet specs 19 amps at 1.484 volts, plus PLL and FSB termination. That's a bit more than 10 watts, but they may have improved their yields to the point that they can turn down the voltage. If so, they'd have to run the processor at a core voltage of just over a volt to hit 10 watts.

          Keep in mind, as always, that the numbers you're quoting are Thermal Design Power, and as Intel notes in Table 23, Footnote 1, Page 70: "The Thermal Design Power (TDP) specification should be used to design the processor thermal solution. The
          TDP is not the maximum theoretical power the processor can dissipate."
          Yagotta B. Kidding
          • The point is, it's lower

            That's fine, the point is that it is significantly less than the current Turion chips. Intel Pentium M 90nm chips are rated at 5 to 27 watts and AMD Turion is rated at 25 to 35 watts.

            I have a hard time believing that AMD will spec their rating more conservatively. They usually do the opposite and over state their specs. AMD has been known to play fast and loose with their own processor speed ratings.
            george_ou
          • Conservatism

            [i]I have a hard time believing that AMD will spec their rating more conservatively. They usually do the opposite and over state their specs. AMD has been known to play fast and loose with their own processor speed ratings.[/i]

            Oddly enough, AMD's track record on this is just the opposite. The TDP for their other processors is the maximum electrical dissipation of the family. In other words, if the 2.4 GHz processor draws 70 amps at 1.2 volts, the whole series is rated at 84 watts, even though the maximum currents and voltages for the lower-speed parts may be lower.

            What this tells me is that Intel is ahead of AMD in recognizing the marketing value of TDP and is sharpening their pencils while AMD is still shipping paper with conservative engineering data.

            The certain take-home, however, is that they define TDP differently so it's foolish to simply compare TDP numbers if you actually care about power, either for cooling or battery life. That in turn takes us back to DB's complaint that OEMs are holding their test results close to their vests, denying consumers information necessary to make informed decisions.
            Yagotta B. Kidding
          • I have a bigger problem with the inflated processor rating numbers

            They've done this more than once in the past. This was done before the Athlon era, and done when AMD announced the first Athlon 3200+. The first 3200+ wasn't even close to the first Intel Pentium 4 3.2 GHz processor.
            george_ou