Who to believe on power consumption? AMD or Intel?

Who to believe on power consumption? AMD or Intel?

Summary: If we are to believe AMD's claims that Intel uses dubious TDP numbers and has more power hungry chipsets on the motherboard, then we should be seeing a difference far lower than 50 watts and perhaps even a wash. But looking at the highlighted figures, AMD's overall power consumption was actually 66 watts higher than Intel! This (based on TomsHardware numbers) would seem to indicate that not only is Intel NOT over exaggerating their power efficiency numbers, but they're actually being more conservative than AMD.

SHARE:
TOPICS: Processors
106

AMD is on a full marketing counter offensive after the unanimous praise heaped upon Intel's new Core2 "Conroe" desktop processor and are claiming that Intel's low-power ratings are based on a skewed metric.  AMD's reasoning is that AMD's TDP (Thermal Design Power) rating is based on maximum power draw while Intel's power rating is based more on typical power consumption and therefore exaggerated.  Furthermore, AMD argues that we must look at the whole picture by measuring the power consumed on the entire computer.  In fact at last month's press conference at AMD's headquarters, everyone was given a power meter to take home to measure power at the wall.  But does AMD's argument have any merit?  TomsHardware seems to have the answer when they measured power consumption of identical systems with the exception of AMD or Intel motherboards and CPUs.

Intel rates their new C2D Extreme x6800 processor at 75 watt TDP while AMD rates their new Athlon 64 FX-62 Extreme processor at 125 watt TDP which has a theoretical difference of 50 watts.  Since Intel uses an external memory controller on the motherboard, that theoretically translates in to an additional 20 watts of maximum power used by Intel based motherboards so the power advantage should be much smaller than 50 watts if we are to believe AMD.  Here are the actual system level measurements from TomsHardware:


* Represents power save idle mode

If we are to believe AMD's claims that Intel uses dubious TDP numbers and has more power hungry chipsets on the motherboard, then we should be seeing a difference far lower than 50 watts and perhaps even a wash.  But looking at the highlighted figures, AMD's overall power consumption was actually 66 watts higher than Intel!  This (based on TomsHardware numbers) would seem to indicate that not only is Intel NOT over exaggerating their power efficiency numbers, but they're actually being more conservative than AMD.  I threw in an overclocked Intel X6800 at a blazing 3.46 GHz for good measure and even that uses 51 watts less power than the AMD FX-62.

If that wasn't bad enough news, we also see that Intel's mid-end E6600 Core2 desktop processor at 2.4 GHz savaged the 2.8 GHz AMD FX-62 on nearly every single benchmark while costing more than three times less.  Note that these performance numbers are being echoed by every other hardware enthusiast site on the web.  If a 2.4 GHz Core2 can do that much damage, it's hard to imagine what a 3.46 or even 4 GHz Core2 can do.  The Core2 processors are already showing some really nice overclocking properties while AMD's FX line doesn't seem to have much room to clock.  When I issued the dire warning on AMD back in March, I was blasted by AMD fans for posting dubious numbers from a benchmark set up by Intel but it looks like those numbers were legitimate after all.  Considering the fact that even the $224 2.13 GHz E6400 Core2 is giving the AMD FX-62 a run for the money on many benchmarks, AMD is truly in a dire situation and they're under immense pressure to cut prices since Intel has already announced massive price cuts on their legacy Netburst CPUs.  Now with the power argument out the window, it would take a miracle to AMD back in the race.

Topic: Processors

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

Talkback

106 comments
Log in or register to join the discussion
  • Well George...

    Fortunate AMD has the Cool 'n Quit technology.. That sure does make a difference in power consumption ...

    What I didn't find was what happenes when 64 bit software is used.

    BTW 1: Did you notice that every link at Toms site, about the compare, now links to an AMD event?

    BTW 2 (George only): I'm writing a response for your comment about Cox. Hopefully I can finish it today, but otherwise it will be tomorrow..
    Arnout Groen
    • And to make sure

      here's the link...

      http://enterpriseevent.amd.com/momentum/?cid=vibrant&lang=en
      Arnout Groen
    • Not really, that was included in the results

      I included AMD's Cool 'n Quiet technology in the idle scores just like I included Intel's Speedstep technology. It didn't really help AMD all that much.
      georgeou
      • That's funny for not helping much how is it

        That the idle power consumptions are IDENTICAL.
        Tell me, how is that a bad thing? Ever watch cpu usage when running your computer. Your "Big story" is, smaller cores use less power. DUH!!!!! When AMD ramps up the same fab size your story will be AMD FINALLY CATCHES UP TO INTEL. I'll reiterate what I said in another thread. You are an INTEL fanboy. You have a personal INTEL computer from the last generation of chips when you should have been using an AMD, since AMD has been the superior cost and power performer for 2 years. Not Intel will be ahead for.. 4 months, 6 months? Something like that.

        And your last post about the 'benchmarking' was refuted by a third party. You'll always be a WINTEL fanboy based on your bias. (LIke that insane spreadsheet you had to test Open Office.)
        Bill F.
        • Die shrinks don't automatically translate in to a big dip

          When Intel went down to 90 nm, they didn't get an automatic power drop. They managed to make some of their Netburst chips a little better with the drop to 65 nm but it wasn't enough to catch AMD. It took a fundamental architecture shift to turn the tables.

          AMD is now in the same situation. They might be able to bring power down a little and ramp up the GHz a little with the shift to 65 nm, but it's doubtful that will be enough to catch the current Core2 products. To make maters worse, Intel is going to be going to 45 nm soon and they've got lots more room to scale with Core2.

          AMD might have something in the works that might turn the tide, but it will take some time before we know if it beats Intel?s next generation product.
          georgeou
        • Not identical on the extreme products

          "That the idle power consumptions are IDENTICAL"

          Only if you're looking at the slower AMD models. We're talking about the highend models here.
          georgeou
  • Haha..AMD caught in their own gimmick!!

    What a riot. AMD's own gimmick shows that Intel ACTUAL power draw is MUCH better using their own metric. AMD is toast in this department.

    While wall power is a very good meaure, it is not the whole story. For one, power supply efficiency at different loads (idle, full on) contributes significantly to the end result.

    Either way, Intel systems are far coolee and give far better performance against the AMD challenger no matter how you measure this.
    Prognosticator
    • Depends....

      "Either way, Intel systems are far coolee and give far better performance against the AMD challenger no matter how you measure this."

      Yeah right.... NOT! Then explain why AMD has had the upper hand against Intel in the last 2 years.
      Arnout Groen
      • Agree, but..

        Ya need to come out from the past, big guy.

        This is about TODAY's products, not yesterdays. I agree AMD's strategy of comparing Opteron against the 5 year old netburst has been wonderful and I am sure they (you) will continue to do that (well, ya should).

        However, today is today and those that really care about this stuff separate the rhetoric from the facts and the facts today are in Intel's favor.
        Prognosticator
        • I agree as well...

          but you didn't state that in your original reply.

          And because it's my job to be annoying, irritating, frustrating.... (How are you doing George ;-) , I posted a reply)

          you're just experiencing I take my job serious.. :-)
          Arnout Groen
      • Wasn't much of an upper hand

        It's true AMD had a performance "edge" the last three years but it wasn't across the board on all benchmarks and it was never this massive. You never had a situation with low end AMD parts running neck and neck with Intel's flagship product.

        On power consumption, AMD had a significant advantage there on desktops and servers but always lagged on Mobile and still do. Intel just turned the table on power across the board.

        So we have a situation now where where is a BIG gap on:

        Overall performance
        Price/performance
        Performance/watt
        Overall watt
        georgeou
        • Maybe George...

          But then explain these 2 'facts'

          (I coppied them from my reply in Adrian Kingsley-Hughes blog "Conroe puts Intel back on top" )

          1. So I won't have to replace my PC after 3 or 4 years! --> I bought my first PC with AMD K6-3D Now 380 mhz in 1999. I replaced it in 2005. My father bought an Intel PC 400mhz 2 month's earlier and replaced it 3 years later.. I've done a lot of CAD design, using my PC. The last version of AutoCad I could use on my PC was 15. my father's PC stopped at 9, because it refused to run the program.

          2. AMD based PC's are a lot cheaper, then you say it is. The FX-62 4200 is available for $440, while the Conroe is available for $430 (the one from TomsHardware, ask George Ou). Motherboards are cheaper for the AMD processors then for the Intel.

          About point 2 George.. WHERE are you buying your hardware? The prices you name sound like a fraud. And another add on: the prices i mentioned are with a 10% discount...:-)which i didn't add in the original reply.
          Arnout Groen
          • Where in the world do you get your pricing?

            $440 for an AMD FX-62? Do you have a time machine to go in to the future? FX-62 is over $1000.

            Conroe 2.4 E6600 has a list price of just over $300 and it blows away the FX-62.
            georgeou
          • Appology George...

            I made a phone call this morning to the company that offered them... Asking for it... ;-)

            They said they made a mistake with the site.. :-(
            So they've taken it offline...

            It should have mentioned the Athlon 64 X2 4200+, for that price, not the FX 62. They did offer it for 900 euro's though, but that's without the discount. That price is more in line with what you pay.. so an appology is at it's place here:

            Sorry George, I should have verified it earlier on.... Now can you? ;-)
            Arnout Groen
          • And what mistake did I make?

            "Sorry George, I should have verified it earlier on.... Now can you?"

            What error did I make? I'd be the first to apologize if you showed me where I got something wrong.
            georgeou
          • George...

            Read my reply "OK...time for a lucky shot..." about craiglist.

            It's all in there...
            Arnout Groen
          • That Pentium D is slower, so what?

            The Pentium D 940 is a slower CPU, what can I say. The other issue is that certain tasks test better on certain CPUs than others. In a synthetic test that happens to exploit the weakness in a certain design or takes advantage of others, you can see some large differences. But it doesn't necessarily translate to the real world. Even when the FX was dominating the Netburst CPUs, there were certain games that did better on Netburst.

            Now until I see some benchmarks with the Core2 against the FX on real world 64 bit applications, we can only speculate. You can say that the Intel EM64T implementation has some differences, but calling it "emulation" is simply not correct.
            georgeou
          • Ehh George....

            I didn't say it was a Pentium D,

            .....it was a Pentium 2.

            BTW: did you understand what i was trying to say with the last comment I made about Craiglist?
            Arnout Groen
    • Tom's hardware = intel fanboys

      I'm curious to know what tom's was doing for this test. They have the highest wattage tests of any site I've seen, the best I've seen has the Athlon 64 FX62 sitting at roughly 190watts with a single core load, and up to 240 watts with a dual core load. And somehow Tom's got the loads up to 280????????? That's a 40 watt increase, I don't know the specs of the test Tom's used on the AMD system, but if they used the same graphics card as the Intel system (x1900X radeon) then the difference between the 7900GT and the X1900X power draw wise is pretty huge. (the AMD tests were done With out CoolnQuiet as well, leaving the idle wattage sitting at roughly 170watts, still, 20 watts less than Toms' reports, imagine with CnQ on...)
      Also I find it odd that they (tom's) didn't use the athlon 64 X2 5000, the new AM2 dual core top, as it's total wattage use, with full dual core load, was roughly 210.
      Tom's also has some of the lowest Intel wattage reports of any I've seen, leading me to believe that their tests were done as such
      AMD tests, CnQ off, full dual core load
      Intel Tests, Speedstep on, single core load.
      Canadian falcon
  • Are these your numbers, George?

    Who did this test George? You? If so, then like all of all your numbers, they are skewed.

    I want to see numbers from a completely neutral, independent test lab. Until these tests are done by labs which have no contact with either Intel nor AMD, theses numbers mean absolutely NOTHING.

    And you can bet your last dollar that AMD has new chips in the works, so your prediction that AMD is going to disappear, is not going to happen. I know you're prejudiced against AMD processors, but come on, use a little common sense for a change!
    linux_for_me