Sign in with
Sign up | Sign in

Low-Power Face-Off: AMD's Athlon X2 Vs. Intel's Core 2 Duo

Low-Power Face-Off: AMD's Athlon X2 Vs. Intel's Core 2 Duo
By

AMD, which dominated the processor market with its Athlon 64 and Athlon 64 X2 just a few years ago, was able to make a return with the recent Phenom II processors. While these new 45 nm CPUs cannot trump Intel's Core i7, they offer performance that can match a Core 2 Quad here and there. Unfortunately, there is no low-power version available yet, so we decided to look at AMD’s remaining low-power option.

The latest Athlon X2 dual-core CPUs are very affordable, and supposedly efficient. Our shootout includes an Athlon X2 5050e (45 W TDP) and a 5600+ (65 W TDP) as well as several Intel processors.

Low Power Versus Efficiency

Power users will certainly focus on performance first, while all other users want to have the best of both worlds: a fast processor that does not consume more power than necessary.

Performance is easy enough to define. It is the time required for a workload to complete, or workload execution within a given time. However, efficiency is still sometimes misunderstood, because efficiency does not equal low power. A low-power device such as Intel’s Atom processor has low power consumption, but it is far from being efficient.

We define efficiency as the power that is required to complete a specific workload. In this case, power is measured in Watt-hours (this isn’t watts per hour, but watts times hours). Some readers had noted that work energy should be stated in joules. One joule equals one watt-second, so a watt-hour equals 1 x 60 x 60 = 3,600 joules. We decided to stay with Watt-hours (Wh), as this is the way we are charged for electricity.

CPU Power Versus Platform Power

On the CPU side, AMD has traditionally had two advantages over Intel. First there is the SOI manufacturing process (at 90, 65, and 45 nm), which ensures low power requirements in idle mode thanks to low leakage power. This makes it very much competitive with Intel’s manufacturing, which has typically been 12-18 months ahead of AMD with regards to transistor size (note that AMD processor manufacturing is being outsourced). Second, AMD processors have come with integrated memory controllers since the introduction of the Athlon 64 in late 2003, which typically provides a power consumption advantage over logic that is part of the Intel chipset. Core i7 has started to introduce a paradigm change in this direction with Intel, but that doesn’t apply to Core 2.

So there is the platform, which contributes to power consumption. Especially if you look at all-in-one integrated chipsets such as the AMD 780G, AMD has offered much better 3D performance and power efficiency than Intel chipsets. We used an integrated AMD 780G motherboard as well as an Intel board based on the latest G45 chipset for this article, as well as two representative AMD processors, two Intel dual-cores chips, and two Intel low-power quad core CPUs.

What To Expect?

 The Athlon X2 5050e is a 2.6 GHz 45 W low power part, while the 5400+ reaches 2.8 GHz within a 65 W TDP. The Athlon X2 5050e is a 2.6 GHz 45 W low power part, while the 5400+ reaches 2.8 GHz within a 65 W TDP.

AMD’s Athlon X2 processor is a rather aged product and cannot compete with any of the Core 2 processors in terms of performance—this is a given. So don’t take this article as a performance shootout, but rather as it was intended: a look into the efficiency of AMD’s low power / low budget offerings, since the AMD Socket AM2+ platform still provides a great basis for office systems in the value range. In the end, performance is still many times better than what you’d get from an Atom nettop. In light of $299-399 netbooks, it makes a lot of sense to consider a low cost PC instead, as you can easily stay below $500 even with a reasonable display.

Display all 46 comments.
This thread is closed for comments
  • 4 Hide
    Anonymous , March 31, 2009 6:27 AM
    45nm high k MG vs. already week 65nm plus core 2 archetectual superiortity = no contest, I really wish AMD had a twin on 45nm for notebooks.
  • 0 Hide
    sonofliberty08 , March 31, 2009 6:40 AM
    PCMark only run good on Intel CPU , they found it on VIA chips .
  • 3 Hide
    bodisan , March 31, 2009 9:00 AM
    There is a problem in your rationale for calculating power efficiency. You measured the power needed for the systems to complete a 3D Mark test, and then you compared the results, but you cannot do that because 3D Mark gets whatever it can from a processor or a graphics card over a fixed period of time, so technically the systems do not have the same task to complete. It would have been more interesting to test on video encoding or archiving software, because no matter what, the task there is exactly the same no matter what. And if you would have done so, I suspect the final results would have been even more in favor of the intel processors. Hope I'm not wrong!
  • 0 Hide
    wdmso , March 31, 2009 9:41 AM
    Who cares about power efficiency. its like getting a big Mac large fries with a diet coke and saying your on a diet. unless all your system parts are purchased with efficiency in mind this comparison is point less .
    who buys a e8500 with out a mid to high range graphics card?
  • 5 Hide
    randomizer , March 31, 2009 11:50 AM
    wdmsowho buys a e8500 with out a mid to high range graphics card?

    Someone interested in real work.
  • 3 Hide
    ifko_pifko , March 31, 2009 12:12 PM
    Somehow I don't understand the Average power consumption... I thought that 3D/PC Mark Vantage can use 4 cores... How can the low power AMD consume more on average than Q9550s?
    That AMD system needs less power in idle and in peak... So it seems that comparing peak power consumption is poinless. Or you have a mistake somewhere else.
  • -2 Hide
    Anonymous , March 31, 2009 12:23 PM
    maybe someone like a photographer working with big raw files might buy a e8500 with a low-mid (if they do not use Photoshop CS4) graphics card.

    i was actually thinking of getting an e8400, 4gig ram and a low end graphics card to do photo editing in lightroom, does anyone know whether a q8200 would be quicker (i doubt it)? i live in south africa, our computer part are quite expensive, so a core i7 920 with 3gb ram (including cheapest gigabyte x58 mb) is nearly double what a mid-range gigabyte 775 mb, e8400 (or q8200 - they cost the same here) and 4gb ddr2 ram is
  • -6 Hide
    lamorpa , March 31, 2009 12:33 PM
    wdmso...its like getting a big Mac large fries with a diet coke and saying your on a diet...

    Your on-a-diet. My on-a-diet. What's this 'on a diet' thing? Or did you mean 'you're on a diet'?
  • 7 Hide
    Anonymous , March 31, 2009 12:38 PM
    Why oh why is the Intel E5200 not included here? Anyone even looking at the AMD 5050e would not be comparing it to an E8500 or anything quad core - even if technically at full load those processors could do more work/watt. If you're looking for a truly low-power build, you're looking at the AMD 5050e and the Intel E5200, and you're not overclocking anything. (Maybe the E7xxx series, maybe)
  • 7 Hide
    Anonymous , March 31, 2009 1:20 PM
    In Hungary the AMD 5050E: under 80USD, 5600+: 75USD,
    Intel: E7200: 130USD, E8500:220USD, Q8200S over 300USD, Q9550S over 475USD...
    So what are we compare? Phenom???
  • 7 Hide
    Seikent , March 31, 2009 1:25 PM
    5050e -> $65
    5400BE -> $58

    e7200 -> ~$110
    e8500 -> $180
    etc...
    it's an old story that core 2 duo/quad cpus are more efficient than the Athlon X2 cpus, so if you're going to compare these processors again, use processors in the same price range, because the difference in power usage isn't going to pay the extra investment under normal circumstances (I think).
  • 2 Hide
    Anonymous , March 31, 2009 1:38 PM
    So typical Tom's putting Intel quads against amd dual core. And even high end core duos when the right competition would've been pentiums.
  • 3 Hide
    bvbellomo , March 31, 2009 1:51 PM
    I really wish performance per watt was measured in total system watts, not just CPU watts.
  • 0 Hide
    sublifer , March 31, 2009 2:09 PM
    As long as you're looking at performance per watt and trying to see if these products make sense, you should include a performance per watt per dollar graph and see where everything ends up. Ideally that chart should be relatively flat to indicate that you get what you pay for.
  • 9 Hide
    sublifer , March 31, 2009 2:28 PM
    Okay, quick price check and a little division later:
    (B4B=bang for buck=perforformance per watt per dollar)

    price 3Dmark B4B PCmark B4B
    Q9550s @ $400 0.470 0.705
    Q8200s @ $265 0.657 1.011
    E8500 @ $180 1.094 1.489
    E7200 @ $120 1.583 2.250
    X25050e@ $65 2.846 2.277
    X25400+@ $58 2.672 2.224

    As we all know, bang for buck gets better when you look at cheaper parts. The X2 5050e actually wins bang for buck and the X2 5400 doesn't do too bad either. Because of the graphically better 780G platform the AMD CPUs leave the Intel CPUs in the dust in 3D bang for buck.
  • -1 Hide
    sublifer , March 31, 2009 2:29 PM
    My formatting went to pot but I imagine most of you can make sense out of it.
  • 0 Hide
    JimmiG , March 31, 2009 3:12 PM
    AMD is really hurting by not having any high-clocked, 45nm dual-core CPUs out. On the other hand, it looks like they are trying to position the Phenom II X3 as a direct competitor to the mid-range Core 2 Duos. How does the efficiency of the X3's (and X4's) compare to those same Intel CPUs?

    The Athlon X2's are still fine CPUs for many users, though. I used a 5600+ in a system I built for my mum. It was very cheap and more than fast enough for her needs, and the 780G is a solid platform.
  • 5 Hide
    shortstuff_mt , March 31, 2009 3:34 PM
    Did Tom's Hardware really just compare 65nm AMD chips against 45nm Intel chips? I want the 10 minutes of my life I just wasted reading this garbage back.
  • 1 Hide
    jeffredo , March 31, 2009 4:54 PM
    It would take a long time to make up for the difference in price with miniscule power savings on a CPU - like forever.
  • 0 Hide
    nerrawg , March 31, 2009 6:14 PM
    I have to say that I agree with sublifer's complaint, I mean if the point is either to save money or be green, either way the intel choices don't all seem to be that relative compared to the cheap $60 AMD processors. Anyone considering those AMD solutions now is probably A) on a tight budget and B) doesn't need the system to play extremely demanding games like crysis or run/compile demanding data.

    So what about this for a cheap Intels build that are energy efficient, fast and a similar budget alternative?:

    CPUs:
    E5200 2.5 Ghz wofldale 45nm - $70
    E1400 2.0 Ghz allendale 65nm - $50

    Mobos:
    ASUS P5KPL-CM LGA 775 Intel G31 - $55
    GIGABYTE GA-G31M-ES2L LGA 775 Intel G31 - $53
    GIGABYTE GA-EG41M-S2H LGA 775 Intel G41 - $75
    ASRock P43Twins1600 LGA 775 Intel P43 - $77

    And if you want a GPU that uses passive cooling and little power:
    SAPPHIRE 100252HDMI Radeon HD 4550 512MB 64-bit GDDR3 PCI Express 2.0 x16 - $50
    POWERCOLOR AX4650 512MD2-S3 Radeon HD 4650 512MB 128-bit GDDR2 PCI Express 2.0 x16 - $70
    SAPPHIRE 100255U Radeon HD 4670 512MB 128-bit GDDR3 PCI Express 2.0 x16 - $70

    So I was wondering if these would possibly make better comparisons for someone looking in the price range of $60 CPUs?

    However if you guys really just intended this article to be all about CPU efficiency then I understand the choice of more expensive Intel chips, its just that AMD really doesn't have anything yet (like you mentioned) to compare with low energy Intel offerings like the Q9400S at only 65W rating. Hope for AMD is still there as I have heard rumors (can't remember where now) that the newer phenom II's quads are going have 940 similar clocks (around 3.0GHz) but 810 similar power rating (95 instead of 125W) and maybe some 65 watters? Would be cool to hear the low down on this.
Display more comments