Phenom Recycled: Athlon X2 7000-Series

Everybody is getting excited about the upcoming Phenom II launch. AMD has already released a bit of information, saying that the 45 nm Phenom II has reached 6.3 GHz cooled with liquid nitrogen, and the pre-release press information indicates that overclocking margins could be as substantial as the potential of a Core 2 processor. However, Phenom II isn't here yet. But AMD {has} just introduced a new Athlon X2 product, which is based on the Phenom's Stars core. We're taking a look at the Athlon X2 7000-series, which brings AMD's dual-core performance up a bit.

The Athlon X2 7000-Series: Phenom Recycled

While AMD's upcoming Phenom II processor represents a new product, shrunk to the new 45 nm DSL SOI process and with more cache memory, the Athlon X2 7000-series is nothing short of the already-well-known Phenom on the bug-fixed B3 stepping. You can tell this by the model numbers, which all end in 50. The 7000 model number most likely was chosen to compete with Intel's Core 2 Duo E7000 series, which includes the same total amount of cache memory. However, while the new Athlon X2 7000 comes with 512 KB L2 cache per core and an additional 2 MB shared L3 cache for both cores, Core 2 Duo E7000 just has its 3 MB shared L2 cache.

The processor die of today's launch is identical to the Phenom X4 and X3, but only two cores are activated. While this is a performance disadvantage for multi-threaded applications, it introduces an advantage in terms of power consumption, as two processing units require less power than three or four. In addition, AMD now has the option to recycle Phenom processors that might have issues with up to two cores. With two cores disabled and re-branded to Athlon X2, the new generation takes on the old Athlon X2 at up to 6400+ speeds, as well as the Core 2 Duo family.

Black Edition Is A Go!

The Black Edition processors (BE) are the most interesting ones, as they come with unlocked multipliers. Hence, it is possible to more easily overclock these. Only the 2.7 GHz top model Athlon X2 7750 is a Black Edition--the two other models are called 7550 and 7450, running at 2.5 and 2.4 GHz. All Athlon X2 7000 processors are at 95 W Thermal Design Power (TDP), although AMD mainstream processors typically stay below that.

Catching Up With Core 2?

The big question is: can the Athlon X2 7000-series catch up with Intel's Core 2 Duo processor family? In the end, the new chip comes with 512 KB L2 cache per core and the Phenom's 2 MB L3 cache, which might provide an advantage. AMD caches are exclusive, which means that contents stored within one level of the cache aren't stored redundantly in a lower level. We'll break the anticipation early: the 45 nm Core 2 family remains faster, but the new X2 7000 still represents an excellent option for users who insist on ideal price/performance ratio, and--more importantly--for upgrade users who already use a socket AM2/AM2+ system.

Create a new thread in the US Reviews comments forum about this subject
This thread is closed for comments
120 comments
    Your comment
    Top Comments
  • Naw-yi
    notice the memory used is 800mzh not 1066mhz, also why would they use different video cards? the drivers used arnt even up to date... test should be run with alike hardware and software(or as close to alike as possible) or its not to be taken seriously.
    14
  • Other Comments
  • davidgbailey
    First Comment!!!

    AMD is competing with Core 2 Duo? Why don't they try to compete with core i7? Oh wait, they can't compete with core i7 being 2 years behind. But AMD FTW against Core 2 Duo 2 years later!!!
    -5
  • curnel_D
    I'd upgrade to the 6400+BE before I'd upgrade to the 7000 series.
    0
  • tipmen
    Poor AMD
    -4
  • royalcrown
    Wow, only 80 bucks to upgrade a four whole fps over my 5400 x2....lemme rush right out, not !
    6
  • urban_black_redneck
    Curnel_DI'd upgrade to the 6400+BE before I'd upgrade to the 7000 series.
    Now the 6000 is just under $80 (same price as the 7750BE) and the 6400 is hard to find.

    I am building my first gaming computer from the ground up, proudly starting with the 7750 Black Edition.
    0
  • amdgamer666
    would like to have seen some overclockability tests but otherwise great article. good to see amd has released kuma (finally) to become slightly more competetive
    6
  • Naw-yi
    notice the memory used is 800mzh not 1066mhz, also why would they use different video cards? the drivers used arnt even up to date... test should be run with alike hardware and software(or as close to alike as possible) or its not to be taken seriously.
    14
  • Naw-yi
    Naw-yinotice the memory used is 800mzh not 1066mhz, test should be run with alike hardware and software(or as close to alike as possible) or its not to be taken seriously.
    -9
  • cangelini
    Naw, I'll ask Patrick to answer the memory questions. However, if you read the Test Setup page, you'll see that only one graphics card was used. Also, given the mainstream nature of these graphics cards versus the very high-end graphics card, there is no way in a million years that the WHQL driver used would have any effect on the outcome of these benchmarks.
    2
  • cangelini
    Naw, ran the memory question past our German team and updated the piece. Same modules were run at 1066 and CAS 5. Thanks for the catch. Hope that answers all of your questions. If there's anything else, please let me know!
    1
  • sighQ2
    ..................................
    A-DATA DDR2-800+ (2x 2GB)
    Setup: DDR2-1066 (CL 5.0-5-5-15-2T)
    VS
    A-DATA DDR3-1600 (2x 2GB)
    Setup: DDR3-1333 (CL 7.0-7-7-20)
    ...................................
    WELL - Perhaps you could do an extensive unbiased review that would SOMEHOW DEMONSTRATE that these 2 mem sets are EQUAL????

    good luck with that :)
    -7
  • sighQ2
    I guess it just wasn't possible to select newer mobos for the AMD setup. Not sure why you used 2 different mobos. One newer mobo with a 790GX/SB750 or 790FX/SB750 chipset, would have been better and would offer EZ-oclox using AOD/ACC.
    Plus the intel boards are newer and FOR A CPU COMPARE TEST it would be better to just use similar ram, so there is no accusation of bias - I bet you would like that.
    But somehow, there's always the 'smoke and mirror' trixxx, = sad.
    Your readers must be idiots.
    -4
  • sighQ2
    Also - in all results, it would be nice if all those cpu's could be listed with CLOCKSPEEDS rather than model numbers so that we could tell that the little AMD is clock for clock blowing the doors off the intel products.
    Perhaps I should be surprised that bigger model numbers are faster. You know, that, I would have guessed. In which case, you could have saved a lot of testing time.
    -11
  • sighQ2
    Have you noticed?

    Whenever I read a test report here, I always have to make posts like this complaining about blatant bias - it's just so obvious to me.

    This gets old and boring for both of us = yes??

    Apparently, your people do not know much about AMD hardware; or, they don't want anyone to know.
    -11
  • sighQ2
    Also lose the synthetic benches - I do not run synthetic benches.
    And please try to avoid cherry picked benches that make intel look good.
    Also try to word things without the obvious slur against AMD.

    A brief review on this cpu here:
    Includes a cpu-z oclox to 3.3 Ghz

    http://www.amdzone.com/index.php/reviews/60/10953-amd-athlon-x2-7750
    -11
  • sighQ2
    Urban_Black_RedneckNow the 6000 is just under $80 (same price as the 7750BE) and the 6400 is hard to find.I am building my first gaming computer from the ground up, proudly starting with the 7750 Black Edition.


    Cool. And you will be able to easily upgrade in a year or two by dropping in any DENEB cpu, whenever you feel like it. (DENEB has dual integrated memory controllers on the cpu for either ddr2 or ddr3 :) and socket AM3 is back-compatible :)
    -5
  • royalcrown
    pPlese ppl ddr 800 or 1066 is gonna make no difference in the real world, and timings schmimings...it's not gonna make the "new" XP's a compelling reason to upgrade.
    AMD needs something that is FAST, not just marginally faster rehashed leftovers.
    2
  • sighQ2
    o THAT reminds me
    In this test report, the power consumption figures are all meaningless because the intel cpu's require power from north bridge re memory controllers which, in AMD the mem controller is internal. So the intel scores are artificially low = add 15-20 watts to include mem controllers.

    More smoke and mirrors. You should have known that too. That's called something worse than bias.
    -11
  • royalcrown
    sighQ2Cool. unless @cangelini: please pass this on, i am typing bling and constantly refreshing and clearing my cach because this page design/forum desigh stinks...using Firefox 3 if the webheads wanna know.I cant really type because the page never loads, my cursor is stuck hard against the left corner and stuff goes left into space when i try to post. This sucks and they need to fix it or junk it !need a new 125-140 watt mb ! ! nd you will be able to easily upgrade in a year or two by dropping in any DENEB cpu, whenever you feel like it. (DENEB has dual integrated memory controllers on the cpu for either ddr2 or ddr3 and socket AM3 is back-compatible
    0
  • sighQ2
    royalcrownpPlese ppl ddr 800 or 1066 is gonna make no difference in the real world, and timings schmimings...it's not gonna make the "new" XP's a compelling reason to upgrade.AMD needs something that is FAST, not just marginally faster rehashed leftovers.


    Well, yes, that is true. But many little biases make for a biased inaccurate review that is not true. So prove your point with a fair test. I know what you mean, but the truth is all I am seeking here, and it is not here!

    The things I am pointing out are common in many socalled tests and reviews. Once you start to see it, YOU benefit.

    C2D might be faster, but not all of them are faster. And in real world computing, something else takes over, and is not about test reports. They don't speak of that either. Lack of stalls, smoother, consistency. Or the fact that a bench test can be cached on first run and reported on second run = faster = lying to make one look better.
    There are a lot of trix.
    -5