• What
    is this?
    You've landed on the AMD Portal on AnandTech. This section is sponsored by AMD. It features a collection of all of our independent AMD content, as well as Tweets & News from AMD directly. AMD will also be running a couple of huge giveaways here so check back for those.
    PRESENTED BY
POST A COMMENT

58 Comments

Back to Article

  • ExodusC - Thursday, April 08, 2010 - link

    This is overdue, but a great thing. Reply
  • pjconoso - Thursday, April 08, 2010 - link

    We're now seeing results from Intel's settlement with AMD :). Is this the one rumored to be $200-300? I want one! Reply
  • wicko - Thursday, April 08, 2010 - link

    So I take it the cores are only overclocked when other cores are not in use? I guess with 6 cores you're bound to have some of them idle. Reply
  • ET - Thursday, April 08, 2010 - link

    The article says twice "3 or fewer cores idle", when it clearly means "3 or more cores idle". From the description, it looks like it's actually just "if there are 3 idle cores, then 3 turbo up". Reply
  • InternetGeek - Thursday, April 08, 2010 - link

    Here's an easier way to read it. The number of idle cores determines the number of cores that will be accelerated if there are more than 3 idle cores. So for 3 idle cores, 3 are accelerated; for 4 idle cores 2 are accelerated and so on. What this means is the more idle cores there are the more the active cores can be accelerated, up to 3.5ghz.

    What I also understand is that there won't be any power saving because regardless of which cores are idle or not, all will be overvolted. Must be because they all share the same power plane.
    Reply
  • JumpingJack - Saturday, April 10, 2010 - link

    This is not they way I understand it, but I could be wrong.... if 4 cores are idle, 3 cores still get accelerated, one is simply not used. Voltage goes up to all cores as well, so this chip will run as hot as 6 cores loaded, or maybe even hotter -- assuming this is indeed how it works.

    Power consumption will be an interesting thing to observe when this chip comes out.
    Reply
  • neoflux - Thursday, April 08, 2010 - link

    Yeah, the slide clearly says "if 3 or MORE are idle", not fewer.

    Also, the article makes it sounds like it's either 3 boost or none boost, whether or not it means that.

    According to the slide it's either 3 idle = 3 boost, 4 idle = 2 boost, or 5 idle = 1 boost. Not as effective or efficient as Intel, but better than nothing.

    PS- would it be terribly hard to keep the article showing along with the commenting box or even the other comments? Hard to give detailed comments when you can't see what you're replying to without opening article again on another tab.
    Reply
  • Anand Lal Shimpi - Thursday, April 08, 2010 - link

    Wow, long night, fixed :)

    And yes we have a lot of changes coming for the comment engine, they just didn't all make the cut for the launch list but we'll be adding things as we go along.

    Take care,
    Anand
    Reply
  • Carleh - Thursday, April 08, 2010 - link

    Would it be possible to add some left and right margins to print page layout? I know it's meant to be printed, but I guess a lot of readers use it to read the whole article at once (me included), and it is slightly inconvenient to read without margins. Thank you. Reply
  • Barneyk - Thursday, April 08, 2010 - link

    Im really impressed by AMD and how they stay competitive with Intel even though Intel has such a huge lead right now.
    To bad they're not competitive enough to actually bring the prices down more on Intels mid-range to high-end levels.
    Reply
  • gruffi - Thursday, April 08, 2010 - link

    "This isn’t nearly as elegant of a solution as Intel’s turbo."

    What a funny joke. What AMD does is more clever, more effective and less complex. And that means it IS more elegant.

    133 or 266 MHz more for an i7 920. That's impressing. And who the hell had the idea to overclock when all cores are under load? Yes, Intel's turbo is a joke.
    Reply
  • silverblue - Thursday, April 08, 2010 - link

    If you really want to judge Intel's Turbo Boost, do so with the Lynnfield implementation. Reply
  • gruffi - Saturday, April 10, 2010 - link

    Say that to people who purchased a Bloomfield system. Even Lynnfield's implementation is not good, just better than Bloomfield's. Reply
  • fitten - Thursday, April 08, 2010 - link

    You have to be completely blinded by fanboi religion to think AMD's solution is "more elegant". Reply
  • MySchizoBuddy - Thursday, April 08, 2010 - link

    can the intel fanboy care to explain how one is better than the other. Reply
  • fitten - Thursday, April 08, 2010 - link

    I'm guessing you aren't talking to me, seeing as how I own more AMD based machines than Intel ones. Reply
  • Samus - Saturday, April 10, 2010 - link

    Where have I heard that line before? Geez. Reply
  • Jaybus - Thursday, April 08, 2010 - link

    Hard to say which is better. Current OS schedulers are designed for symmetric processor cores. Threads with a higher priority are more likely to get assigned to a core, but there is no way to tell the scheduler to assign certain threads to the faster cores. A thread's core affinity can be set, but there is no way to know if the core it is set to will be one of the fast ones. So AMD's approach is less useful than it could be for the time being. However, schedulers that can handle asymmetric cores are coming, since they will certainly be needed for scheduling GPU cores that may be on different video cards or integrated on the CPU silicon and will differ in performance. Only benchmarking will tell if one boost method is better. Most likely, one will be better on one OS and one on another OS because of differences in the schedulers. Reply
  • hyc - Thursday, April 08, 2010 - link

    You're missing the point. The OS doesn't need to care which core it should assign to; the core will boost automatically if it's needed and there are sufficient idle cores. Reply
  • nichow03 - Friday, April 09, 2010 - link

    Actually he has a very good point. With the AMD approach there may be times when it is more efficient for the scheduler to only schedule on 3 cores instead of 6 to initiate turbo.

    Image a theoretical app/game that uses three cores but no more. The scheduler could run a low priority thread (say the zero page thread in Windows) on the forth core, preventing turbo mode and slowing down the app/game you care about.
    Reply
  • strikeback03 - Friday, April 09, 2010 - link

    Yeah, I mean who could possibly want more speed when you are already using all four cores? *rolls eyes* Reply
  • ash9 - Thursday, April 08, 2010 - link

    seems to me this speaks volumes to their 'M-Space' designs Reply
  • Pessimism - Thursday, April 08, 2010 - link

    Officially branded the Phenom II X6, AMD won’t be launching these processors until some time in the future.

    I stopped reading there.
    Reply
  • Dzban - Thursday, April 08, 2010 - link

    No need to. It is supposed to launch 24th of April. Reply
  • phenom_x8 - Friday, April 16, 2010 - link

    WHy there is still no review about this new processor?? Anand, ha you ever check that in my country, it have been around for quite some time, especially the 1055T and 1035 T since 12th of April. A lot of people also sell it on many forum only for $218 . This is the link http://www.kaskus.us/showthread.php?t=3876654 (Sorry, its in Indonesian)
    REVIEW PLEASE, IMPORT IT IF YOU HAVE TO (Dont wait for any sample from AMD)
    Reply
  • phenom_x8 - Friday, April 16, 2010 - link

    WHy there is still no review about this new processor?? Anand, have you ever check that in my country, its have been around for quite some time(12th of April), especially the 1055T and 1035 T .
    A lot of people also sell it on many forum only for $218 . This is the link http://www.kaskus.us/showthread.php?t=3876654 (Sorry, its in Indonesian)
    REVIEW PLEASE, IMPORT IT IF YOU HAVE TO (Dont wait for any sample from AMD)
    Reply
  • vol7ron - Thursday, April 08, 2010 - link

    This is interesting - I've never purchased an AMD CPU, but I might give it a try.

    I'm curious why AMD decided that the voltage needed to go up. If some cores are not being utilized, then the current should flow to the used cores. Is this what you mean by power gating?

    Also, my memory does not serve me as well as it used to. With all these new processors, I thought it was advised to get three sticks of RAM for maximum performance. I'm not sure if that was catered to the 3 or 6 core CPUs, or if it had something to do with the North Bridge. Perhaps someone could clear this up for me?

    Thanks,
    vol7ron
    Reply
  • Taft12 - Thursday, April 08, 2010 - link

    The 3-sticks-of-RAM floating around in your head is from the triple-channel architecture of Intel's socket-1366 Core i7 platform. Barely any benefit over dual-channel, but 6 slots in the motherboards are very nice! Reply
  • Zaitsev - Saturday, April 10, 2010 - link

    "I'm curious why AMD decided that the voltage needed to go up. If some cores are not being utilized, then the current should flow to the used cores. Is this what you mean by power gating?"

    That is exactly what intel does, and I think that IS the reason AMD can't use the same method. You know Intel has patents on the powergate tech.

    But you're right, it makes more sense to decrease voltage to the un-used cores.
    Reply
  • ltcommanderdata - Thursday, April 08, 2010 - link

    Maybe this will finally push Intel to release mainstream Gulftown's either 6 cores or harvested quad cores to replace the Bloomfield Core i7. Bloomfield's 1 or 2 multiplier Turbo Boost and consistent high 130W TDP even on the lower-end models isn't very impressive compared to Lynnfield and now Thuban. Reply
  • rickcain2320 - Thursday, April 08, 2010 - link

    Anything less than Intel's exorbitant $999 I suspect, but here's hoping its $299 or less. I can't wait. Reply
  • frozentundra123456 - Thursday, April 08, 2010 - link

    Glad to see that there are several chips in the line up instead of just a super high end like intel did.
    Hopefully the price of at least some of the models will make them practical for mainstream use.
    Maybe this will force intel to release more hexa cores too.
    Reply
  • Eeqmcsq - Thursday, April 08, 2010 - link

    Good to see AMD bring the Turbo feature to their lineup. Can't wait to see these CPUs trickle down into even lower price segments later on.

    But moving forward, now both Intel and AMD will have variable clocked CPUs. Benchmarking and determining which CPU performs "better" will be even more complicated compared to constant clocked CPUS. On a variable clocked CPU, performance will actually drop when more tasks are being run due to the CPU clocking down from turbo, whereas on a constant clocked CPU, performance remains constant as long as there are cores available to complete the current tasks. Just more things to check for when picking a CPU.
    Reply
  • Smoked Tails - Thursday, April 08, 2010 - link

    It doesn't make evaluating and comparing any more difficult than it used to be. I remember back when reviewers had to double check the operating speeds of motherboards for reviews simply because some manufacturers would squeeze in a few extra MHz here and there.

    For this I can see tweaking the suite to include things to compare the performance of Intel's vs. AMD's implementations. But, I'd like to see the majority of testing with it off. That's just me being odd I guess. I think it's easier to quantify performance/watt or $ that way. Although, keeping the focus off of the turbo could potentially eliminate methods of benchmark tampering that ATi and Nvidia used to use. I'd hate to see a driver update that disabled two cores for Vantage so that you get 4 cores at 3.x GHz.

    Anyways, just my two cents.
    Reply
  • TranceCode - Thursday, April 08, 2010 - link

    Could we have a comment button to 'mark spam' please?
    A way for the community to help remove comment spam would be a great feature.
    Reply
  • JarredWalton - Thursday, April 08, 2010 - link

    The admins have that feature, and when used three times on an account the account should be deactivated (as well as deleting the spam). Reply
  • silverblue - Thursday, April 08, 2010 - link

    2 extra cores will help in situations where you can throw as many cores as you like at a task. We may also get to see what keeping the L3 cache at 6MB will mean for performance; probably very little unless AMD has tinkered around with the latencies (unlikely). I doubt it'll perform better at the same speed as a current X4 on a core-per-core basis.

    My concern lies with a) software that utilises exactly 4 cores, and b) the deviation from the power-of-2 system. It's entirely feasible that the Phenom II X4 965 BE will outperform AMD Phenom II X6 1090T in either of these situations due to its higher core speed plus the fact that the software won't scale up from 4 to 6 cores. For anything that doesn't tax all four cores of the 965 BE, or has no limit on the number of cores it can use, the 1075T and 1090T will generally win out. So, AMD have seemingly tackled both ends of the spectrum but due to the fact that the CPU itself doesn't have a number of speed bins for each number of active cores (like Nehalem), there's no middle ground because there's no graceful drop in speed when four or five cores are being taxed. 333 and 166MHz boosts for 4 and 5 cores respectively would've been better in my opinion, however perhaps AMD thinks that if an application uses all four cores on an X4, it will gratefully utilise the other two on an X6. We'll have to wait for the benchmarks to see if this was a good approach or not.

    As it is, this is what we're getting, and it should generally be a better product than the current X4s as well as the early Nehalems. Watch the i7-920 and i5-750 fall in price at the very least.
    Reply
  • smokedturkey - Thursday, April 08, 2010 - link

    "AMD won’t be launching these processors until some time in the future"...

    ya think? ;o)
    Reply
  • DarkKnight - Thursday, April 08, 2010 - link

    Will this work on am2+ motherboards like the Phenom II X4? Reply
  • DarkKnight - Thursday, April 08, 2010 - link

    nevermind, just read the 2nd last sentence. Reply
  • rns.sr71 - Thursday, April 08, 2010 - link

    I read early reports(...rumors?) that the HT link speed(and maybe NB) of the desktop hex-cores would be a tad higher than the quads- 2400mhz vs 2000mhz. i know a person can overclock em that much on there own, but if the STOCK speeds are higher on the HT/NB within the same tdp then you're better off. that alone would give a nice and meaningful boost in preformance at the same CORE clock speed.
    has anyone heard any reliable info on this?
    its too bad that denebs(and all variants) NB speeds is so slow and only 64 bits wide- see this link: http://it.anandtech.com/show/2978/amd-s-12-core-ma...
    amd needs to fix this with bulldozer. i'd like to see 128 bit wide NB running no slower than whatever the HT link speed will be. maybe 2400-2600mhz.
    Reply
  • formulav8 - Thursday, April 08, 2010 - link

    The Northbridge is still 2ghz. The report about higher NB was on wikipedia and has since been debunked by Hexus I think?? It was some site that apparantly had some internal info.

    I hope they do increase the nb speed of course, but its not very likely.

    Jason
    Reply
  • rns.sr71 - Thursday, April 08, 2010 - link

    thats lame. NB speed is one thing thats holds back amd cpus right now. it will be even more noticeable with six cores(provided that they are in use) trying to use the NB and the cache. plus any bump of the NB speed, provided it stays within the tdp limit, will help make up for the fact that they didn't increase L3 cache size. they really need to work on the speed. Reply
  • nemetsk - Thursday, April 08, 2010 - link

    It's been years since multi-core was intro'd, and it will still be MANY years until software is properly threaded (read: tailored) for all these cores. When they went multi-core "because they couldn't clock the core any higher" it sounded like a classic cop-out to real innovation. Seriously? Ya know the French are still working on cold fusion.... I hope someone is still working on a single core :)

    My advice: buy a value-priced 2-core cpu and add an ssd.

    Peace.
    Reply
  • ash9 - Thursday, April 08, 2010 - link

    Let me see if I get this right
    Turbo boost---------------- Intel--------AMD
    Activated by------------------ O/S--------CPU
    Cores affected-------------- 2------------ 3
    logically it makes sense that minimum is 3 cores –Asus’s new Phenom II X6 motherboards have “Core Unlocker” suggesting that Phenom II X3 processors derive from X6’s, meaning (logical deduction) that core 3 cpus will be include turbo boost.

    seems logical to me
    asH
    Reply
  • rns.sr71 - Thursday, April 08, 2010 - link

    no. Phenom II X3 processors derive from PhenomII x4's- one core disabled. the x6's are a new cpu for the desktop that are drived from a 6 core server cpu. Reply
  • ash9 - Thursday, April 08, 2010 - link

    I'm suggesting a core 6X with 3 disabled cores or more

    New Asus Motherboards Unlock Latent AMD CPU Cores
    http://www.techtree.com/India/News/New_Asus_Mother...

    New M4A89GTD PRO Series motherboards are offered in two versions and the only difference between both almost identical motherboards is the USB 3.0 port support. One of the distinct features of these motherboards is the new Core Unlocker switch that activates up to four latent AMD CPU cores of the AMD's triple-core and quad-core processors. Remember the days when BIOS hack unlocked fourth core in AMD Phenom II X3 processor.

    activates up to four latent AMD CPU -
    the only way to ACTIVATE UP TO 4 LATENT - is if 4 are latent , then you can activate up to 4 cores

    If this 6X design can compete within a reasonable rate to the i7's then why not disable 4 cores for 2X, 3 cores for 3X - hence-activates up to four latent AMD CPU, or unlocking 1 of 4 can activate Tturbo Core.
    After all, 3 cores can be turbo boosted compared to Intel's 2

    deduction my dear Watson
    (I could be wrong, but the logic appears sound, in light of the current information)
    asH
    Reply
  • Schmich - Friday, April 09, 2010 - link

    On one of the slides it says that Turbo CORE works with ALL am3 motherboards. Does that mean that it does NOT work with all am2+ boards? Reply
  • futurepastnow - Saturday, April 10, 2010 - link

    If it doesn't, they've lost a sale from me. Reply
  • spigzone - Monday, April 12, 2010 - link

    All AM2+ boards with an updated Bios?

    Two of my 4 AMD AM2+ motherboards have an Thuban Bios update currently available, two do not. Yet.
    Reply
  • jretter07 - Sunday, April 11, 2010 - link

    The leakage of an idle core is 2-3W at nominal voltage. In Turbo mode, voltage is increased for all cores, so the leakage of the three clock gated cores could reach 5W each. That's just wasted power. I would rather go for Intel's solution, where idle cores don't run up my electricity bill.

    -J
    Reply
  • qwertymac93 - Monday, April 12, 2010 - link

    IIRC amd cpus have the ability to lower voltages to cores individually. while the whole cpu would see a slight voltage increase, the idle cores would still have lowered voltages(.85v?) Reply
  • stardude82 - Sunday, April 11, 2010 - link

    The K10 is way behind Nahelem on a per clock basis (and behind Core also). If you look at the 975 and 980x benchmarks, the only thing that the extra core helps with is encoding. So after a year and a half, AMD finally has an answer to the 920. Reply
  • TheFlyingSquirrel - Sunday, April 11, 2010 - link

    AMD's great support for each socket is what keeps me with them at the moment
    performance great for the price and a plethora of CPU's
    can always count on AMD for a great value
    Reply
  • apriant - Thursday, April 15, 2010 - link

    i buy phenom II x4 965 few days ago,n these 6's will came out soon..damn..
    but thx to amd with their supported socket am3/am2..thats how i loved amd :D
    Reply
  • phenom_x8 - Saturday, April 17, 2010 - link

    WHy there is still no review about this new processor?? Anand, have you ever check that in my country, its have been around for quite some time(12th of April), especially the 1055T and 1035 T .
    A lot of people also sell it on many forum only for $218 . This is the link http://www.kaskus.us/showthread.php?t=3876654 (Sorry, its in Indonesian)
    REVIEW PLEASE, IMPORT IT IF YOU HAVE TO (Dont wait for any sample from AMD)
    Reply
  • warmon6 - Friday, April 23, 2010 - link

    im guessing they already have the cpu and benched-mark it, although the reason why they posted nothing is because AMD has tech web sites like Anandtech, toms hardware, ect are on a NDA with them.

    So even if they have the stuff ready to post, that NDA prevent them posting the info up. Breaking the NDA is something they wont do as it will create many problems like not receiving samples of cpu's.
    Reply
  • Wolfpup - Friday, April 23, 2010 - link

    Wow, wasn't expecting to see 6-core CPUs from AMD so fast. These could actually be fairly interesting and competative. I mean probably Intel will just introduce cheaper 6-core CPUs, but right now at least a 6 core one with six less powerful cores might still be intersting versus a 4-core with more powerful cores.

    At any rate, it's great to see.
    Reply

Log in

Don't have an account? Sign up now