AMD’s Phenom II X4 980 Black Edition processor

AMD has taken to doing these 100MHz “speed bumps” lately, and frankly, it doesn’t exactly make for compelling theater. When a processor clocked north of 3GHz gains 100MHz and replaces the prior model at the same price, pretty much nada shows up on the ol’ seismograph we have hooked up to the CPU landscape. Yes, AMD just grew little more competitive—or, perhaps more aptly, a little less uncompetitive—versus the Intel juggernaut.

And yes, AMD needs to claw out every inch of space it can muster in this competitive environment, so adding another 100MHz whenever possible arguably makes sense.

Forgive us for not being more excited about watching a two-year-old, 45-nm quad-core adding a few percentage points worth of clock ticks each second. We’re just pretty sure we know what will happen when that puppy steps into the ring with the Sandy Bridge buzz saw: small chunks of Deneb silicon will be flying in every direction in a fine, black mist.

With that illustrious introduction, we give you the Phenom II X4 980 Black Edition CPU. With four cores clocked at 3.7GHz, 6MB of L3 cache, and a 125W TDP rating, the X4 980 is a direct successor to the Phenom II X4 975 Black Edition. The X4 980 deviates slightly from the blueprint by adding 100MHz and coming in at $185, or $10 less than the prior top-end Phenom II X4. AMD tells us the X4 980’s debut will push the the X4 975 down to $175, the X4 970 down to $155, and the X4 965 down to $135. We presume the senior member of the group, the X4 955, will have its case heard before an ObamaCare cost-reduction panel. Say your goodbyes now.

All of those members of the Phenom II X4 900 series, including the 980, are Black Editions, which means you get to picture them wearing trendy, black turtlenecks like Steve Jobs. Really, that’s the major perk. The other, more minor perk? An unlocked multiplier, intended to make overclocking as free and easy as a circa-2005 home loan.

We’ve run the X4 980 through our CPU test suite—you can see the full setup documented right here in our Core i7-990X review—and we figure we might as well drop the performance summary on you immediately. In fact, we’ve compiled an overall performance index based on our test suite, and we’ve mashed it up with the latest pricing from AMD and Intel, for a single image that tells the value story for the current desktop CPU market. Behold:

The move from 3.6GHz to 3.7GHz grants the X4 980 about the sort of advantage you’d expect versus the X4 975—that is, not much. We’ll refer you to our Sandy Bridge review for a full set of test results. Just imagine the X4 980 as a smidgen quicker than the X4 975 in any benchmark we ran.

Even among AMD’s own CPU offerings, the X4 980’s value proposition looks rather raw next to the Phenom II X6 1075T, which also participates in the price cuts today. The six-core 1075T has the same 125W TDP as the X4 980 and, thanks to the Turbo Core feature that the X4 980 lacks, the 1075T’s peak clock speed is 3.5GHz. The 1075T will drop down to its base frequency of 3GHz when all six cores are busy, but its combination of more cores and flexible frequencies gives it a clear lead over the X4 980 in our overall performance index. That tells you something of the X4 980’s flavor; products with lower core counts generally offer higher peak frequencies for the same price, but AMD’s Turbo Core has rendered the X4 900-series’ appeal suspect.

Meanwhile, the $184 Core i5-2400 offers substantially higher overall performance than the X4 980 or the X6 1075T—for the same money, and with a lower 95W power envelope. If you look beyond the overall summary, you’ll find that the Core i5-2400 generally outperforms the X4 980 in workloads that are both lightly and heavily threaded.

Gaming performance

Rather than encapsulate everything into a performance summary and move on, we’ll make a brief tour through our gaming benchmarks, so you can see the practical impact of the X4 980’s faster frequency.

Yep, we’re talking about a frame per second of difference, or sometimes a fraction of that. Our Bad Company 2 test, which involves logging of frame rates during a manual gameplay session, has enough variability that the X4 975 finishes ahead of the X4 980 in the overall average. At the end of the day, the difference between these two processors is imperceptible in today’s games. The only good news here is that both of them can run these games at more-than-acceptable frame rates.

Of course, for gaming, the $117 Core i3-2100 is arguably faster than either of them.

Power consumption and efficiency

The X4 980’s TDP rating of 125W puts in firmly in a higher weight class than its most direct price and performance competitors from Intel, all of which are 95W parts—despite the fact that only the Intel processors have a built-in GPU. Nevertheless, power ratings and true power consumption are tricky and can differ substantially in some cases. Here are our results for the X4 980 against, well, nearly everything.

Our X4 980-based test system draws roughly 20W more when idling and 60W more while rendering than a comparable Core i5-2400-based system. Power draw under load has risen only 7W versus the Phenom II X4 975, but AMD wasn’t exactly winning this fight prior to the X4 980’s arrival.

This puppy isn’t going to win any power efficiency awards for this rendering workload, either. Even the old Core 2 Quad Q9400—which, after all, was fabbed on a fairly similar 45-nm process—is more efficient than the X4 980. Only AMD’s hexa-core Phenom IIs can come close to Intel’s best 45-nm chips, because those Phenoms don’t have to live so far up the voltage/frequency curve.

And then there are Intel’s 32-nm chips, which occupy the top six spots.

Overclocking

Our copy of the Phenom II X4 980 happily overclocked by 500MHz, to 4.2GHz, at 1.5V—only a tenth of a volt above stock. When we tried to push further, our chip was a bit of a tease. It would POST and boot into Windows without complaint at 4.4GHz, but one of the four cores consistently threw an error in our Prime95 torture test. We tried incrementally cranking up the voltage, all the way to 1.65V, but that didn’t help. Despite a tantalizing glimpse at 4.4GHz, we had to settle for 4.2.

The additional 500MHz improves the X4 980’s performance to the point where it gives the Core i5-2400 a true challenge—but the Phenom needs the help of overclocking to get there. Even with the frequency boost, you’d still be better off in most of these apps (which are nicely parallelized) armed with the two additional cores of the Phenom II X6 1075T.

Conclusions

I’m trying to think of an appropriate prospective buyer for the Phenom II X4 980, but I can’t really come up with one. If you are the owner of an existing Socket AM3 system looking for an upgrade in this price range, you really should go for the Phenom II X6 1075T. Yes, the 1075T’s base clock is a little lower, but two more cores and a peak Turbo Core frequency just 200MHz shy of the X4 980’s speed easily make up for that—and the 1075T drew less power under load in our tests.

If you are dying for more single-threaded performance—or if you just want the best all-around CPU—then one of Intel’s Sandy Bridge offerings near $200 would clearly be the better choice. The Core i5-2400 costs the same as the X4 980 and handily outperforms any desktop processor AMD makes, while at about $30 more, the Core i5-2500K is even faster and will let you get your OC on with a fully unlocked upper multiplier, just like AMD’s Black Editions. In between those two is the non-K version of the Core i5-2500, with the same performance as the K-series CPU at $205. We’d pony up the 10 bucks for the unlocked K series, but if you’re really pressed for cash, Intel’s model lineup has you covered. All of these Sandy Bridge processors are more power-efficient than the X4 980, too.

If you’re bound and determined to buy an AMD processor and are looking for a good value compared to the Intel incumbent, we’d suggest looking at the Phenom II X4 840. Yes, the Core i3-2100 is faster in our overall performance index, but the X4 840 has four true cores to the i3-2100’s two, so the Phenom is quicker in many widely multithreaded applications. That chip more than any other today represents AMD’s traditional position as a value alternative to Intel.

All of which leaves us wondering what, exactly, is the point of this little product refresh. Yes, AMD’s top quad-core product is a tiny little bit improved over the prior model, but even in the small picture, nothing much has changed at all.

Update on 5/3/11: AMD provided us with incorrect pricing information prior to the initial publication of this review. Shortly thereafter, AMD revealed a lower (by $10) price for the Phenom II X4 980 and complementary reductions across much of its Phenom II lineup. We have revised the text and value scatter plot to reflect the new pricing in this updated version of the review.

Comments closed
    • jimmy900623
    • 8 years ago
    • Damage
    • 9 years ago

    Update on 5/3/11: AMD provided us with incorrect pricing information prior to the initial publication of this review. Shortly thereafter, AMD revealed a lower (by $10) price for the Phenom II X4 980 and complementary reductions across much of its Phenom II lineup. We have revised the text and value scatter plot to reflect the new pricing in this updated version of the review.

    • potatochobit
    • 9 years ago

    the sandy bridge only costs less…. when you don’t include a motherboard

    • ew
    • 9 years ago

    Right now I have an AM2+ motherboard with an Athlon x2. I’m planning on upgrading to a Phenom II x4 or x6 once Bulldozer is released. Used Phenoms should be dirt cheap at that point. Upgrading to Sandy would be great but I just can’t justify the cost.

    • Delphis
    • 9 years ago

    Only came to say that I love the macro close-up title pic, very creatively shot.

    • eitje
    • 9 years ago

    Is there a high-res version of that first pin pic that you could link? 🙂

      • Damage
      • 9 years ago

      Wow, the pins are popular. I’ve added an image gallery with a number of hi-res images.

        • SomeOtherGeek
        • 9 years ago

        Well, cool pix get a lot of responses. And of course, thanks!

        • matic
        • 9 years ago

        Got hooked by the pins pic on the home page, gone through the interesting review and finally got my wallpaper for my new Ubuntu install. Thanks!

        • eitje
        • 9 years ago

        That 6th pic is SAUCY! awesome.

          • Damage
          • 9 years ago

          Thanks! That’s the EFS 17-85 with a broader focal plane. Can’t get quite as close with it, but it still gives good pin.

        • Bensam123
        • 9 years ago

        It’s the highlight of the article. :p

    • ronch
    • 9 years ago

    I’m sure AMD is itching in the behind to get Bulldozer out for the longest time. But they simply can’t yet. I hope BD is in good shape. Heck, if I was dying I’d fight for my life just to see the day BD comes out with benchmarks.

    • Silus
    • 9 years ago

    Pretty pointless product, but then again, this is AMD’s strategy for the high-end for a while now. Increase frequency of their fastest processor and hope for the best! Bulldozer either delivers or AMD’s CPU division is in a world of hurt.

      • OneArmedScissor
      • 9 years ago

      If AMD did not even offer Bulldozer for desktops, I don’t think they’d exactly be in a “world of hurt.” High end desktop CPUs are an extremely small part of the market that exists for pretty much no reason other than allowing both Intel and AMD to “beta test” their server parts with little risk.

        • Silus
        • 9 years ago

        Well, you got the point: the server market.
        If Bulldozer doesn’t deliver, their server products will not be able to make a dent on Intel’s market share in either the desktop and more importantly, in the server market. And that’s where the world of hurt comes in.

          • swaaye
          • 9 years ago

          Yup. Considering Bulldozer is primarily targeting the server market with its new architecture one can only hope that it does deliver there because it’s probably not going to be that neat for the other segments. Can’t wait to see how it goes.

          • ronch
          • 9 years ago

          I tend to agree. But additionally, no matter what, AMD’s and Intel’s high end processors will determine not only the battle in the high end, but eventually in the midrange and low end as well, as those high end processors will eventually trickle down the market segments. K10 initially debuted in the high end server market but right now AMD is only second best in the midrange market (and can’t even reach the high end) primarily because its former high end product (Barcelona) wasn’t as strong as Intel’s Core architecture even when it came out. AMD can’t come out with new architectures every two years like Intel does, so when it comes out with a new architecture, it better be really powerful, as they probably would stick with that architecture for a longer time than Intel does with their own. All the power and performance they introduce with a new architecture will eventually be ‘inherited’ by all the lower product lines later on, and that will either bolster or kill their entire lineup.

      • ronch
      • 9 years ago

      What this really means for us is AMD is dropping prices across the board. Who wouldn’t love that?

    • neon
    • 9 years ago

    bottom of page 4:
    [quote<]the two additional cores of the Phenom II X4 1075T.[/quote<] should refer to Phenom II X6 1075T. Like the macro photography on the pins 🙂

    • anotherengineer
    • 9 years ago

    Nice close up pic of the pins.

    • dpaus
    • 9 years ago

    Why is this review?

      • ColeLT1
      • 9 years ago

      Why is this review (blank)?
      I’m dying to know!

        • dpaus
        • 9 years ago

        Why bother to review a hyper-small spec-bump in an end-of-life processor only a few weeks before it’s successor is released, when just a few weeks ago it was clearly established that this 4-year-old design is indeed outclassed by Intel’s 4-month-old design? Why??

      • Meadows
      • 9 years ago

      Why is this comment?

        • dpaus
        • 9 years ago

        If a single new character is released for an older game, should a complete review be written comparing the old game to a brand-new one?

          • DancinJack
          • 9 years ago

          To be fair, there haven’t been any releases in the CPU arena in the same time-frame have there? Also I don’t think it’s a complete review. Definitely shorter than a typical TR CPU review.

    • ronch
    • 9 years ago

    I check TR every day hoping AMD will give more information about Bulldozer. And instead I get this.

      • PRIME1
      • 9 years ago

      Here is Bulldozer taking on Sandy Bridge

      [url<]http://i.imgur.com/FJSxH.jpg[/url<]

        • ronch
        • 9 years ago

        That’s taking on Sandy Bridge, alright.

    • Bensam123
    • 9 years ago

    Same old-new processors… lovely pictures though.

    • swampfox
    • 9 years ago

    Just FYI, at the bottom of the first page you say the i5-2400 has Hyper-Threading, which I don’t think is true (I checked the SB review, and it says only the 17-2600 has it…)

      • sweatshopking
      • 9 years ago

      You’re correct. the 2600 is the first SB with HT

      • Damage
      • 9 years ago

      Doh, thanks. Fixed. Was thinking Sandy Bridge generally, not focusing enough on the i5-2400.

    • RtFusion
    • 9 years ago

    Quick comment:

    What lens and DSLR did you use for the closeup of the pins?

      • JustAnEngineer
      • 9 years ago

      It’s right there in the EXIF embedded in the image:

      Canon EOS Rebel T3i
      55mm, f/8, 1/40s

        • swaaye
        • 9 years ago

        I think they really really like their light box. 🙂

          • Damage
          • 9 years ago

          No light box. Two nice umbrella lamps and some white poster board!

          Wanted to play with the macro capability of the T3i’s kit lens. I’ve been using a (generally nicer) EFS 17-85 lens with a Rebel XT for years, and I largely prefer it on the T3i, as well. However, the kit lens seems to offer some nice possibilities for close-up macro shots with narrow focal planes to make things interesting.

          Overall, I’d prefer a great close-up with a broader focal plane, if you can manage it. Blur in product shots generally isn’t a goal around here. I think my 17-85 with the T3i may allow for some amazing things on that front.

          This was a fun little project, though.

            • ew
            • 9 years ago

            You’ll want to stop down a lot more then f/8 if you want a larger depth of field. Exposure time will go way up so you’ll have to use a tripod if you’re not already. Also, in terms of value nothing beats a cheap set of extension tubes for macro photography. [url<]http://en.wikipedia.org/wiki/Extension_tubes[/url<]

            • Damage
            • 9 years ago

            Like I said, I wasn’t going for a large focal plane.

            • rhema83
            • 9 years ago

            In any case that is a lovely close-up shot of the processor pins. Quite a refreshing and unsual perspective (in contrast to AMD’s lackluster offering being reviewed).

            Your T3i can make awesome pictures. If you have the spare cash for photographic equipment, the EF 100mm f/2.8 Macro is a wonderful lens to use for product photography. It can achieve 1:1 magnification (that’s what Macro lenses are for) and has almost negligible distortion. Of course, double up the pony and you get the L version with IS, but if you are using it on a tripod in a wind-less room, I don’t think you should spend money on features you don’t need.

            Cheers for more cool photos!

    • Arag0n
    • 9 years ago

    I feel funny how no-one ever points how Metro 2033 it’s pretty much more stable using a x6 1100 than the new Core i5 2500 or whatever, with a pretty higher low-frames score. That’s usually what i really care when i see this kind of benchmarks. I don’t really care if single core performance it’s high as hell if once the game needs the good multithreading technology to speed up some slow parts of the game or physics the game expirience becomes crappy, and seems that this is what happens with the metro 2033. I still belive that range the CPU’s from the top to bottom by lowest frame rate it’s better than by average…

      • odizzido
      • 9 years ago

      I care mostly about lowest frame rates as well. Low frame rates usually happen when you most need high FPS as well.

      • Anonymous Coward
      • 9 years ago

      Ordering performance by the minimum observed number would be less reliable (at the same level of benchmarking effort) because just about anything could cause a momentary performance dip. The reviewer would have to ensure that performance dips were repeatable, at the least. I imagine some clever application of statistics could get around it, if there was a suitable dataset to work with.

        • cygnus1
        • 9 years ago

        It wouldn’t be that difficult to do multiple (somewhere between 5 and 10) runs of the benchmark and average the lows. I think you’d get a much better idea of how a game handles on a given piece of hardware that way.

          • Anonymous Coward
          • 9 years ago

          5 or 10 runs is a significant increase in workload for the reviewers. (Also I think even if you did that, taking special note of the lows would still give a more error-prone result.)

        • Bensam123
        • 9 years ago

        Believe standard deviation is what you’re looking for and no it’s not less reliable. It’s no different now then causing a high frame rate to drop momentarily. This all starts dipping into variability and probability, which TR doesn’t cover a whole lot.

        When you get too far into statistics it makes most people head swim though. It’s hard for people to visualize what you put on paper, but I agree. I’ve also talked to Geoff (or Scott, I can’t remember), about this years agot and they said something to the effect that this is very similar to the approach of testing games with vsync on. I don’t think its’ the same case at all, but that’s what I was told at the time

        It would be very nice to test games in a inverse manner.

          • Arag0n
          • 9 years ago

          I don’t think so, the results of the 1100 are consitent with the results of the 1075. Both score 24 and 22 respectively while the new core i2X00’s 16-14. I really think that there is going something in the low frame rates area of the game that makes the game perform slow into intel parts that doesn’t affect AMD x6 parts. You can see that it’s almost a perfect speed up for cores, coming from x4 parts to x6, they come from ~14 to ~24, and it has sense, since if 1 core is designed to do a task and OS tasks and the other 3 are designed to do the frame calculations, then, from 4 to 6 you have 3 vs 5 useful cores, so a 66% speed-up, that means 14->23.24, it makes so much sense to me that something makes use of the number of real cores in the CPU in the game that the HT technology of intel can’t deal, and that’s why the HT and non-HT parts perform almost the same score.

            • OneArmedScissor
            • 9 years ago

            “I really think that there is going something in the low frame rates area of the game that makes the game perform slow into intel parts that doesn’t affect AMD x6 parts.”

            When the GPU is actually pushed, this happenes with other games between the older i7s and X4s, as well, and my understanding is that it’s because of differences in their memory systems.

            Phenom IIs have a 48 way associative L3 cache, while Intel’s L3 is 16 way associative. Phenom IIs actually have slightly worse overall memory latency (I believe the L2 has a bit of an issue), but their particular configuration seems to be more likely to pull what it’s looking from into the memory.

            It has little to do with “speed.” Accuracy might be a better word, but I don’t know that there’s any point in even figuring it out. I find even the 1920×1200, no-AA “compromise” CPU benchmarks comparing a dozen of the same chips completely misleading. I’ve seen numerous tests where lowering CPUs down to 2 GHz has zero impact on framerate.

            What seems to be most important are the delays encountered by needed information in the overall memory structure. Sandy Bridge may be “fast,” but it’s still got a complex system to pipe things through. What I’m curious to see is how Llano stacks up, seeing as it has fairly significant L2 cache, which is something that hasn’t been seen in a while, or with an integrated memory controller, and it should also have lowered L2 latency.

      • indeego
      • 9 years ago

      I think the gameplay literally being on rails makes the game crappy.

    • UberGerbil
    • 9 years ago

    Pretty pictures, though.

    • loophole
    • 9 years ago

    The only place I can see this CPU perhaps being a good buy is as an upgrade for a system that’s got an AM2+ motherboard whose manufacturer never put out a BIOS update for Thuban support (but obviously the board has to support C3 Denebs) and the user has a boatload of good quality DDR2 memory that they don’t want to upgrade from yet (probably 8GB). But even then I’d probably go for a 955BE in that case and put the savings into an upgrade of something else.

      • Alouette Radeon 4870
      • 9 years ago

      Yeah, I’m in that category. I have an MSI K9A2 Platinum motherboard that’s AM2+, 8GB of OCZ HPC DDR2-800 and I LOVE my 790FX chipset because I can mount up to 4 Radeons in Quad-CrossfireX (currently, I have 2 HD 4870s). To put things in better perspective however, I still use my original Phenom II X4 940 (3.0GHz) and it’s still so fast for pretty much everything I do (including gaming) that I haven’t yet felt the need to even overclock it. These reviews are generally just excuses for people to write stuff and make comments on regardless of the fact that we’re really splitting hairs here. What I’ll probably do is upgrade to the fastest X6 available when I finally decide that my CPU is too slow. I expect that to be more than a year away.

    • JustAnEngineer
    • 9 years ago

    One thing that Phenom II does still have going for it is that it rides on much less expensive motherboards than the Intel processors do.

      • paulWTAMU
      • 9 years ago

      It’s not enough to entirely offset it at this point, but it does alter the value proposition somewhat.

      • flip-mode
      • 9 years ago

      If you’re cheaping out that much then you’re not going to be interesting in the 980. But, along those lines, with the 980 you can still rock your DDR2.

        • JustAnEngineer
        • 9 years ago

        There’s probably a reason that I picked the Phenom II X4 955 Black Edition (least expensive quad-core Black Edition processor) instead of the one with faster stock speed and higher price tag. 🙂

          • paulWTAMU
          • 9 years ago

          I’m debating one of those (series not that particular one nesseccarily) into my motherboard ATM actually. I’ve almost never upgraded a CPU without replacing most of the stuff but I just don’t need to…but there’s been a few times I’ve kind of wanted more RAM and CPU power. I’m running a Athlon II X4 640 now…don’t know how much I can upgrade without having to swap out more than I want to.

    • UberGerbil
    • 9 years ago

    The last hurrah of the old regime…

      • Rakhmaninov3
      • 9 years ago

      The Muammar Qaddafi of the mid-range

      • JustAnEngineer
      • 9 years ago

      The question is whether or not Llano significantly changes the landscape.

        • UberGerbil
        • 9 years ago

        I was thinking more about bulldozer. I realize that this is a desktop processor, and BD initially is more of a server product… but then, so was the K8 Opteron line that this Phenom brings to symbolic close. Llano is of course part of that same lineage, and I guess could be considered a bonus hurrah, but with the GPU grafted on it’s more of a dress rehearsal for all tomorrow’s parties than the 980 Black’s last gasp of yesterday’s.

          • dpaus
          • 9 years ago

          Which leads one to ponder when we’ll see BD derivatives with workstation-class embedded graphics.

            • UberGerbil
            • 9 years ago

            The last time I saw an AMD roadmap (haven’t gone looking in a while) the first BD+graphics “fusion” product was Trinity in 2012 — and it’s still at 32nm. And who knows what the graphics will be like (I’m not ever sure I know what “workstation-class” graphics means anymore: that used to denote something higher-end than consumer, but gaming graphics are the top of the heap now and “workstations” just require drivers without the cheats).

            But really, in the context of this “Phenom II X4 980 Black Edition” processor (that is the article we’re at here, afterall) neither Llano or Trinity are in the conversation — this 980 would get paired with an appropriate GPU and so would Bulldozer. Any GPU on the die would be an afterthought at best.

      • FuturePastNow
      • 9 years ago

      I bet there’ll be one more X6 before it is done, too.

Pin It on Pinterest

Share This