AMD lifts the curtain a little bit on its Bristol Ridge APUs

AMD is giving the world a sneak peek at its upcoming Bristol Ridge APU lineup this morning. Here's what we know. Bristol Ridge parts will still be fabricated on a 28-nm process, and they'll still use Excavator CPU cores and Radeon R5 or R7 integrated graphics. All that may sound familiar from Carrizo, but Bristol Ridge APUs still promise some performance improvements over their predecessors.

AMD makes different performance claims about Bristol Ridge using parts with different TDPs, so the first task is to pit apples against apples. The company sticks to 15W parts across the board for its graphics performance comparisons. Intel's Skylake Core i7-6500U CPU serves as the blue team's representative in this fight. That chip's Intel HD Graphics 520 IGP isn't going to blow back one's hair to start with.

Using the 3DMark 11 Performance benchmark to test the Bristol Ridge FX APU and the Core i7-6500U, AMD says the Intel chip scored 1605 while the Bristol Ridge APU scored 2409, or 1.5 times the Core i7's result. Within AMD's APU lineup, a 15W Bristol Ridge chip is supposed to deliver 23% better graphics performance than a 15W Carrizo part, too. The Bristol Ridge platform used DDR4-1866 RAM, while the Skylake platform used DDR3-1600. The move to DDR4 is probably one of the biggest changes for this next-gen APU platform.

It's worth noting that 3DMark 11 Performance is an older benchmark that runs at 1280×720. Between that benchmark and the price point of the Intel notebook AMD chose—a $500-ish Asus laptop with a 1366×768 display—we have a pretty good picture of the systems where AMD expects these APUs to end up.

AMD also notes that in the PCMark 8 v2 Home Accelerated test, which runs with OpenCL support enabled, the 15W Bristol Ridge part stays within 5% of the Core i7-6500U's index score. It also performs about 5% better than a Carrizo chip in the same test. We're fans of the all-important per-thread CPU performance measure, though, and it's there that the picture gets (or perhaps remains) more murky for this new APU.

AMD says a 35W Bristol Ridge APU turned in a Cinebench R15 single-threaded score of 93.24 in its press materials. Compare that to the 15W Core i5-6500U's Cinebench 1T score of 125, and we end up with a somewhat less rosy picture. While the company says Bristol Ridge APUs deliver up to 50% more compute performance over Kaveri APUs in the 35W power envelope, AMD still has a ways to go to catch up to Intel when it comes to IPC.

Performance numbers aside, AMD is showing off one major design win for Bristol Ridge today. HP is using these APUs to power a version of its 15.6" Envy x360 convertible notebook. AMD says Envy x360s will be available with a range of dual-core or quad-core FX APUs, although it's not discussing TDP numbers for those parts. Even so, the x360 sounds attractive. It offers 1080p or 4K display options, mechanical and PCIe solid-state storage options, and an optional infrared camera for Windows Hello. The x360 measures in at 0.74" (18.8 mm) thick and weighs 4.8 pounds (2.2 kg).

AMD says it's shipping Bristol Ridge dual- and quad-core parts to OEMs "in volume" today, and it plans to give the APU family a full introduction at Computex 2016 alongside a range of products built around those parts. We expect to learn more about the changes AMD has made inside this chip to boost its performance then.

Comments closed
    • NoOne ButMe
    • 4 years ago

    YAWN.

    • ronch
    • 4 years ago

    Whoa look… [b<][u<][url=http://tinypic.com/r/2ecgksg/9<]40%!!![/b<][/u<][/url<]

    • ronch
    • 4 years ago

    I sneaked, I peeked, and I saw the same Carizzo 28nm chip using the same Excavator cores and the same R5/R7 graphics.

      • DPete27
      • 4 years ago

      Yeah, but that BLISTERING fast DDR4-1866 support!!!

      …. oh wait…you say Skylake supports DDR4-2133? Why was it tested with DDR3-1600? Wouldn’t have been to pad the performance comparison….would it?

        • chuckula
        • 4 years ago

        NO NO NO NO NO NO… yes.
        Yes, a bit, a bit.

        But Skylake does have a wart!

        • MOSFET
        • 4 years ago

        Probably because that is what is normally shipped with 6xxx-U parts.

    • flip-mode
    • 4 years ago

    [quote<] Bristol Ridge parts will still be fabricated on a 28-nm process, and they'll still use Excavator CPU cores and Radeon R5 or R7 integrated graphics.[/quote<] Slow clap. Shake head. Read no further.

      • ronch
      • 4 years ago

      Well, there’s probably 1 guy in the world who finds this interesting, right?

    • cygnus1
    • 4 years ago

    [quote<] it plans to give the APU family a full introduction at Computex 2016 [/quote<] Could at least say in which month Computex 2016 is planned... Not everyone memorized when industry trade shows are scheduled.

      • maxxcool
      • 4 years ago

      😉 [url<]https://www.google.com/search?q=Computex+2016&oq=Computex+2016&aqs=chrome..69i57.4478060j0j1&sourceid=chrome&ie=UTF-8[/url<] /ducks for cover/

        • cygnus1
        • 4 years ago

        Lol, I know how the googles work… Not the point 🙂

          • maxxcool
          • 4 years ago

          heh. sorry .. tuesday .. lowhanging fruit.. and a hangover 😀 /cheesegrin/

    • Chrispy_
    • 4 years ago

    I’d really love it if TR could review one of these things.

    The problem with a lot of AMD APUs is that they don’t get much review coverage, and the review coverage they get is for parts that don’t end up on shelves in stores.

    A TR ‘real world test’ of a common Bristol Ridge model alongside an equivalent midrange Intel laptop with run of the mill HD 520 IGP would be awesome. For people that don’t game much, (but do still game) it’d be nice to see what the drawbacks are for getting improved gaming performance in a real-world scenario.

      • odizzido
      • 4 years ago

      Yeah I’d like to see that as well.

      • EndlessWaves
      • 4 years ago

      Improved gaming performance? What improved gaming performance?

      Skylake’s mid-range integrated graphics is similar to Carizzo’s best. The R6 and R7 are somewhere between the 520 and 540. As Bristol Ridge seems likely to be just a refresh (a clock speed bump) I wouldn’t expect much difference.

      I’d give good odds that a 540 will be as good as Bristol Ridge R7.

      Obviously there’s the question of availability but I know HP were offering the i3-5157U on some laptops so hopefully the i3-6167U will become available too.

        • Chrispy_
        • 4 years ago

        Not improved compared to previous gen AMD, Improved compared to [i<]current-gen Intel[/i<]. In the past, it was easy; Intel IGPs could not run AAA games, AMD's could - so your only IGP gaming option was AMD. Now, Intel's IGPs run games. The HD4600 can play most AAA games at 720p 'low' detail at 20fps. Hardly stellar, but good enough to qualify as 'playable' The HD 520 is a little bit better still, and benefits from DDR4 in some laptops too. What I (and I presume others) want to know is whether it's still worth trading Intel's CPU advantage for AMD's IGP advantage.

    • mkk
    • 4 years ago

    Can’t wait for them to lift the skirt a little bit on Zen.
    Ahem…

      • chuckula
      • 4 years ago

      Here’s all you need to know: [b<]40%!!!!!![/b<]

        • ronch
        • 4 years ago

        Incidentally, I’m 40% sure it’ll catch up to Broadwell or Skylake.

          • Ninjitsu
          • 4 years ago

          It’ll have to be Kaby Lake now…

            • ronch
            • 4 years ago

            AMD always catches the previous generation/s. That’s still something, I guess.

          • Anonymous Coward
          • 4 years ago

          Maybe it will cost AMD only 40% as much to fab.

            • ronch
            • 4 years ago

            Or cost 40% more to fab, 40% more likely.

            • Anonymous Coward
            • 4 years ago

            Meh, if AMD does one thing right with this new design, it’ll be like Bobcat/Jag and be a highly fab-able chip.

            • ronch
            • 4 years ago

            Yes, it’ll be easily portable across process technologies, but even so, AMD’s fab partners will need to earn some money. At the end of the day, it’s about how much Intel saves by fabbing chips by themselves and how much more AMD spends to have someone else make chips for them. Of course, no one will really have a concrete answer to this.

            • Anonymous Coward
            • 4 years ago

            I don’t expect they will undercut Intel’s manufacturing prices, but they perhaps don’t need to. Getting a similar cost, at 80% the performance, in large volumes, I bet they would do well. Margins are high, lots of room to play, if they have the product.

      • albundy
      • 4 years ago

      It’s the bruce jenner of brands. once well respected for its performance, but now a laughing stock.

      • Unknown-Error
      • 4 years ago

      perv!

    • Anonymous Coward
    • 4 years ago

    So… any particular changes in the silicon?

      • Jeff Kampman
      • 4 years ago

      Almost certainly, but AMD isn’t discussing them today. We’ll have to wait until Computex.

        • NTMBK
        • 4 years ago

        Isn’t Bristol Ridge just improved power management firmware, plus an enabled DDR4 controller? AMD ran a talk titled “Increasing the Performance of a 28nm x86-64 Microprocessor
        Through System Power Management” about it.

        • Ninjitsu
        • 4 years ago

        I’m curious as to what AMD is doing at Computex, I thought it was Intel’s event. (Intel and Taiwanese partners, primarily).

          • chuckula
          • 4 years ago

          Computex isn’t any one company’s event, it’s a very large trade show that has heavy representation from pretty much everybody except for Apple.

            • Ninjitsu
            • 4 years ago

            AMD doesn’t appear to be a regular (see link). Wasn’t there when I went in 2014 , either, afaik.
            [url<]https://www.computextaipei.com.tw/en_US/show/info.html?id=F1970639CF716213D0636733C6861689&sFuncID=F1970639CF716213D0636733C6861689[/url<]

      • ronch
      • 4 years ago

      Probably about as many changes in the silicon between an FX-8350 and an FX-9590.

    • ronch
    • 4 years ago

    Looking at that ‘7th generation’ slide reminds me how sleezy those AMD marketers are. Kabini and Beema are included to make this iteration look more far ahead in the evolution of APUs? Really. It’s just the 4th generation if we only include desktop APUs of which Carizzo and Bristol Ridge are really part of (and only the 3rd if we base our criteria only on Bulldozer-derived APUs). It’s like saying I’m the 1st generation and my kids are the 4th generation because of my cousins and their kids. Yes, BR now goes down the TDP ladder to Kabini levels but this is still a desktop part downclocked to reach those TDP levels. I dunno. Doesn’t seem right to me.

      • bittermann
      • 4 years ago

      Did AMD kick your dog or steal your girlfriend? Seems like it by looking at your posts.

        • ronch
        • 4 years ago

        No, but they need better marketers.

          • maxxcool
          • 4 years ago

          Ye godz yes. for the sum of a half eaten kitkat i would drown (buy my own hand) their entire marketing team in a tub of warm fetid vomit, feces, and blood of their own excretion.

            • Kretschmer
            • 4 years ago

            That’s…a bit extreme.

            • maxxcool
            • 4 years ago

            Yes, but fully counterweighed and in balance with the marketing teams sleaze in the past. To be fair I equally bashed Intel on PrescHot fiasco as well.

            • Welch
            • 4 years ago

            More extreme than these new APUs…..

        • TheMonkeyKing
        • 4 years ago

        No, but he still has high hopes for the company. It’s sad to see him crushed when he gets let down.

        #IGaveUpIn2008

      • auxy
      • 4 years ago

      Llano → Trinity → Richland → Kaveri → Godavari → Carrizo → Bristol Ridge, looks like 7th to me! ┐( ̄ヮ ̄)┌
      I would allow for leaving out Richland and Godavari, since Intel doesn’t call Devil’s Canyon a ‘generation’, but even in that case it’s still 5th generation. Pretty obvious anti-AMD bias there, ronch. (´・ω・`)

        • chuckula
        • 4 years ago

        Trinty –> Richland: Not really new.

        Kaveri –> Godavari: Not really new.

        Chorrizo –> Bristol Ridge: Not really new.

        So it’s more realistically: Llano –> Trinity/Richland –> Kaveri/Godzillavardi –> Chorrizo/Bristol Ridge.

        But then again, Skylake is the sixth generation Core while AMD wanted a number higher than 6 so: mission accomplished.

          • auxy
          • 4 years ago

          Bristol Ridge has DDR4, Carrizo didn’t. So clearly it’s a more significant change than Kaveri/Godavari, nya. ( ̄ω ̄)

            • chuckula
            • 4 years ago

            [quote<]Bristol Ridge has DDR4, Carrizo didn't. [/quote<] That's actually not true. Carrizo does technically include a DDR4 controller and embedded versions of Carrizo do support DDR4. It's just that commercially available products (of which there were very VERY few models available) stuck with DDR3 because CHEAP. [url<]http://www.anandtech.com/show/9726/amd-launches-excavator-based-rseries-apus-for-embedded-with-ddr4-support[/url<]

          • NoOne ButMe
          • 4 years ago

          The secret is AMD count small core parts also. Bobcat -> Llano –> Jaguar –> Trinity –> Kaveri –> Carrizo –> Bristol Ridge.

          Although Bristol shouldn’t count as the 7th. 6.1 maybe.

        • ronch
        • 4 years ago

        That’s not what it says in the slide.

          • auxy
          • 4 years ago

          So?

            • ronch
            • 4 years ago

            If you read my post correctly, I was talking about how they slipped Kabini and Beema in there to get to 7. Personally I would’ve been more OK with Richland and Godaveri in there. They’re not exactly what I’d think having enough evolutionary changes to warrant being called a generation step forward but at least they were part of the evolution of desktop APUs.

            Edit – just to make it clear once and for all to you guys out there who think I’m anti-AMD: I’m honestly a big fan of AMD since the Am486 days. I have been watching the company generation after generation but what they’re doing these days, especially their marketing people, is just stupid most of the time. Crashing Nvidia’s party, cranking out those stupid 220w TDP chips just to make some noise (probably Roy Taylor’s idea), not even knowing how many transistors Bulldozer had when it came out, marketing people like Adam Kozak and Roy Taylor who look like they should’ve been game show hosts instead, their below-the-belt Fixer videos, poking fun at or openly taking advantage of their competitors’ misfortunes (Sandy Bridge chipset problems, etc.), etc. Not very professional.

            • Ninjitsu
            • 4 years ago

            I remember you being pro AMD and have seen you slowly getting pissed with them. 🙂

            • Rza79
            • 4 years ago

            Ronch, you forgot their biggest scandal. Calling a four core cpu, a eight core cpu. That’s the winner.

            • ronch
            • 4 years ago

            Well, I’d like to think of my FX-8350 as really having 8 cores, so I’ll stick with this belief for now. When I replace it with Zen+++++ somewhere down the road, that’s when I’ll tell myself, “It wasn’t really an 8-core CPU. What you have NOW IS a true 8-core CPU.” 😀

      • NTMBK
      • 4 years ago

      AMD marketing are always working in Intel’s shadow. Just look at the product names they introduced with Llano! Intel has an i3, i5 and i7, so AMD have to bring out an A4, A6 and A8. They’re all one higher than Intel, and have an A for AMD instead of an i for Intel. Pretty pathetic.

      I miss the Phenom II X4 naming. First part tells you what generation this chip is, second part tells you how many cores. Dead simple.

        • ronch
        • 4 years ago

        I think Intel’s i3, i5 and i7 monikers were influenced by BMW. AMD’s A4, A6, A8 and A10 were influenced by a certain other German luxury car maker. And there’s Celeron -> Sempron, Pentium -> Phenom, ‘K’ used as a suffix to indicate unlocked multipliers, Summit Ridge and Bristol Ridge being somewhat similar to Sandy Bridge/Ivy Bridge, etc. I don’t think those AMD trademarks/codenames were purely coincidental. Of course, there’s nothing wrong with this but it’s just noticeable.

          • NoOne ButMe
          • 4 years ago

          Rathe have influence marketing that taking IMC, x86-64, etc.

          In all fairness all of these companies ‘steal’ ideas from each other. Part of the reason AMD CPU side got so screwed is trying to not steal ideas lol! And the other Part Intel has all the monies (so design talent and fabs and sadly at times bribes).

        • chuckula
        • 4 years ago

        AMD GOES TO ELEVEN!!!

        • JMccovery
        • 4 years ago

        Actually, the first digit of the Phenom II/Athlon X4 models generally denoted the number of cores, and the next two digits mainly denoted family position based on clockspeed, amount/absence of L3.

        Though, there were several times where AMD screwed up naming (960T and 800-series, I’m looking at you).

          • ronch
          • 4 years ago

          People complained about how long Phenom’s full product names were (e.g. Phenom II X4 955) but I honestly found them easier to understand and remember than today’s product names (Core i5-4460.. A8-7600.. FX-6300… although the FX’s first digit clearly denoted core count until the ‘Centurions’ threw out everything we knew).

            • JMccovery
            • 4 years ago

            I mainly named them based on the ‘brand’ and model number: Athlon 640, Phenom 555/955 (unlocked cores, baby!). Yeah the ‘II X#’ parts of the names were just annoying to say all the time.

            TBH, ID rather go back to name and number products, than the ‘random letter+random number – number string’ products we have now, because things like the A10 and A12 series seem like unneeded product segmentation.

            • ronch
            • 4 years ago

            Honestly, AMD’s and Intel’s naming conventions these days are pretty much senseless. A mobile Core i7 is probably slower than a desktop Core i3, which makes the i3/i5/i7 branding not very helpful to anyone who isn’t an enthusiast, and their 4-digit codes are also all over the place. I have 2 cousins who have a Core i3-4170 and a Core i3-4340. About the only thing that makes sense to me is the i3 part because they’re both desktop parts (which tells me both are 2C/4T parts), and the first digit of the 4-digit code (which indicates they’re Haswell parts). In terms of clock speeds and cache, both are pretty much the same. One has 1MB more L2 and the other makes up by running 0.1GHz faster. But of course I never would’ve guessed those by just looking at the model numbers. Thank goodness for ARK. AMD doesn’t have an ARK facility so it’s CPU World there.

        • Duck
        • 4 years ago

        It’s like older parts were named by engineers whereas newer parts are from a marketing department with no money; or is that an AMD with no money?

    • anotherengineer
    • 4 years ago

    I wonder if any of these will come with a dual channel memory controller??

    I wonder if makers will just make single channel motherboards for them??

    edit – and I wonder if the smallest ones will based of puma and a true quad core like the “carrizo A8-7410”??

      • RAGEPRO
      • 4 years ago

      All of these processors should sport dual-channel memory; these are based on the ‘heavy equipment’ cores, not the ‘cat’ cores.

        • ermo
        • 4 years ago

        Not too long ago, there was a long article on another tech site describing how vendors mostly use single-channel RAM for their AMD offerings because it allows them to use a single motherboard platform for both the small (Beema) and the big AMD APUs (Carrizo).

        This was with DDR3, but it still sucks as you only get decent APU performance from the big APUs in the 15-35W range using a dual channel setup.

        • anotherengineer
        • 4 years ago

        So now we just have to hope the laptop mobo are not configured into single channel mode now.

        [url<]http://www.extremetech.com/computing/222662-poor-oem-designs-single-channel-ram-sabotage-amds-carrizo[/url<] [url<]http://fudzilla.com/news/processors/39903-oems-appear-to-kill-amd-s-carrizo[/url<]

    • ronch
    • 4 years ago

    These chips hardly make my heart beat a little faster but given my boring PC usage scenarios I’d actually consider these if my CPU and mobo both quit tomorrow.

    • chuckula
    • 4 years ago

    Thanks AMD but I think I’ll leave the hardware testing to TR.

    You just focus on actually launching the chips, TR will take it from there.

      • RAGEPRO
      • 4 years ago

      I’m looking forward to seeing how these things actually run. This will be the last gasp of the Bulldozer family and I’d like to see it able to go out with its head held high … or at least not crying in the corner.

        • auxy
        • 4 years ago

        You foolish fool! ψ(`∇´)ψ Such vain hopes. You know that even if the CPU performance were up to snuff, the platform is too garbage! Crap memory performance! Toilet-tier caching!

        Search your feelings. You know it to be true! 。゚(TヮT)゚。

          • RAGEPRO
          • 4 years ago

          Yeah. Probably. It’s a nice thought, though.

          • ronch
          • 4 years ago

          Come on, Auxy, yes I know the Bulldozer lineage is pretty far behind by all measures but it’s not really THAT bad in real world use. In a 10-car comparison shootout over at, say, Car & Driver, you can get the #1 car or you can get the #10 car, and while the #10 car will probably sound a little thrashier, a little thirstier, a little slower to 60mp/h, and look a little dorkier, in the grand scheme of things, they’ll all get you from Point A to Point B. Just get what makes you happy.

            • auxy
            • 4 years ago

            I’m real-life friends with RAGEPRO (who is a writer on this site now!) for nigh-on 20 years, so I was kind of ribbing him a little. He’s always been a bit of an idealist and he gets hopeful about things without really considering the facts, whereas I was once an idealist who had her hopes crushed too many times and am now a cynical realist. (゚∀゚)

            The thing is, I’m a the hardest of hardcore gamers, ronch, and I accept nothing but the most optimal experience possible. For the money, there’s just no reason to consider an APU, as a gamer, except possibly in a NUC-like or laptop form factor. As a desktop machine, even though the CPU performance really is pretty much there in large part, especially multi-threaded, the machines just don’t do well in games when you look at the frame latency metrics. (;¬_¬)

            There are hundreds of videos all over Youtube that compare the APUs to things like 3rd-4th gen Core i3s, and say “look at the FPS!” but in the 30fps video you can’t see the stuttering and hitching that the APU machine is doing. I’ve built the machines myself, I’ve seen them running side by side, I know the difference. Whether it’s because of the single-threaded CPU performance or the memory/caching issues or whatever, APUs suck for anything that’s latency-sensitive — like gaming. I’ve even tested a Richland APU at 4.4Ghz with a discrete GPU, and then taken that same GPU and stuck it into a Pentium G3220 machine, and observed smoother, better gameplay with the Pentium. Same RAM, too, by the way.

            It’s a shame, but that’s the reality. (・へ・) That #10 car might get you there, but you’re not gonna enjoy the experience like you would the #1 car, and in this case, it’s just #1 and #2. I think I’ll stick with #1. (And you know what an AMD fangirl I am!)

            • ronch
            • 4 years ago

            OK, got it, Auxy. Yes we all want #1 or #2. I know #10 sucks but I guess I’m just being kinda simple with life these days and trying to be thankful I even have a decent car and a decent computer. I thought, if I won the lottery (not that i play, it’s just to illustrate a point), I’d get a cheap but functional Suzuki instead of a BMW. How many people would actually do that?

      • Anovoca
      • 4 years ago

      [quote<][s<]AMD[/s<] [i<]Scott[/i<] says the Intel chip scored 1605 while the Bristol Ridge APU scored 2409[/quote<]

        • chuckula
        • 4 years ago

        Don’t go putting words in Scott’s mouth!
        And remember, Scott doesn’t work for AMD… he works for the Radeon Technology Group!

          • Anovoca
          • 4 years ago

          That’s why he only ran the 3dMark. Somebody else runs the Cinebench tests

            • chuckula
            • 4 years ago

            His clicking finger was kidnapped!

Pin It on Pinterest

Share This