AMD previews Carrizo APU, offers insights into power savings

“Carrizo” is the code name of AMD’s next-generation CPU for notebooks and convertible PCs. This chip has been on AMD’s roadmap for some time now as the successor to the Kaveri chip that powers the firm’s current lineup of A-series APU products.  We even got an early look at the first working Carrizo silicon at CES in January.

Still, many of the details about Carrizo and its next-generation “Excavator” CPU cores have been shrouded in mystery to date. Fortunately, that’s changing. In conjunction with the International Solid State Circuits Conference, AMD has begun to tell the story of how it has achieved major improvements in power efficiency and performance with Carrizo, even though the chip is built on a 28-nm fabrication process like Kaveri before it. AMD Corporate Fellow Sam Naffziger and Senior Director of Client Products Kevin Lensing briefed us ahead of ISSC and shared some fascinating information about Carrizo’s new technology.

Introducing Carrizo

For the uninitiated, Carrizo is AMD’s answer to Intel’s Broadwell chips, and it’s expected to arrive in consumer systems around the middle of this year. Like AMD’s other “Accelerated Processing Units,” or APUs, Carrizo combines CPU cores and graphics on the same piece of silicon. In fact, Carrizo is almost a complete PC system on a single chip, and nearly every major component onboard has been updated compared to the prior generation.

We don’t yet have all of the details, but Carrizo combines an evolved version of the Bulldozer CPU core known as Excavator, “next generation” Radeon graphics based on the GCN architecture, and an updated UVD accelerator block capable of handling H.265 video. Carrizo is also the first “big” AMD APU to integrate the traditional south bridge I/O functions (like USB and SATA), making it a true system on a chip.

Thanks to this change, Carrizo is able to share the same pinout and motherboard infrastructure as AMD’s low-cost, low-power product, known as Carrizo-L. (Carrizo-L is similar to Beema and Mullins and will likely compete with Intel’s Bay Trail and Cherry Trail products.) Lensing told us AMD hopes the shared infrastructure between Carrizo and Carrizo-L will allow the company to capture more of the available market. PC makers should be able to offer systems across a broad range of price and performance levels based on the same basic chassis and motherboard.

AMD claims Carrizo is “the first processor in the world with HSA 1.0 support,” referring to its Heterogeneous Systems Architecture effort to enable converged CPU-and-GPU computing. That claim is a bit confusing since AMD said something similar about Kaveri, which it touted as the first architecturally complete HSA development platform. In this case, the mention of the HSA 1.0 spec is important. That spec has long been a work in progress and is only just being finalized. Perhaps it’s no surprise that only Carrizo meets its full demands. More concretely, Carrizo adds at least one relevant HSA feature that Kaveri lacks: GPU context switching for multiple processes. When it arrives, Carrizo will surely become the reference platform of choice for HSA development. (Whether or not HSA will gain any great traction with software developers, of course, is another question.)

Carrizo’s real magic isn’t listed in its spec sheet, though. Instead, it has to do with how AMD tackled a daunting engineering challenge: delivering meaningful improvements in chip density, performance, and power efficiency over Kaveri without the benefit of a die shrink. After all, this chip has to compete with Intel’s Broadwell, which is fabricated on a much more advanced 14-nm process with second-gen tri-gate transistors. Carrizo is built on a 28-nm process using traditional planar transistors.

Yet AMD claims it has managed to squeeze out some substantial improvements over Kaveri. Overall, Carrizo weighs in at roughly 3.1 billion transistors, or 29% more than Kaveri, with “approximately the same” die area. Power use and performance, two sides of the same coin, are also apparently much improved in this new chip. The firm has achieved these gains using careful tuning for laptop-class power envelopes bolstered by various innovative techniques—and that’s what AMD is sharing this week at ISSCC.

Excavator: heavy equipment gets streamlined

The Excavator CPU cores in Carrizo are the fourth generation of cores based on the initial Bulldozer microarchitecture. Each generation has improved per-clock instruction throughput and power efficiency over the last one, and Excavator is no exception. AMD estimates a 5% overall gain in per-clock instruction throughput over the prior-gen Steamroller core thanks to various changes.

We don’t know what all of those tweaks are yet, but Naffziger did mention one change in particular: the L1 data cache has doubled in size while maintaining the same access latency. He also alluded to support for new instructions in Excavator, but without offering any further details. Excavator is rumored to add support for AVX2, which would boost performance in specific code paths that use SSE or the like, but the new instructions wouldn’t contribute to a general performance increase. At any rate, we don’t expect dramatic changes on the CPU architecture front from this generation of AMD tech. Those are likely reserved for the upcoming Zen microarchitecture, an all-new, x86-compatible core expected to supersede the Bulldozer family next year.

The most notable changes in Carrizo come not in architecture, but design. Naffziger said the Excavator team “stole some plays from the GPU playbook” by adopting a high-density design library traditionally used for GPUs. This library packs quite a bit more logic into a given amount of chip area. The examples below show some important parts of the Excavator core when laid out using a high-performance library a la Steamroller and a high-density library a la Carrizo.

The overall improvement in density is even more dramatic than one might expect from casual inspection of the examples above. The dual-core CPU modules on Carrizo occupy 23% less area than Kaveri’s, even with the added features and the doubling of the L1 data cache’s capacity—all on a 28-nm process.

That said, we are talking about very different looking chips, when all is said and done. The images below illustrate the layers used in a CPU-focused metal stack versus those used in a GPU-focused stack.

The Excavator team made a trade-off here, choosing the higher logic density of a GPU-style design over the clock frequency headroom afforded by a CPU-style design.

As the plot above attests, that trade-off makes sense for Carrizo because the chip is targeted to laptop-class power envelopes of about 15W. Not only does the high-density library reduce the chip area required for each CPU core (thus saving on costs), but it also yields some nice reductions in power use during low-wattage operation.

Notice that the crossover point where Excavator no longer beats Steamroller is at about 20W per dual-core module. That fact may help explain why AMD hasn’t articulated plans to produce a socketed version of Carrizo for desktop systems. The chip’s tuning probably doesn’t translate well into desktop-class power envelopes of 65W or higher. Carrizo’s benefits over Kaveri may be questionable in such scenarios.

 

GCN goes low-power

AMD has done some power optimization on the GPU side of things, as well. Again, the changes were all intended to help tailor Carrizo for its intended power envelope. The team tuned Carrizo’s GPU cores for low-power operation by reducing their reliance on high-performance devices that bleed more power in the form of leakage. By selecting cooler, lower-power options from the suite of devices available in the 28-nm process, they were able to realize substantial efficiency gains: either a 20% power savings at the same clock speed or a 10% higher operating frequency in the same power budget.

These optimizations allowed AMD to enable all eight of the GCN compute units simultaneously in the low-power version of Carrizo. Previously, in the low-power Kaveri, they had to limit the chip to six CUs at once in order to avoid exceeding the APU’s power budget.

Again, optimizing an integrated GPU in this fashion involves a trade-off. The peak operating frequencies of the graphics cores in Carrizo are likely lower than Kaveri’s, which would translate into lower peak performance at higher power levels.

Voltage-adaptive operation

Chips need a certain minimum amount of voltage in order to operate properly without crashing. Unfortunately, the voltage supplied to a chip in a typical system isn’t always perfectly steady. To avoid problems, chipmakers typically supply a little extra voltage to their chips. Naffziger told us AMD has generally overvolted by about 10% in order to compensate for potential voltage droop. That may not sound like much, but a chip’s power draw is determined in large part by the square of the voltage, so keeping voltage low is a critical goal.

AMD has an innovative solution to this problem: voltage-adaptive operation. The firm’s CPU cores have the ability to track the supplied voltage in real time, at “sub-nanosecond” speeds, in order to detect when voltage droops. The chip can then reduce its operating speed briefly in response to the voltage reduction, preventing a crash.

Naffziger notes that voltage droop happens “less than one percent of the time,” so voltage-adaptive operation should have no noticeable impact on performance. Instead, he says, “we just get a bunch of that power waste back” by not needing to overvolt the silicon to ensure stability. The firm can choose to turn the power savings into higher clock speeds, too, if it wishes.

Voltage-adaptive operation was built into Kaveri’s CPU cores, but in Carrizo, it’s been incorporated into the graphics CUs, as well. AMD estimates this technique reduces power consumption by up to 19% for its CPU cores and by as much as 10% for its integrated graphics.

Adaptive voltage and frequency scaling

AMD has also sought to reduce voltage in Carrizo by giving the chip the ability to tune itself. More precisely, this optimization, known as Adaptive Voltage and Frequency Scaling (AVFS) currently applies only to the Excavator CPU cores.

Each Excavator core includes a scattered collection of AVFS “modules” that include replica versions of critical logic pathways in the CPU. The AVFS modules can test these pathways for stability at different voltage levels in real time as the chip operates. AVFS thus allows the CPU core to know its lowest safe operating voltage for the present conditions, including the clock speed and temperature.

AMD estimates that the voltage reductions made possible by AVFS can cut power consumption between five and 15% at a given clock speed. As with many of the other optimizations in Carrizo, the biggest gains from AVFS come at lower power levels.

Low-power standby

For several generations, AMD’s mobile platforms have been missing support for a Windows 8 feature known as Connected Standby mode. This is the mode that allows Windows-based systems to mimic the behavior of phones and tablets by dropping into a deep sleep state that uses very little power. The system can then wake up briefly to check for any incoming messages and notify the user, or it can simply awaken very quickly when the user is ready to resume working.

AMD didn’t talk about Connected Standby in its Carrizo presentation, but it did confirm that the chip will support the primary power state needed for this feature, a state known as S0i3. In this state, the power management subsystem gates off power to nearly all of the APU’s silicon, reducing chip-wide power consumption to less than 50 milliwatts. The time required to enter and exit S0i3 mode is much lower than the time needed to transition to the traditional S3 standby mode. AMD says the APU can drop into S0i3 in less than a second.

This addition doesn’t yet give AMD’s big APU all of the aggressive “active idle” capabilities that Intel built into its mobile Haswell offerings, in part because Carrizo lacks the fast state transitions made possible by Haswell’s integrated voltage regulation. Intel has said that Haswell can wake from being completely powered down in under three milliseconds. Still, the addition of S0i3 to Carrizo is a step in the right direction.

 

The future is still fusion?

Back in 2012, AMD’s then-new CTO Mark Papermaster laid out a development direction for AMD that involved SoC-style design principles. Most mobile SoCs are built using a particular approach where discrete functional blocks are glued together using a common interconnect, like ARM’s AMBA spec. This approach, combining modularity with a common interconnect, can allow for rapid development of new chips, with shorter design cycles and more flexibility. Since Papermaster joined AMD, the firm has put together a pretty good track record of delivering new APUs on a fairly regular basis, but it hasn’t yet produced a big x86 APU based on SoC-style design principles.

For example, for most intents and purposes, AMD’s Kaveri is kind of a Radeon glued to a Bulldozer core. Rather than sharing a memory controller over an interconnect fabric that maintains memory coherency—as one would expect with the SoC approach—Kaveri’s GPU has three paths to memory: a 512-bit Radeon memory bus, a 256-bit “Fusion Compute Link” into CPU-owned memory, and another 256-bit link into CPU-owned memory that maintains coherency. In theory, with a proper coherent fabric, these three links could be merged into one, saving power, reducing complexity, and quite probably improving performance. The use of a proper interconnect fabric would also allow AMD to swap in newer or larger graphics IP without requiring as much customization.

AMD surely chose to build Kaveri as it did for good reasons, most notably because it needed to deliver a product to the market in a certain time frame. Still, one can’t help but note that Intel’s original Sandy Bridge chip had a common ring interconnect joining together the CPU cores, graphics, shared last-level cache, and I/O. From a certain perspective, although it wasn’t meant for this mission, Sandy Bridge’s basic architecture was arguably a better fit for AMD’s HSA execution model than Kaveri.

We don’t yet know all of the details about AMD’s new APU, but the firm confirmed that Carrizo follows the same basic implementation style as Kaveri. Carrizo doesn’t yet embrace the SoC-style design methodology AMD intends to adopt.

That said, AMD’s new APU doesn’t have to follow this approach in order to be a great product for 15W laptops. I’m persuaded AMD’s power-optimizations efforts are a smart choice for this generation. Given that it takes about four years to build a chip of this class from scratch, I would expect next year’s models finally to make the transition to a more modular layout.

Zen and the art of feature matrices

Speaking of the future, one of the most interesting slides AMD had to share with us was the following matrix of product features, including some features that are coming in future products. Have a look, since it offers a few juicy tidbits of new information.

I’m not really clear on how all of these time frames map to products, exactly, but AMD told me the features included in Carrizo count as “in product” options in the blue boxes. Some of these things we already know about, such as AVFS and voltage-adaptive operation. Others are still mysterious, like inter-frame power gating, which could be a GPU-focused optimization. (Hmm!) I suspect we’ll know more about these technologies once Carrizo hits the market mid-year.

The “in development” features in purple are probably giving us some of our first insights into the chips based on AMD’s upcoming Zen microarchitecture. Most notable among them is integrated voltage regulation, a la Intel’s Haswell and Broadwell CPUs. Some of the other features listed look to be power-management capabilities made possible by the faster switching and finer granularity of integrated voltage regulation. For instance, per-IP adaptive voltage likely means separate supply rails for various on-chip units. Environment and reliability-aware boost could be an extension of AVFS. The workload-aware energy optimization sounds a lot like the Energy Efficient Turbo feature Intel built into its Haswell-EP processors; this feature monitors stalls on the CPU and reduces clock speeds if the CPU core’s performance is limited by external factors.

I suspect “advanced bandwidth compression” is a GPU feature. AMD introduced much-improved frame buffer compression in its Tonga GPU last fall. Looks like that capability could make it into APU silicon in 2016 alongside Zen. I hate to steal the spotlight from Carrizo, but Zen could be where AMD really catches up to Intel in a big way.

A big win for efficiency in laptops

We don’t have much in the way of specifics about the full extent of Carrizo’s power efficiency gains over the prior generation just yet, but AMD is claiming “double digit” increases in both performance and battery life.

The power efficiency plot above helps tell that tale visually. We’ll probably have to wait until closer to Carrizo’s release before we have more specific numbers. Regardless, this power-efficiency progress is all part of AMD’s goal, stated last year, to make its products 25 times more power-efficient by the year 2020.

AMD tells us Carrizo parts will fit into power envelopes ranging from 12 to 35W. That is the “breadth of the design space” for this chip, although, cryptically, the firm also says that “doesn’t mean it doesn’t scale beyond that.” I’m not sure what to make of that statement. Either it’s purely an engineering sentiment, or it might mean we could eventually see Carrizo-based products that push below 12W or above 35W, though presumably they wouldn’t be entirely optimal implementations of the chip.

Whatever the case, AMD has pretty clearly aimed for a specific target with this generation of its APU technology. In Naffziger’s words, “We optimized for a range and knocked it out of the park.” The result may be a chip that competes credibly against Intel’s 14-nm mobile Broadwell offerings. If AMD can manage that feat with 28-nm silicon, it will be quite an achievement, given everything.

Comments closed
    • WillBach
    • 5 years ago

    I wrote a test harness for some of the HSA stuff. Maybe I should do an “ask me anything.”

    • Chrispy_
    • 5 years ago

    I don’t like the idea of AMD sacrificing clockspeed to get greater transistor density.

    The single largest failure of AMD’s architecture over the last five years has been low IPC. Now they’re reducing clockspeeds too?

      • Meadows
      • 5 years ago

      They don’t sacrifice clock speed. They’re sacrificing *potential* clock speed, i.e. these chips will be physically incapable of hitting 3-4 GHz in stable operation.

      Which is just as well, because these types of chips never run at speeds like that to begin with.

        • Chrispy_
        • 5 years ago

        I thought GPUs were really hard to run at more than 1.2GHz though.
        Maybe I’m reading too much into the “CPU built on GPU tech” part.

          • Meadows
          • 5 years ago

          I’m no authority on the matter but I believe the two can’t be compared that easily. The type of architecture in question probably has an effect on clock speed ceilings too.

    • ermo
    • 5 years ago

    A ThinkPad T450 or T450s-ish model with Carrizo, 8GB of RAM, a 14.1″ 1600×900 or 1920x1080p IPS panel and a 500MB MX200 SSD might not be a half bad proposition then?

    It might even compete favorably with the i3 option in the T450 on both price and performance?

    • UnfriendlyFire
    • 5 years ago

    HP: “Let’s stick them in our Elitebooks… And price them to the point where our i5 Elitebooks and competitors’ cheaper business notebooks with dedicated graphics are competing in the same price range!”

    Others: “768p TN 15.6″ display with 4 hours battery life, 5 lb, and bad trackpad/keyboard is good enough. Price it at $400 to $700. Oh, and maybe throw in a low-range dedicated GPU so we can advertise it as DUAL GRAPHICS!!!”

    Asus: “Let’s stick a Temash (or some other tablet APU)… in a 17″ laptop. And still advertise it as having an i7 power because we’re too lazy to update the marketing material.”

    (There was an Asus laptop with a 1 GHz Temash chip. And yes, the marketing said it had an i7 in it. I’m sure Intel would have something to say to Asus if they stuck a Cherry Trail or Braswell CPU in an 17″ laptop)

      • sschaem
      • 5 years ago

      I dont think AMD got a clue about this situation…

      They have decent chips for some markets, but OEM produce crap nobody want to touch.

      Tweak a few things to highlight what those APU are all about,
      and some model would find their niche.

      BTW, those are the moment when a R&D department realize how important it is to have a functioning sales department not managed by monkeys.

      AMD Chief Marketing Officer & Chief Strategy Officer are out, hopefully its a sign of some fresh blood with brains coming in.

    • LoneWolf15
    • 5 years ago

    Thanks, guys, for the article. Interesting and appreciated.

    To me, this indicates one of several things I think AMD needs to do -become as SoC with their CPU/APU units as possible.

    One of the biggest drawbacks I see with AMD in the past number of years isn’t just performance-per-watt; it’s that when you have OEMs using your processor in the budget segment, you have them cutting corners on the infrastructure. You have to be concerned that your processor is going to look bad because the vendor put it inside of a box/laptop with a cheap/cheesy mainboard, a poor wireless card, etc.
    The more control you exert over this by bringing components onto your chip, the less you have to worry about your product being crappy because of the vendor. I really hope that at some point, AMD looks at doing some wireless as well (akin to Intel’s Centrino-style branding) to ensure they aren’t dependent on whatever Broadcom or Qualcomm-Atheros can sell the cheapest.

    I’m excited to see what the new offerings do in the mobile market. We don’t need them to be the fastest –but if they’re “good enough” in a reasonable price range, I’d love to own an AMD device again.

    • sschaem
    • 5 years ago

    Nice presentation on the Carrizo data Scott.

    My take away. Lots of work did go into this, and many steps are commendable by AMD engineering. .. but… yep a but, I’m not convince much of all this will translate into much better CPU performance.

    I think efficiency improvement will be visible (battery life),
    yet I dont expect any miracle for the CPU side of a 35w carrizo compared to a 35w kaveri.

    Then again, thats only a problem really when you compare benchmark numbers with Intel,
    for a most users this class of CPU is fine. (if the performance was to low, Intel wouldn’t be selling atom or pentium / low end i3 in mobile devices)

    So, if the battery life of a 28nm carrizo laptop is within 20% (net/office profile) of a 14nm broadwell AMD might have some customers. If not, this AMD R&D effort would have been in vane.

    My hope here is that OEM stayed clear off kaveri for some technical reasons, and carrizo fixed all their concerns, and AMD got contracts pending to make this design refresh worthwhile.

    I just want AMD to have some money to live another day so they can release at least a 490x 🙂

    • slaimus
    • 5 years ago

    I feel like this large Carrizo core has very little chance to find a design. 25-35W consumer laptops have all but disappeared, except for expensive business models. The ones for sale are either 13-17W laptops with Ultrabook processors, 7-10W budget laptops with netbook processors, or 47W quad core desktop replacement notebooks. Carrizo-L is already slotting into the Ultrabook and netbook range.

    • cmrcmk
    • 5 years ago

    Has there been any discussion of number and type of ports that hang off the integrated south bridge? If this is aimed at only tablets and laptops, I wouldn’t be surprised if it only has 2x SATA III and 4x USB3.

    I’d love it if this had something more like the full desktop compliment of expansion ports: 6x SATA III and 6x USB3 (plus a smattering of USB 2 and 1.1 as well). It’d be great for a low power HTPC or NAS without requiring extra controllers.

    • ronch
    • 5 years ago

    If you go to [url<]http://www.amd.com/en-us[/url<] and scroll down to the bottom, there's a section about Mark Papermaster's tech discussion. He looks tired and his hair looks a lot thinner, like a vampire sucked all his juice. I guess working on Zen and Chorizo can really suck the life out of you.

    • ronch
    • 5 years ago

    I find it a bit curious as to why AMD would go through all the trouble to practically re-implement the latest incarnation of the fundamental Bulldozer architecture using high density libraries. Yes, they’ve been working on it since Piledriver came out but is it really worth it? And doesn’t it distract them from focusing on Zen and stretch them a little thinner? And does anyone really seriously think this will put a dent on Broadwell-U sales? Unless benchmarks and power efficiency tests show that Chorizo blows Intel’s similar offerings out of the water, I reckon Chorizo will be nothing more than a niche lineup.

    Then again, maybe AMD understands the demands of its customers better but as a longtime AMD fan I’m not sure APUs are really going anywhere, at least not when they’re still using Bulldozers derived CPU cores.

    Edit – I see people down thumb my seemingly anti-AMD posts lately. Not hating on AMD, just airing out some real concerns here.

      • VincentHanna
      • 5 years ago

      If nothing else, it means more chips per silicon wafer. More chips per wafer means less cost per chip.

        • ronch
        • 5 years ago

        Making more chips per wafer that few people want to buy is futile. I think they should have opted to produce faster chips that more people would not only actually want to buy, but pay good money for them.

        This is exactly what costs AMD an arm and a leg these days. They wanted to produce a compact CPU core to maximize the number of chips per wafer. Given AMD’s historical process node disadvantage against Intel, this looked like it made perfect sense. However, in doing so, AMD seems to have focused too much on making a compact core, leading to the idea of shared resources which killed per-core performance, so AMD just hoped that software developers will have learned to utilize a ton of cores for everything from gaming to transcoding by the time Bulldozer comes out. Guess what happened.

          • sschaem
          • 5 years ago

          Do you have access to AMD cost analysis documents?

          AMD doesn’t get the chips for free.
          30% smaller chips directly translate into 30% less cost to acquire the dies.

          Also AMD had to update the die to make Kaveri remotely competitive.
          Because AMD cant sell Kaveri in this market, and they cant wait another 2 years for a new architecture.

          So AMD got a stop gap solution that greatly improve efficiency for mobile devices.
          This saving greatly increase the value of the chip to OEM over kaveri.
          (Those chips are TDP limited, and this was the true performance ceiling.)

          Also realize that the market doesnt need speed deamon for all products.

          Look at ultrabooks. the APU that goes int hose are pretty slow if you just look at relative benchmark numbers.
          Also why would Intel even bother making a 800mhz Broadwell U if speed what all that matter?

          Reality is even low end AMD APU are plenty fast for “90%” of laptop models.
          The issue is TDP/battery life. and carizzo is a big step forward (on paper)

          So for this market the CPU performance threshold has been reached…
          AMD is right to focus on efficiency and TDP, their module provide plenty of horse power already for the common windows application and OS.

          AMD does a LOT of stupid things, but carrizo is not one of them.

            • ronch
            • 5 years ago

            Spending a bit more on die space is worth it if it means you can sell the chips for twice or thrice the price, isn’t it? I imagine the Bulldozer core is roughly as big as a Sandy Bridge core but sacrifices too much. I think the current situation tells it well enough.

      • deruberhanyok
      • 5 years ago

      Not the first time they’ve done something like this. Remember the switch to VLIW4 for a handful of parts in the HD 6000 series, only to then launch GCN in the HD 7000 series?

    • deruberhanyok
    • 5 years ago

    I don’t doubt AMD’s engineering capabilities.

    I do doubt whether we’ll actually see this technology in the marketplace in any kind of reasonable timeframe.

    With the desktop Kaveri parts the launch was just really screwed, so much they had to second launch 6 months after the first.

    With these… if they’re not available on time, then OEMs won’t pick them up, and they’re left with a technology that only makes it to market in super expensive machines (the kind that they like to use to say “look what our tech can do!”) or super cheap machines (the kind that have crap quality 1366×768 TN displays, 2GB of DDR3-1333, mechanical hard drives, battery life under 3 hours and still weight 7 pounds because it’s a solid block of plastic).

    I really hope that doesn’t happen – I’d even like to see embedded desktop solutions using the higher-power versions of this – but I’m not holding my breath.

      • _ppi
      • 5 years ago

      Mind you, when they are two process generations back compared to Intel, even if they were CPU architecture masterminds, they just cannot realease competitive product. At all.

        • deruberhanyok
        • 5 years ago

        Idunno, for the price, I think their current offerings are competitive. They were just impossible to find at retail for a while.

        As software gets further optimized I think you’ll see that they can even beat Intel in price/performance with current tech – early D3D12 tests have shown an A8-7600 to be keeping pace with an i3-4330 with a discrete GPU:

        [url<]http://anandtech.com/show/8968/star-swarm-directx-12-amd-apu-performance[/url<] So if, for example, Windows 10 was out today and games were making use of D3D12, the A8 would be the better deal - it's significantly cheaper. But it isn't, so my point is just: Even if they stay two process nodes behind, as software is better optimized and if they keep delivering small but steady performance gains, then yes, I think they can release a price-competitive product.

    • I.S.T.
    • 5 years ago

    So… Is the 8350 still going to be AMD’s main high end CPU(Not counting the extreme edition style 9350 or whatever it was called) for all of 2015? It hasn’t been updated in nearly two and a half years, despite Steamroller coming out for the Kaveri models…

      • Deanjo
      • 5 years ago

      This surprises you why? The AM3+ platform hasn’t seen a chipset update in 5 years.

      • maxxcool
      • 5 years ago

      *2017* fixed that for you ….

    • tanker27
    • 5 years ago

    I must be hungry because I keep reading the name in my head as Chorizo.

      • derFunkenstein
      • 5 years ago

      Chuckula has been calling it that all along, so that probably doesn’t help.

        • chuckula
        • 5 years ago

        I’M IN UR HEADZ!

      • anotherengineer
      • 5 years ago

      *googles Chorizo*

      hmmmm kinda looks like Mr. Hankey…………..

        • tanker27
        • 5 years ago

        You really should try some. The most common form of Chorizo found in the U.S. is mexican. It goes great with scrambled eggs or as a spicy alternative filler for burritos or tacos.

        The other one I love is the Spanish (Spain) version which is very spicy. Its a bit harder to find but some specialty grocery stores do carry it.

          • anotherengineer
          • 5 years ago

          Thanks.

          I like spicy things, I will take a look next time I’m at the deli in the grocery store, I find spicy things less common up here in northern Canada.

    • the
    • 5 years ago

    I’m really curious at what clock speeds this design will be shipping at. I wouldn’t expect a reduction from Steamroller in mobile but the Bulldozer linage was designed for high frequencies. I’m impressed that AMD has been able to utilize it as well has they have in mobile but it is mainly a desktop/server part. And as alluded to in the article, they likely cancelled the desktop parts as it wouldn’t be hitting the same frequencies at the highend. The IPC improvements would be nice but could be negated by the lower clock speeds. AMD needed to do both on the desktop.

    Three features on the last slide stand out to me: integrated voltage regulation, advanced bandwidth compression, and interframe power gating. The IVR is just surprising I thought we’d see with a more mobile focused part. I’d be able to further reduce power consumption due using similar techniques that Intel adopted in Haswell. The bandwidth compression is likely what was used in Tonga. Carrizo’s GPU could be in a weird spot between Hawaii and Tonga in terms of capabilities. If the GPU doesn’t have the same compression as Tonga, I’d expect more of an evolutionary jump in GPU performance and nothing surprising like Tonga fitting in between the R9 280 and R9 280X. Then again, this could be an attempt to apply similar memory compression to the CPU or IO side of things. That would be interesting. Lastly, interframe power gating sounds like a technique used by nVidia’s Battery Boost technology. Essentially synching GPU workloads to a lower frame rate that runs in sync with the display. IE no need to waste power trying to render a frame that wouldn’t be displayed if V-Sync is enabled. Or it could mean something else entirely.

    Overall Carrizo doesn’t sound like a bad part. The worst thing about it is when it’ll be released: middle of 2015. It really needed to be a late 2014 product. AMD is positioning it as a Broadwell competitor and with Broadwell’s delays, it very well could have been. However, Intel appears to be rather aggressive with Sky Lake’s release and not letting the Broadwell delays affect it. Thus shortly after Carrizo arrives, we could be seeing Sky Lake systems. Even if AMD’s hype lives up to the claims of Carrizo being a Broadwell competitor, Intel will have leaped a generation ahead.

    • K-L-Waster
    • 5 years ago

    In principle, this sounds like what AMD needs to sell to ensure the company remains viable. It’s not what enthusiasts want (based on the posts in the GPU articles it sounds like the most vocal AMD fans want the computing equivalent of a 500 cu. in. V8 with twin super chargers, gas mileage and handling be damned…) but something like this is what will actually sell enough units to be profitable.

    Now, the big question is whether or not the actual delivered silicon is able to achieve what they say it will. Hopefully yes, but only time will tell.

      • dragontamer5788
      • 5 years ago

      What AMD needs is their marketing team to sell chips in general to OEMs.

      I think AMD in general has decent APU offerings (I have a Richland laptop, and it came at a great cost/performance ratio especially after I upgraded its hard drive to an SSD). Obviously they’re inferior products to Intel, but modern laptops / devices are more about SSD, RAM and other components.

      I think the consumer has too much emphasis on the CPU right now however, which makes these chips a hard sell. The “i3/i5/i7” brand is very strong.

      • Essence
      • 5 years ago

      The question is; can AMD deliver it on time (I doubt it), and if so; what will Intel competing hardware prices and performance be like!

      It depends on how it compares with competing Intel gear, as has always been the case.

    • anotherengineer
    • 5 years ago

    Nice report, slides, info, etc.

    However,

    Will we see this? I have a hard time finding anything here in Canada with Kaveri, while Beema is a little easier to find in a notebook.

    It doesn’t matter how awesome something is if it’s vaporware, cause you can’t sell or make a profit on vaporware.

    I hope we see it though, and in good numbers, options in the market are always nice.

    • sweatshopking
    • 5 years ago

    I’M FROM THE FUTURE GUYS!
    HERE’S WHAT HAPPENED:
    AMD MADE A NEW CHIP, AND IF IT LAUNCHED A YEAR AGO WOULD HAVE BEEN STELLAR! BUT IT DIDN’T, AND SO IT WAS TOO LITTLE TOO LATE AND NO OEM’S USED IT! IT COULDN’T COMPETE WITH INTEL’S STUFF, AND SO WAS LARGELY WRITTEN OFF EXCEPT WITH A FEW CHEAP LAPTOPS, AND SINCE THEY WERE CHEAP, THEY SUFFER FROM POOR BUILD QUALITY AND THINGS LIKE MECHANICAL HARD DRIVES. CUSTOMERS WERE DISAPPOINTED AND DECIDED TO BUY INTEL NEXT TIME CAUSE THAT WAS LIKELY THE DIFFERENCE!

      • chuckula
      • 5 years ago

      In a world where case-sensitivity have been outlawed… one poster will find a way to save us all!

      • Deanjo
      • 5 years ago

      HAVE THE LEAFS WON THE CUP YET?

    • Anovoca
    • 5 years ago

    Glad to see AMD engineers innovating ways to do more with less, if only I could make the same sentiment for their board of directors.

    • dodozoid
    • 5 years ago

    I am somehow a fan of AMD. I tend to expect great improvements from them. I look forward to every piece of news about their upcomming products. I own an AMD R9 290 (with ivy i7 CPU though).
    But with every iteration of their CPU design i keep hearing “The next one will definitely kill Intel”
    Its allways: “Steamroller will solve this!”
    When it came out it turned out to be a mere evolution, a decent one, but still an evolution with a promise the mighty Excavator will change the game and slay the blue dragon.
    And here it is… another evolutionary step with a promise the mighty Zen will change the game.
    I realy hope THIS time, their new CPU arch will finaly live up to its hype because it is probably the last try AMD has.

    Sorry for weird english…

    • Klimax
    • 5 years ago

    If they have any brains those GCN improvements are already present in 300 series in all chips to be released. (equivalent of Fermi 2.0)

    As for their catching up with Intel? I would talk hop when they do jump. Their track record suggest that optimism is not warranted. That goes for both Carrizo and Zen.

      • nanoflower
      • 5 years ago

      I can believe they’ve made the improvements mentioned in the article since they’ve long talked about the tradeoffs of designing for high performance vs low power. However it does sound like they are suggesting this new Carrizo will be great when it will only be great if power usage is of overwhelming concern as this new design isn’t going to perform as well at compute tasks as Broadwell. At GPU related tasks I can see it beating Broadwell but in the mobile area how many people are sticking only with IGP/APU AND are concerned about GPU performance? That seems to be a very small market for this product.

        • xeridea
        • 5 years ago

        [quote<]in the mobile area how many people are sticking only with IGP/APU AND are concerned about GPU performance?[/quote<] Anyone who wants to do light-medium gaming but doesn't want to spend $1000 on a laptop.

          • Klimax
          • 5 years ago

          Intel won’t be that far behind.

          Anyway, we’ll see…

    • Wirko
    • 5 years ago

    Not sure what to make of it: in the power saving feature matrix, every “in market” feature ends in 2013 or mid-2014. If we leave out the past and the future, is AMD in the market or not?

      • ronch
      • 5 years ago

      It’s an AMD PowerPoint slide! It’s supposed to be vague and confusing but at the same time exciting!

        • Deanjo
        • 5 years ago

        [quote<]It's an AMD PowerPoint slide! It's supposed to be vague and confusing but at the same time exciting![/quote<] And in AMD history, amended many times going forward.

    • Meadows
    • 5 years ago

    Looks pretty good at first glance, and I can only wonder how much better it would’ve looked at a smaller process node.

    An interesting word caught my eye on the power efficiency image too. What the hell was “Tigris”? I don’t remember that one.

      • ronch
      • 5 years ago

      Tigris was AMD’s laptop platform back in 2009.

    • ronch
    • 5 years ago

    [quote<]AMD claims Carrizo is "the first processor in the world with HSA 1.0 support," referring to its Heterogeneous Systems Architecture effort to enable converged CPU-and-GPU computing. That claim is a bit confusing since AMD said something similar about Kaveri, which it touted as the first architecturally complete HSA development platform.[/quote<] They are now also saying that Kaveri couldn't really use all of its 8 GPU clusters concurrently and it had to fall back to using just 6 of those CUs to keep the chip from blowing up (exaggeration). (Source: Extremetech.com) So why did they put in 8 GPU compute clusters in Kaveri if they're not gonna be used concurrently constantly anyway? Know what? Why is AMD like this? It's like every time they have a new product or new iteration of a product, they reveal new information about the previous product that wasn't clearly revealed when it came out. It's usually about a limitation. For example, I don't remember reading anything about only 6 of those 8 CUs in Kaveri being able to work concurrently for any length of time when Kaveri came out. In fact, they advertised it as a 12-core APU (4 CPU cores + 8 GPU cores)! Also, when Kaveri came out it was generally publicized that it was the first fully-HSA-compliant APU, only to be dispelled nowadays when Carrizo is making headlines, saying Carrizo is the first HSA 1.0-compliant APU (so Kaveri was fully compliant with what?). And this: Since 1999 the fundamental K7 architecture was publicized and believed to have 3 ALUs and 3 AGUs. Guess what? When Bulldozer came out AMD then said that one of those AGUs was included only to simplify chip layout, so there were really, effectively, 3 ALUs and just 2 AGUs. Glad that's cleared up... after 12 years. I don't know if AMD is/was deliberately keeping information from going public during product launches to help sell chips (saying Kaveri is fully HSA compliant when it is in fact not, to help make it look more compelling), just making old products look worse to help make the new product look better, or AMD is just being run by monkeys who couldn't even get simple information across themselves right (remember when they miscommunicated Bulldozer's transistor count?). Edit - Joel Hruska, the author of the ET article, replied to my comment, which is an almost exact replica of what I said here. Joel: HSA 1.0 is just a marketing bit. There's no published or claimed difference. Also, it was K10 that had the third AGU that was superfluous, not K8. The 6 vs. 8 GPU bit is interesting, but it wasn't fully spelled out if the eight GCN cores ever functioned or not. Edit 2 - Joel Hruska cleared it up a bit in his reply to my comment. From Joel: [quote<]I've just realized that I wasn't as clear as I should've been regarding the state of the CU units in Kaveri. AMD's 17W and 20W APUs are *sold* as having 6 CUs or fewer. In reality, all eight CUs are present on the die, but two are fused off for power-saving purposes. With Carrizo, AMD is saying it can enable all eight CUs at lower TDPs. So no -- they didn't mis-state the APU's graphics capabilities at low TDP -- they just built hardware on die that they couldn't activate at their lowest TDPs.[/quote<] So that's that.

      • Klimax
      • 5 years ago

      And people were lambasting NVidia for incorrect count of L2/ROPs…

      • NTMBK
      • 5 years ago

      They were talking about mobile Kaveri. 95W Kaveri was perfectly capable of using all 8 CUs.

        • chuckula
        • 5 years ago

        The GTX-980 is perfectly capable of using all 4GB of RAM so the complaints about the GTX-970 are unjustified…. oh wait.

      • chuckula
      • 5 years ago

      Interestly quote from that extremetech article:

      [quote<]One other major change actually addresses an issue in the lowest-power Kaveri cores that we weren’t previously aware of. Some of AMD’s high-end mobile cores had eight GPU clusters for a total of 512 cores — but didn’t actually enable all eight Compute Units (CUs) at the same time. These systems are effectively limited to six CUs to keep the core from overheating. Carrizo removes this limitation, which should substantially improve performance.[/quote<] If this were Nvidia...

        • Lans
        • 5 years ago

        [url<]https://techreport.com/review/26528/a-first-look-at-amd-kaveri-apu-for-notebooks[/url<]

          • chuckula
          • 5 years ago

          Yes, I read it the first time when it came out.
          Note that the issue raised by Extremetech is [b<]not[/b<] that some models come with fewer CUs enabled, but that the FX-7600P appears to only really be able to use 6 CUs at any one time when AMD advertised that 8 CUs are available. It also makes sense when you look at the rumored GPU gains of Carrizo over Kaveri: Carrizo actually has all the hardware turned on, which logically leads to a performance boost.

            • Lans
            • 5 years ago

            I haven’t seen actual proof in the Extremetech.com article that I could find via google with your quote and it doesn’t even list FX-7600P (“[s]ome of AMD’s high-end mobile cores” only). It does make such claim but there is no pretty pictures. But without any proof or detail to go on, it could be just something on how the thermal budget is used like no CPU turbo for graphically intensive task and only CPU turbo if less than all CUs are used. Maybe there is a bomb shell to be dropped. but we’ll have to see.

            I would give you, if proven, it would be like the GTX 970 where benchmarks won’t change but spec is misstated.

        • techbot
        • 5 years ago

        This has gone too far really, one would expect posters in a tech forum to actually have a clue, but apparently this is asking for too much!

        The low power mobile Kaveri variant, FX-7500, is advertised as a 6 GCN part, because, guess what? It can only run with 6 CUs at this TDP, contrary to the 35w FX-7600P which utilizes the 8 CUs and is advertised as such!

        Sorry for crashing your party but no Kaveri part is falsely advertised when it comes to active CU count.

          • chuckula
          • 5 years ago

          Wrong: I was talking about the 7600P: Advertised as having 8 CUs but apparently only 6 worked at a time.

          I’m a little tired of the “this has gone too far” crap from people who don’t even know what an ROP is but act like Nvidia committed war crimes with the GTX-970.

          If AMD lied about the FX-7600P they need to be held just as accountable as Nvidia… now if you want to be a non-hypocrite and say that the issue with the GTX-970 is no big deal then I’m OK with that, but I’m REALLY over the whole fanboy schtick of acting like literally everything Company A does is pure-evil while literally everything company B does is pure Unicorn-rainbow lovefest.

            • techbot
            • 5 years ago

            No, you are wrong, it is super clear they were talking about the high end low power parts, which is the highest FX part in the 19w TDP range.

            A quick look at any of the graphics benchmarks will tell you the 7600p works as advertised, I cannot fathom how you can come to such a wierd conclusion.

            As for the 970, I have no stance on that, I think suing Nvidia over this is a bit ridiculous.

            • sschaem
            • 5 years ago

            Yes, if AMD lied about the FX-7600P. NO ONE is arguing about setting different standard here.

            But so far all this seem to be a miss quote and this is where the argument is, not about “AMD doesn’t have to answer to anyone, but nvidia does”

            So you can save your typing and start to link to facts that the 7600p is a 6CU part.

            You want some fact ? check
            [url<]http://www.pcper.com/reviews/Processors/AMD-Kaveri-Mobile-Preview-AMD-FX-7600P-Performance/PCMark8-and-GPU-Performance-Co[/url<] The 35W FX- 7600P is FASTER then a45W a8-7600 , even so the FX-7600p got a lower GPU clock. Cloud gate graphics : 9981 (8cu 686mhz) vs 9166 (6cu 720mhz) If the FX-7600p had 6cu, it would score closer to 8000 not 10000.

          • ronch
          • 5 years ago

          From the Extremetech article:

          [quote<]One other major change actually addresses an issue in the lowest-power Kaveri cores that we weren’t previously aware of. Some of AMD’s high-end mobile cores had eight GPU clusters for a total of 512 cores — but didn’t actually enable all eight Compute Units (CUs) at the same time. These systems are effectively limited to six CUs to keep the core from overheating. Carrizo removes this limitation, which should substantially improve performance.[/quote<] "... didn’t actually enable all eight Compute Units (CUs) at the same time." If a particular SKU was advertised as a 6-CU part, there's no way AMD will allow all 8 CUs to work together, period. But the way ET tells it, it seems to indicate that 8 CUs were advertised but they don't all work together all the time. Doesn't it remind you of Nvidia's recent issues?

            • Antimatter
            • 5 years ago

            AnandTech phrases it differently:
            [quote<]The high density, power optimized design also plays a role in the GPU segment of Carrizo, offering lower leakage at high voltages as well as allowing a full 8 GCN core design at 20W. This is an improvement from Kaveri, which due to power consumption only allowed a 6 GCN design at the same power without compromising performance.[/quote<] Perhaps Extremetech are misinterpreting what AMD said, or is AnandTech mistaken?

            • nanoflower
            • 5 years ago

            Well a quick google turned up this quote “At 35W and with four CPU cores allied to eight GPU Compute Units, the FX-7600P is well positioned to bring mainstream performance, general-purpose computing and gaming capabilities to affordable mid-range laptops.” Which I believe came from AMD.

            [url<]http://hexus.net/tech/reviews/cpu/70653-amd-fx-7600p-28nm-kaveri/?page=8[/url<] Also look at AMD's own web page where they say 4 CPU + 8 GPU for the FX-7600P and FX-7500. [url<]http://www.amd.com/en-us/products/processors/notebook-tablet/fx-apu#[/url<] (Yes they do put the caveat there of up to but that means that either the FX-7500 has 8 GPU cores or they played a marketing game of saying up 8 when they knew it could never be more than 6 in actual usage. Kind of what Nvidia did being technically correct in saying the 970 has 4 GB of memory available but not going into the limitations.)

            • Zizy
            • 5 years ago

            7500 is advertised as 4C+6CU. This up to is a classic marketing BS. Check any 7500 material and you will see 4+6 => 10 cores marketing.
            7600P is advertised as 4C+8CU and has all 8CU working.

            AnandTech is probably correct here. 8 on die is unlikely. Most of the models have 6CU so there is almost surely 6CU die. Harvesting functional 8CU and dropping 2 CU to make 7500 is illogical, as all good 8CU are needed for 7600P, which is a higher end model of the two.

            Not that it really matters as no laptops come with either processor.

            • techbot
            • 5 years ago

            There is only one single SKU advertised as with 8 CUs, only one, the high end is the FX series, which has two SKUs, one 35w with 8 CUs and one 19w with 6 CUs.

            Some journalist explained that in bad language or even didn’t understand it well so let’s all go and call it a scandal

            in revenge of the 970 nvidia issue maybe? Since you keep bringing it up in the when I didn’t even mention it? Or even care as I explained earlier?

            • sschaem
            • 5 years ago

            It seem to be a really bad attempt at deflecting attention.

            Because REPUTABLE sites (like PCPer) have shown that the 35w kaveri FX-7600p ***with a slower GPU clock*** beat a 45w kaveri 6cu desktop part.

            So I’m not sure why all this benchmark data become irrelevant because some tech blogger cant understand AMD presentation….

            • sschaem
            • 5 years ago

            No benchmark support this … I have to wonder , why in hell are you spreading BS like this ?

            • ronch
            • 5 years ago

            I’m not spreading BS, I’m basing this on the Extremetech article. I usually check out TR and ET for my tech news. Besides, even excluding the 6/8 CU issue, don’t you think I’m still right when it comes to what AMD did with K7 info (2 AGUs) and HSA support on Kaveri?

        • NeelyCam
        • 5 years ago

        Yes, but no lawyer would launch a class-action lawsuit against AMD, because AMD doesn’t have enough money to settle.

          • chuckula
          • 5 years ago

          And here I though it was because AMD [b<][i<]never settles![/i<][/b<]

      • Lans
      • 5 years ago

      “first architecturally complete HSA development platform” vs. “HSA 1.0 support” I guess that could be confusing but to me is like “beta release” and “public release”. The former is saying all the features is there but might be buggy while latter is saying all the features have been tested and working to spec.

        • ronch
        • 5 years ago

        So they went right ahead and told the world that Kaveri is the first HSA-compliant chip? Also, if it had a beta/buggy implementation of HSA, doesn’t that make Kaveri not officially fully HSA-compliant?

          • Lans
          • 5 years ago

          Ok, maybe bad explanation/software analogy but I would say “architecturally complete HSA development platform” should mean it has all the HSA features implemented and working but why would they call it “development platform” as opposed to “HSA 1.0 support”? I don’t know the real reason but my guess would be sub-optimal implementation because they do claim “architecturally complete”. Perhaps they had/have bugs which caused them to use sub-optimal solution to produce the same result but slower/less efficiently?

          Anyhow, I agree it can be confusing but I don’t see problem with “first architecturally complete HSA development platform” for Kaveri and “the first processor in the world with HSA 1.0 support” for Carrizo.

            • ronch
            • 5 years ago

            Architecturally complete is good enough to be a development platform even if there are bugs to work out? That’s a bit like jumping the gun, isn’t it? Seems there’s a fine line between ‘full HSA support’ and ‘HSA 1.0 complaint’, but AMD didn’t tell us about it. That’s a bit disingenuous in my book.

      • derFunkenstein
      • 5 years ago

      That probably explains why the A8-7600 and the A10-7850K have very similar game performance with the integrated GPU.

        • Damage
        • 5 years ago

        I really don’t think so. I think the low-power Kaveri parts for ultrathin laptops and such were restricted to 6 CUs for power reasons, and that doesn’t have to happen at 15W in Carrizo.

        I think the reason higher-power Kaveris aren’t much faster with 8 CUs than 6 CUs mostly has to do with sharing two key constraints in a CPU socket: the same TDP and the same amount of memory bandwidth from dual channels of DDR3.

          • derFunkenstein
          • 5 years ago

          OK that makes sense, too. I just never could figure out why the difference is so small, and jumped to a conclusion.

      • sschaem
      • 5 years ago

      So much anger over misreading information….

      How about you check benchmarks ? you might see that a lower clocked 35w kaveri with 8cu does indeed beat at graphics a higher clocked 45w 6cu kaveri desktop part.

        • ronch
        • 5 years ago

        Look, Joel of ET admits that he wasn’t clear about those CUs. Did I misread the information? Should I really read in-depth on Kaveri or Carizzo? People are focusing too much on the 6/8-CU but I sure hope you guys get my point. AMD has the habit of miscommunicating information. Are you ok with that? Personally I find it disingenuous. I don’t like AMD’s marketers, they’re giving AMD a less than stellar image.

          • ermo
          • 5 years ago

          In fairness, I think AMD is giving AMD a less than stellar image. It’s just that their marketers aren’t exactly helping!

          [quote<]"(...) I don't like AMD's marketers, they're giving AMD a less than stellar image.[/quote<] You have to wonder if Lisa Su being at the helm will result in changes to how AMD approaches designs and the engineering culture backing them. So far, Carrizo looks promising enough all things considered. Someone made the point that AMD have recently designed chips which have been very hard to actually manufacture and ramp up, both in volume and performance. Could someone in here shed some light on this? I've never really thought about it this way before...

    • odizzido
    • 5 years ago

    Looks neat. I’d like to see some lower power chips for light fanless systems though.

    • tbone8ty
    • 5 years ago

    Can someone make a gaming specific laptop with these apu’s in them please? Let’s go marketing department!

    Judging by how long it took kaveri and temash to be seen in actual mobile products I sure hope carrizo fairs better.

    • Airmantharp
    • 5 years ago

    You know what jumped off the page for me?

    AMD expects to still be around in 2020!

      • brucethemoose
      • 5 years ago

      Of course they will be around!

      But who will own them?

      • Essence
      • 5 years ago

      We have had the same rubbish being spouted for the last decade or so, and yet AMD is still here (Most of it anyway). I worry more about Nvidia to be honest, as they do not have an X86 licence and are competing with Apple, Samsung and Qualcomm etc, and they are much larger (big money) companies then Nvidia.

      There is a reason Nvidia are making their own tablets, and they have my total respect for that. AMD could learn a few tricks from Nvidia to be honest, but I doubt that very much since its AMD (Run by Monkeys in suits) were talking about here .

    • chuckula
    • 5 years ago

    Seeing AMD’s design choices, I’m pretty impressed with the results as long as you remember that Carrizo is a mobile-oriented chip. AMD got strong density and pretty strong power efficiency.

    It’s also blatantly obvious why Carrizo never became a socketed desktop processor: the exact same design choices that help it in mobile also hobble its abilities to ramp up in a higher power evelope system.

      • DancinJack
      • 5 years ago

      I’m sorry, but how can you say that when we haven’t even seen concrete, third-party numbers yet? PowerPoint slides are pretty and all, but we have seen promises from everyone, including AMD, before and then fail to deliver.

      It all looks promising, but they really do need to succeed on the implementation.

        • chuckula
        • 5 years ago

        I’m willing to give AMD the following:
        1. At 35 watts they’ll have a very strong GPU (20-30% faster than the 7600P).
        2. At any wattage the cpu will be pretty anemic, but that’s almost a given.
        3. The GPU performance should scale down ok at lower wattages although the cpu may end up being weak enough to be the limiting factor in many practical gaming situtations.

          • sschaem
          • 5 years ago

          for 3.) Dx12 to the rescue?

      • _ppi
      • 5 years ago

      25-35W is quite a lot for current laptops (especially if you want ultrabook competitor).

      However, with its all power-saving stuff at idle, it could be very interesting HTPC chip.

        • UnfriendlyFire
        • 5 years ago

        Some of the Iris Pro CPUs have 28W TDP, and I’ve seen 13″ laptops with those ultra expensive CPUs.

        But I wouldn’t expect the OEMs to give the same treatment to AMD, and rather stick those in 15.6″ or 17″ plastic laptops.

        • Zizy
        • 5 years ago

        Depends on the laptop. Those larger ones with 40cm screen diagonal should work very nice within this power bracket. Most of those come with low power Intel stuff and NV GPU. If AMD offers a single chip solution that is as good for gaming and better for movies, it could get lots of recommendations.
        But who am I kidding, like we have ever seen a non-crap AMD laptop. Beer on the pattern to repeat yet again.

    • JosiahBradley
    • 5 years ago

    I’m just really excited for HSA. The future is fusion :b And remember ARM is behind HSA too and the road maps have ARM cores planned as well.

      • Klimax
      • 5 years ago

      And yet best results usually are on Intel CPU/IGP combinations. (In OpenCL best with AMD drivers…) Terribly weak CPU will kill you no matter what gimmick you try.

    • Sam125
    • 5 years ago

    It’s decision time for me. I want a new laptop and Carrizo is positioned for the 25W and 35W category laptops. I do a moderate amount of gaming so hybrid crossfire will more than likely be in the cards for me. So it’s either Carrizo or wait for Zen and/or next year’s or 2017’s APU. I don’t like buying new tech things more than every few years so this is a tough-y. Guess I’ll wait for the reviews.

      • brucethemoose
      • 5 years ago

      Hybrid crossfire isn’t all it’s cracked up to be. APUs are best used by themself, so you can get decent graphics without a dGPU sucking power when it’s on (and sometimes when it’s off).

      Personally, I’m not touching bulldozer with a 10 foot pole. I’m waiting to see if Zen is a worthy Llano successor, as that’s when all the big changes are coming.

        • NTMBK
        • 5 years ago

        Hybrid Crossfire got better with Kaveri. It added XDMA, the “secret sauce” out of Hawaii which makes Crossfire over PCIe with no Crossfire bridge not suck, and makes frame delivery much smoother.

        • Sam125
        • 5 years ago

        I’ve been following hybrid crossfire for notebooks since Llano and there are definite benefits for crossfire in laptops. That’s the main advantage for AMD’s A-series for mobility. The last time HP had a hybrid crossfire setup with a Trinity saw nearly a 50% boost in scores with 3DMark and various gaming benchmarks. That’s a respectable increase for ~$100 it costs to add a discrete graphics card.

        In all honesty though, it depends on what HP offers. I only buy HP laptops so they’d need to offer Carrizo with hybrid crossfire before I’d be able to buy one. I’m keeping my fingers crossed since it seems AMD has a respectable offering for the last BD APU. So…yeah, there’s a good chance I’ll end up just waiting for Zen and APUs derived from it, but Carrizo is at least intriguing.

      • smilingcrow
      • 5 years ago

      If you have the money current best option for mobile gaming is Intel + nVidia and this isn’t going to change that. Wait for next year if you must go the AMD route.

        • Sam125
        • 5 years ago

        Yeah, one could go the Intel+nVidia route as money isn’t really a huge issue for even higher end laptops from companies that don’t gouge the buyer. However, I’ve been buying AMD for 11 years now and I don’t really intend to stop because I don’t have a compelling reason to.

        I could’ve switched to Intel a long time ago but haven’t, simply because I don’t want to. It makes my life a bit more difficult but meh. I never follow the crowd anyway.

    • Geonerd
    • 5 years ago

    My first reaction is that unlabeled graph axes (“Typical-use energy efficiency”) are the work of the Devil… or perhaps the AMD marketing department.

    Miserable DDR3 bandwidth is still going to choke graphics and, to a degree, OpenCL performance.

    With all that saved space, would it have absolutely killed AMD to toss in another one or two x86 modues? At this point, ANY up/side-grade path from AM3.x would be welcomed with canonfire and trumpets.

    Before I get too excited about the hardware improvements, SHOW ME THE CLOCKS!

      • yuhong
      • 5 years ago

      Hopefully the compression support in Tonga will help, though lack of DDR4 is unfortunate.

      • Flatland_Spider
      • 5 years ago

      Clocks: [url<]https://encrypted.google.com/search?tbm=isch&q=clocks&tbs=imgo:1[/url<] You can get excited now. 🙂

      • Antimatter
      • 5 years ago

      I don’t think additional x86 cores would really help as Carrizo is targeted towards laptops and is unlikely to arrive on desktops.

    • mako
    • 5 years ago

    Minor typo: ‘reduce is operating speed’.

    • DragonDaddyBear
    • 5 years ago

    If it is as good as it seems (not always the case with Team Red) then maybe there will be some decent SFF or convertibles that are produced using this APU. One can dream, right?

      • dragontamer5788
      • 5 years ago

      These look like they’re for Carrizo, not Carrizo-L. Fortunately, Carrizo and Carrizo-L share the same socket now, so maybe SFF or convertibles will benefit (Temash / Kabini).

      Otherwise, I’m thinking this is all still about “20W+ Laptops”.

    • Deanjo
    • 5 years ago

    [quote<]In Naffziger's words, "We optimized for a range and knocked it out of the park."[/quote<] Pardon me for taking those words with a grain of salt. AMD over the recent years have always made big claims only to let down the faithful upon release.

      • DancinJack
      • 5 years ago

      No pardon needed. You are on point.

      • chuckula
      • 5 years ago

      Carrizo blows away Broadwell in every dimension.

      [url<]http://www.youtube.com/watch?v=pSw2GTLHtQo[/url<]

      • the
      • 5 years ago

      Oh, AMD has been knocking it out the park. The problem is that they’re playing baseball in a football stadium: the competition has moved on to a different game.

        • K-L-Waster
        • 5 years ago

        Actually, they’ve been knocking it out of the park, but then realizing that due to all the cost cutting they now only own *one* baseball — so now they have to leave the park and go find the darn thing….

          • LoneWolf15
          • 5 years ago

          Deflate-AMD-Gate? 😉 Yes, yes, I’ll show myself out.

Pin It on Pinterest

Share This