Tiny Radeon R9 Nano to pack a wallop at $650

When AMD announced its lineup of graphics cards based on the Fiji GPU, the firm said it would eventually be offering two distinctive products in addition to the Radeon R9 Fury and R9 Fury X. One of those cards will be a dual-GPU monster in the vein of the Radeon R9 295 X2, and the other is an ultra-compact powerhouse known as the Radeon R9 Nano.

Today, AMD is pulling back the curtains on the R9 Nano, revealing a full set of specs and some preliminary performance numbers for this pint-sized overachiever. These cards aren’t expected to hit online store shelves until next month, on September 10, but we have quite a bit of info to share today.

The basic appeal of the Nano is its unique combination of power and size. Thanks to a compact physical footprint made possible by HBM and by the Fiji GPU’s modest power consumption, the R9 Nano crams Fury-like performance into a card that’s just six inches long. AMD says the Nano is well-suited for 4K gaming, and it expects to see Nano-based PCs built around the Mini-ITX standard for compact systems.

Here’s a quick look at the Nano’s full specs, according to AMD:

The big news here is the fact that the Nano packs a fully enabled Fiji GPU, with the 4096 stream processors and the same 1GHz peak clock speed as the Radeon R9 Fury. The HBM-based memory subsystem has the same 4GB capacity and 512GB/s of bandwidth as the Fury’s, too. Although this card is small in stature, its GPU is by no means cut down.

Perhaps the most eye-popping number above is the Nano’s typical board power of just 175W. That’s fully 100W lower than the R9 Fury’s power spec, which is kind of bonkers. And it means the Nano requires only a single eight-pin power input.

AMD says it expects Fury-like performance from the R9 Nano, which may seem like some kind of dark voodoo magic. In reality, though, the Nano likely benefits from one of the most prominent realities in high-performance chips today: in order to eke out the last five to ten percent of performance, one often has raise power consumption dramatically. Power draw is determined by frequency times the square of voltage times the number of transistors cycling. As a result, even a relatively modest decrease in voltage and clock speed can work wonders.

In order to keep power draw in check, AMD has almost surely chosen to run the ultra-wide Fiji engine at somewhat relaxed clocks and voltages. The formula could pay off. The firm claims the R9 Nano is “up to 30% faster” than the Radeon R9 290X while being based on a “40% shorter”board.

I’d be careful not to focus too much on the “up to 1000MHz” clock speed spec in the table above, though. AMD has got to be capping the Nano’s power consumption using its PowerTune algorithm, which decides from moment to moment how fast the GPU can run based on a number of inputs, including the GPU workload, temperature, power limit, and so on. As a result, with a 175W limit, the Nano is likely to run at somewhat lower delivered clock speeds than the R9 Fury with its big, beefy cooler and 275W cap.

One of the firm’s reps told us the Nano’s clock speed will vary according to the workload. For some things, we might see 900MHz. In other cases, the GPU “might even peak to 1000MHz.” Nevertheless, AMD projects performance “more or less” like the R9 Fury’s. We’ll have to get our hands on a card in order to quantify the true meaning of that fuzzy statement.

The Nano’s product manager was also careful to dispel the notion that “this card is going to throttle.” I’m not sure what “throttling” versus “not throttling” means under the control of a complex algorithm like PowerTune, but the Nano should be well-equipped to keep itself cool under load. The card is expected to have a 75°C typical operating temperature, which AMD points out is 20°C lower than the R9 390X’s operating temp.

Cooling duties are handled by a hybrid setup that combines a vapor chamber base with multiple heatpipes snaking into horizonal fins. It’s nothing exotic as these things go, but it should suffice for a 175W card. Notably, the Nano’s cooling solution has a separate heat pipe and set of fins meant expressly to cool the card’s power-delivery circuitry, something the liquid-cooling setup on the Fury X seemed to neglect.

The R9 Nano is claimed to be “library quiet” at 42 dBA, which AMD says is 16 dBA quieter than the R9 290X. Also, unlike the Fury X, the R9 Nano is open for modification by board vendors, so we could see custom versions from the folks at Sapphire, Asus, and so on. AMD tells us some of its partners are prepping designs with water blocks for liquid cooling.

The R9 Nano’s distinctive combination of size and competence won’t come cheaply, though. AMD says the Nano is a sort of co-flagship with the R9 Fury X, so it has given the Nano the same price: $649.99.

I expect the scarcity of R9 Fury and Fury X cards has something to do with the Nano’s healthy price tag. AMD has no reason to ask less if it’s selling all of the cards it can produce. The firm says it is “saving up quantities” of the Nano ahead of its launch date in the hopes that the supply will be sufficient to meet demand.

About those performance numbers

Although R9 Nano cards won’t be in the hands of reviewers until a later date, AMD has supplied some of its own internal benchmark results comparing the Nano against its closest competition, a mini-ITX version of the GeForce GTX 970. AMD is claiming 30% faster performance than the GTX 970 ITX.

I don’t doubt the Nano could outperform the GTX 970, but the results above come with some serious caveats.

You will recall that AMD also released benchmarks for the R9 Fury X ahead of the reviews, and those numbers showed the R9 Fury X consistently beating the GeForce GTX 980 Ti across a range of games. But when we measured the Fury X’s performance ourselves, the numbers told a different story:


In the case of 99th-percentile frame times, the story was dramatically different.

The reason for the difference in FPS averages was clear when we looked back at the footnotes of AMD’s document supplying the numbers. The firm tested the cards with a very particular formula in order to achieve its results for the Fury X—and it has carried over that formula when generating the R9 Nano results you see above. The basic approach is to test exclusively in 4K with high-quality shader effects and post-process anti-aliasing, but to soft-pedal on texture filtering and mulitsampled edge anti-aliasing. 

Here’s how AMD says it tested the games in the Nano benchmark results above, based on the footnotes of its presentation:

  Display

resolution

Quality

preset

Shader

AA

MSAA Anisotropic

filtering

Battlefield 4 3840×2160 High FXAA 0X 0X
Crysis 3 3840×2160 High 0X 0X
Far Cry 4 3840×2160 High SMAA 0X 0X
GTA V 3840×2160 High 0X 4X
Shadow of Mordor 3840×2160 High 0X 0X
The Witcher 3 3840×2160 High 0X 0X

That’s a pretty weird combination of settings, all things considered. 4K is one of the highest resolutions you can get in a display today, and yet we’ve compromised dramatically on image quality via low-rent texture filtering and edge AA methods.

Here’s a quick illustration of the difference in texture quality between 0X and 16X anisotropic filtering:

Without anisotropic filtering

With 16X anisotropic filtering

The difference in quality between post-process AA methods like SMAA/FXAA and more traditional methods like MSAA isn’t quite so dramatic, but post-process AA methods aren’t aware of sub-pixel geometry, so they sometimes get things wrong. For example:

FXAA (left) cannot resolve the fine geometry that a sub-pixel-aware technique like MSAA or SSAA can (right)

I can’t imagine building a high-end gaming rig, getting a 4K monitor and a fancy video card for it, and then tweaking games to compromise image quality in such basic ways.

So why is AMD testing in this fashion? Probably because it plays to the Fiji GPU’s strengths—that enormous shader array—while not leaning so hard on its potential areas of relative weakness, like ROP throughput (for MSAA resolve), texturing (for anisotropic filtering), and small triangles (since polys are relatively smaller at lower resolutions). Taking this peculiar path likely puts the R9 Nano in the best possible competitive light. In the absence of independent reviews, those AMD-provided numbers are sure to circulate widely online. Trouble is, AMD could be setting up the R9 Nano to be a disappointment once it gets into the hands of reviewers and consumers.

So there you have AMD’s performance numbers with some proper context. My advice: wait for the reviews before you decide to jump on an R9 Nano. It may well be a very good product. Heck, it looks super-cool. But manufacturer-provided performance numbers often tell a different story than one might hope.

Comments closed
    • Mr Bill
    • 4 years ago

    I just realized that Sapphire makes an ITX Compact R9 380. although, it has only 2GB. Its of a size to go in a mITX review.

    • NoOne ButMe
    • 4 years ago

    Well, average of 25-26% faster than the 970 at 4K. From techPowerUp* test.

    AMD’s 30% is spot on if you exclude games that have heavy tessellation. Looks like AMD was honest with performance.

      • Mr Bill
      • 4 years ago

      Did somebody get a peak at the review?
      “Potential Catastrophe: Inside-The-Nano, Second TR Review”

        • NoOne ButMe
        • 4 years ago

        No, I just started typing “tech” on my phone and had to do something else. Came back and wrote TechReport instead of TechPowerUp.

    • ThorAxe
    • 4 years ago

    I refuse to buy a new GPU without HDMI 2.0.

      • Mr Bill
      • 4 years ago

      DisplayPort

    • BaronMatrix
    • 4 years ago

    The UltraBook of GPUs…?

    • Mr Bill
    • 4 years ago

    You guys are just funny. TR posts a roundup of GTX 980Ti cards and it gets 36 posts. But here you all are at 168 posts still arguing vehemently about whether the nano is a worthy opponent at any price.

      • Meadows
      • 4 years ago

      The roundup is meaningless. The main review of the 980Ti received 269 comments.

        • Mr Bill
        • 4 years ago

        +++

    • bhappy
    • 4 years ago

    AMD as a whole has devolved into a bad joke as a company. With a management team and marketing department that seem to exist in an alternate reality. Does AMD really think that consumers are stupid enough to believe their marketing slides again, when their benchmarks have repeatedly shown in the past and present to not reflect real world gaming conditions that gamers would actually use in games? Who would actually buy a mid-high end graphics card and turn AF completely off? Even in their best cherry picked benchmarks its not dramatically faster than its competition which is roughly half its price. They also forgot to mention its lack of HDMI 2.0 which means for 4k gaming on a 4k TV it will be almost completely useless unless that TV supports DP which rarely any TVs do. If this card was priced competitively to its competition I could see a case for it but at close to double the price of its competition, the only market for this card would be AMD fanboys that would buy anything with a AMD logo on it.

      • TopHatKiller
      • 4 years ago

      getting there… almost,… yep, and nope… somepeople; if performance was lacking…that’s not true, but continue… well; hdmi2 lack is weird [i think the design is older then amd wants to admit] there are a bunch of quad-hd tv’s that don’t support hdmi2 – they’re dumb too…. uhm, at this size? uhm? you’ve lost me there.
      [Where’s my bourbon?]

    • Arclight
    • 4 years ago

    If it’s the full chip the only thing they spared money on was the cooler and the slightly smaller PCB (which could either be more or less expensive to manufacture, idk). Given these facts, it should at least cost less than the Fury X which has a way more expensive water cooling. Still, it’s difficult to argue that it should cost about the same as a GTX 970 for example since, again, it’s the full chip onboard. Nobody would make the same demands if nvidia did a simillar product with the full GTX 980 ti chip.

    • Johnny Rotten
    • 4 years ago

    As one who was left atx and matx and fully embraced mini-itx, I definitely appreciate the sentiment but ultimately this card is a miss for me. Fact of the matter is ive got a full sized 980GTX inside a small Node 304 with 3 SSD drives and a 600W power supply. The whole system is about the size of a toaster. And the thing is, theres LOTS of space inside. You’re starting to see more console sized cases too that *still* can accommodate a full size card plus a sfx power supply so I’m not sure how much longer there is going to be a need for a mini-sized super card.

    • TheMonkeyKing
    • 4 years ago

    Awaiting January 2016 price drops and if they are significant enough to place it better in the rankings…

    • K-L-Waster
    • 4 years ago

    I’m going to go out on a limb and guess that HBM is still expensive to manufacture compared to GDDR5. AMD usually prices their products aggressively — the fact they are not suggests that they may not have any room to do so.

    Since HBM is the only thing in the Fury that is a major departure, seems reasonable to suspect it is the cause of the extra cost.

    Presumably as production ramps up for HBM economies of scale and process improvements will kick in and drop those costs to a more manageable level.

      • Ninjitsu
      • 4 years ago

      I doubt such a large GPU in itself is cheap either…

        • NoOne ButMe
        • 4 years ago

        AMD has historically had better yields on silicon than Nvidia. Both overall and for similar sized silicon.

        It ought to cost no more than the GM200 silicon does for NVidia. Given they use the same 28nm process variant from TSMC. Given a square die, AMD even should have 1-2 more dice per wafer.

          • Ninjitsu
          • 4 years ago

          Yeah, but I doubt GM200 is cheap either!

            • NoOne ButMe
            • 4 years ago

            Yes, but it’s also no as expensive for these parts are everyone seems to think. Given the immaturity the first 3-6 months of the 7970 that silicon probably costed more than GM200 or Fiji does now.

            With binning Nvidia/AMD are likely getting at least 70+ Dice out of a 80-90 candidates. I imagine given maturity and how big customers they are their costs are under $100/die.

            • Ninjitsu
            • 4 years ago

            Point being that if GM200 and Fiji cost the same, AMD has less wiggle room to cut prices thanks to HBM.

            That’s all I’m saying. I don’t know what it costs them in absolute terms.

          • K-L-Waster
          • 4 years ago

          The numbers suggest Fiji should be competing with GM204, not GM200. GM200 *is* expensive, but so are the products based on it (980 TI and Titan X).

            • NoOne ButMe
            • 4 years ago

            What numbers? Thanks.i

            • PixelArmy
            • 4 years ago

            [url=https://techreport.com/discussion/28912/tiny-radeon-r9-nano-to-pack-a-wallop-at-650?post=933571<]This guy did the math and concluded performance would be ~GM204.[/url<]

            • NoOne ButMe
            • 4 years ago

            Yes, using AMD’s numbers (cut about in half). And, that’s for the weakest Fiji card.

            That’s due to the lower clockspeeds. Of course, with a heavy overclock GM204 IS competitive with Fiji cards. Although, at that point the efficiency tends to be lower.

            Overall, Fiji is nearly competitive with GM200 at stock and noticably faster than GM204 at stock.

    • USAFTW
    • 4 years ago

    You know what they say… The smaller your graphics card,
    The higher your frame latency.

    • maxxcool
    • 4 years ago

    … 650$ ??? ummm … SLI two Gigbyte 970 and SAVE 50$ ??

    • AJSB
    • 4 years ago

    For a *true* mini-ITX build, this (should be) the best card money can buy. Period.

    Meanwhile, Green Team drivers for Windows 10 seem to be destroying monitors like a kid eating candy:
    [url<]https://forums.geforce.com/default/topic/862417/windows-10-official-353-62-drivers-are-killing-samsung-and-lg-notebook-lcd-display-panels/[/url<] Have a nice day :") Edit: I believe that i can reach a new world record of downvotes >:))

      • chuckula
      • 4 years ago

      Notice how the issue revolves around AlienWare?

      Think about the plausibility of your theory that every GPU from NVidia is intentionally designed to destroy monitors because apparently that makes Nvidia rich or something vs. the plausibility of Alienware sucking.

      Hint: The plausibility of Alienware sucking is about 100%.

        • AJSB
        • 4 years ago

        LOL but no.
        The portables were all running well for a long time (some already out of warranty) with ZERO issues under W7 and W8.x with respective NVIDIA updated drivers, so, NO, it can’t be a hardware issue per se …the s**t hit the fan when installed W10 and new NVIDIA specific drivers.

        NVIDIA drivers must be doing now something unusual that is affect specific *brands* and *models* of monitors.

        As for revolving around AW, didn’t read all pages to see if other OEMs are affected or not…it would also be important to check what other OEMS might be using *same* LCD models and if they are affected or not and only then you can say if AW is c**p or not.

      • K-L-Waster
      • 4 years ago

      [quote<]Edit: I believe that i can reach a new world record of downvotes >:))[/quote<] You need to bust out an NGreedia if you want to do that....

    • ish718
    • 4 years ago

    1% of PC gamers will be pleased with their purchase.

      • remon
      • 4 years ago

      According to steam survey, 3.97% of its users have a 970, which is currently the best selling card.

    • Firestarter
    • 4 years ago

    so, given that AMD would have had to do some binning to get dies with good thermals, these might end up being pretty decent overclockers, right? I mean, I don’t think it would make much sense as you’d have to get a custom (water)cooler anyway and the power circuitry might blow up, but at least the silicon itself should be of good quality

      • NoOne ButMe
      • 4 years ago

      the dies that overclock the best typically have worse thermals/etc. Look up stuff surrounding the 390/390x. They’re leakier (as a whole) versions of the original Hawaii chips with some minor modifications to the memory and possibly some other things.

    • wiak
    • 4 years ago

    well every manufacturer will use different settings, just read *reviewers guides or benchmark results from nvidia* they are also different from review sites

    • Krogoth
    • 4 years ago

    The last exciting GPU has been the 970 and arguably the 980.

    Everything else since then has been underwhelming and/or overpriced.

      • sweatshopking
      • 4 years ago

      the only thing exciting about the 970 was the 3.5 fiasco. mediocre 28nm card sticking around too long.

        • NoOne ButMe
        • 4 years ago

        it undercut AMD quite a bit and brought a high end performance at $330. at low power.

        It could easily be overclocked (Raising power a noticeably amount) to be a ultra-high end performer.

        It was very impressive.

        • Krogoth
        • 4 years ago

        290X and 780Ti level of performance for ~$300 when it launched while consuming less power and space.

        3.5GiB issue was Nvidia’s marketing dropping the ball. The engineers and developers knew about it and in the grand scheme of things. It doesn’t matter since by the time the 970 is forced to use more than 3.5GiB of VRAM. The fill-rate and shading power becomes a bottleneck.

          • NoOne ButMe
          • 4 years ago

          No, the 3.5GB issue was just “we don’t think anyone will notice”.

          That’s a HUGE PROBLEM. Technologically speaking, fully agree. It is a non-issue in terms of performance.

          (edited to clarify)

            • Krogoth
            • 4 years ago

            Wrong, the problem is that Nvidia’s Marketing failed to disclose the problem at 970’s debut while the rest of the team knew about it.

            Nvidia had a similar issue with GK106 chips (650-660) and their asymmetrical memory controller setup. The white papers and marketing were forthcoming on the issues that could arise with such a setup at launch.

            If Nvidia’s marketing was just as forthcoming with 970’s launch none of the stupid 3.5GiB debacle would have happened.

            • NoOne ButMe
            • 4 years ago

            Yes, and, they did make comments about the 650/660 regarding that. Nvidia told at least one major site BEFORE the reviews went up.

            If you seriously think that no one who knew the specifications at NVidia read any reviews previous to someone finding out about it via a bandwidth testing program than you’re being silly.

            Nvidia choose to stay quite. Either they didn’t see it as an issue or didn’t care. Well, or no one who knew the specifications read any reviews and realized that no one mentioned that in the specifications or the writing.

            I’m guessing it’s one of the former two. Maybe a combination of them. As, technically speaking, it really doesn’t matter.

            • Krogoth
            • 4 years ago

            It was the [b<]marketing team[/b<] that was out of the loop until the public discover the issue and they were forced to make an addendum on the 970.

            • NoOne ButMe
            • 4 years ago

            And you’re saying that no one who was in engineering/management/etc who knew the real specs of the card read a single review?

            Either they didn’t think it was important, or didn’t want to tell anyone.

        • TopHatKiller
        • 4 years ago

        huh. the 3 1/2meg fiasco wasn’t exciting… just another example of nv talking thru its own ass.

      • Ninjitsu
      • 4 years ago

      980 was overpriced because they could, but 970 was fairly disruptive and has remained so, in my opinion.

        • Krogoth
        • 4 years ago

        That’s why I said 980 was arguable.

        It did change the game somewhat by offering superior performance then 780Ti in most stuff and situation for less power and space. Same that Nvidia decided to price it at $599 at launch instead of $499 because it would have perfectly mirrored the 670/680’s launch.

          • NoOne ButMe
          • 4 years ago

          Huh, the 980 launched at $550…???

            • Krogoth
            • 4 years ago

            It didn’t stop etailers from marking them $599 though. They are hard to get for a while to for those who wanted them.

      • Freon
      • 4 years ago

      I think the 980 Ti is pretty remarkable. It’s price/performance ratio compared to the 970/980 is actually still quite good, not that typical exponential decay we see as you spend more and more money.

      I almost wonder if NV didn’t shoot too low on the price in fear over the Fury. If they put $749 on the box I don’t think it would’ve been out of line. Still ~$250 less than a Titan X and it’s more or less traded for 6GB of RAM you don’t need anyway.

      I at least give credit to AMD for pushing HBM and giving it a shot, if perhaps slightly prematurely. It could have been much worse.

      • HisDivineOrder
      • 4 years ago

      I agree. I think the best days of GPU’s are when great performance and a great price come together. When you get into the high priced, uber-elite gear, it’s neat and all, but you’re talking gear few’ll get to use.

      I’m more excited by tech that most of us mere peons will get to use and that is still remotely impressive.

    • Pez
    • 4 years ago

    Look at the little thing ain’t it cute **tickle**

      • ImSpartacus
      • 4 years ago

      I want one just to put it on my wall. It’s adorable.

    • Mr Bill
    • 4 years ago

    Too bad Mork is not around for the product launch. Can hardly wait for TR to break out the nanometer and benchmark this with DX12.
    [url<]https://www.youtube.com/watch?v=PAFx0_-FvXo[/url<]

    • fhohj
    • 4 years ago

    Good writeup Scott.

    I was interested in this card because it’s tiny, and I love it, but it seems it’s too pricey for its performance. I guess that was pretty likely but it’s here now.

    • LoneWolf15
    • 4 years ago

    Packs a wallop, yep. Right in the wallet.

    EDIT: Ninja’d.

    • Thbbft
    • 4 years ago

    Taking a page from Nvidia – don’t leave money lying on the table.

    Why not $650, if they’re supply constrained and are selling all the Fury and Fury X cards they can produce, they will be able to sell all the Nano cards they can produce to the SFF crowd. Adjust prices as necessary.

      • HisDivineOrder
      • 4 years ago

      Except the SFF crowd is more likely to value HDMI 2.0 as a forward-thinking feature for 4K…

    • brothergc
    • 4 years ago

    “Radeon R9 Nano to pack a wallop at $650”

    Maybe the headline should read :
    “Radeon R9 Nano to pack a wallop on your wallet at $650 ”
    LMAF

    • Chrispy_
    • 4 years ago

    I had a lot of success downvolting and downclocking my old 290X back in the day, there was a forum thread where some people showed similar power-savings whilst others did not.

    I guess if you cherry pick chips that are efficient, you can get 90% of the performance for only 60% of the power consumption, precisely because AMD have already pushed their silicon beyond the voltage/clock sweet spot to get the clocks they have. Dialling back to the sweet spot delivers lower overall performance but much better performance/Watt – which is what mITX form factors are all about.

      • Mr Bill
      • 4 years ago

      I thought it might be a binning thing. I’d like to see where the mITX GTX-970 plots on the performance vs price graph.

        • Chrispy_
        • 4 years ago

        Well, it would only be fair to test the Nano against an mITX 970 anyway.

        If it’s not mITX length, it ain’t even in this race.

    • Silus
    • 4 years ago

    This is clearly connected to supply. That’s why they want to charge so much for a card that’s slated to perform a bit worse or on par with a $550 card (the R9 Fury)

    Plus, they are comparing it to a $300-350 card ? (the GTX 970) …lol
    Yeah, the Nano sure better beat the GTX 970….

    AMD keeps surprising in the worst possible way.

    • DrDominodog51
    • 4 years ago

    Was I the only person who noticed that board partners are able to modify the nano? The difference between the Nano and Fury X is only clockspeed. I bet board partners will be creating full sized nanos with great cooling and similar clockspeeds to the Fury X.

      • Meadows
      • 4 years ago

      Sure, for $800. No thanks.

      • Shobai
      • 4 years ago

      I think the Nano’s down to 4 supply phases, from 6 on the Fury X. Reckon we’ll see Nano variants with 6 phases?

        • DrDominodog51
        • 4 years ago

        Sure. I think that’s a PCB design thing not a chip limitation.

    • Meadows
    • 4 years ago

    The card looks very appealing to me, but the price is ludicrous.

    I’m also not convinced about its performance if they were able to reduce power consumption by a whopping 36%.

      • Mr Bill
      • 4 years ago

      But, it can reach ludicrous speed.
      [url<]https://www.youtube.com/watch?v=ygE01sOhzz0[/url<]

    • Meadows
    • 4 years ago

    [quote<]"small triangles (since polys are relatively smaller at lower resolutions)"[/quote<] Might be just me, but I was under the impression that the [b<][i<]size[/i<][/b<] of the polygons is completely inconsequential.

      • LordVTP
      • 4 years ago

      You’d be wrong there. The size of a triangle (onscreen, in terms of pixel space) is directly proportional to the raster process cost (consumes fillrate)- Quick video of slowed down triangle raster

      [url<]https://www.youtube.com/watch?v=IqhRrySbRDs[/url<]

        • Meadows
        • 4 years ago

        Ah yes, wasn’t thinking of fill rate at the time but polygon throughput.

    • shank15217
    • 4 years ago

    The big unknown is DX12 performance and recent benchmarks with DX12 enabled games show AMD in a much better light thanks to the same reasons its not as strong as Maxwell in DX11.

    • shank15217
    • 4 years ago

    So basically 2 of these in crossfire is what AMD’s flagship should have been. Maybe slightly lower clock to hit a 300 watt TDP.

    • beck2448
    • 4 years ago

    Tech Report does a great job looking inside the numbers as always.

    • HisDivineOrder
    • 4 years ago

    A swing and… a miss.

    AMD, why do you keep making these obvious and stupid mistakes that anyone can see are going to hurt you? It’s like you’re a drug addict family member and we gather around you and we tell you, “Hey, man. We need you. We love you, man. Just c’mon, dude. Stop doing the crack. It’s bad for you. Just put it down…”

    But you won’t stop. We need you to stop, AMD. If not for your own good, for ours. We need you to stop being stupid. Please.

    At this point, I’m begging.

      • Duck
      • 4 years ago

      AMD… not even once.

      • fhohj
      • 4 years ago

      Wow that was harsh. I think I just heard the sound of you being listed somewhere.

      • TopHatKiller
      • 4 years ago

      Don’t bother begging. There’s no point – amd’s brilliant engineers will continue in a miasmic delusion of incompetent marketing.
      ‘Nano’ should’ve been released with ‘fury x’. Months ago when both would have seemed superb counters to GM200.
      Still, let’s remember hmb1 was actually designed years ago… Kaveri was meant to include it, [if not before that design] but amd wanted to save money… so didn’t.
      Nv have released three discreet designs this year… against amd’s one. Again money.
      ….Over the last three years plus amd have cancelled more designs then I’ve had hot dinners [possibly] ——
      Marketing Moron: ‘let’s save money by not releasing anything new!’
      Board Member: ‘Brilliant sir! I promote you to the board!”
      Engineer: ‘Excuse me; we seem to be selling the same thing over and over? Isn’t that a bit of
      an issue?’
      Exec.Financial Moron: ‘No! It’s not. And I’ll sack you too, if you continue with this defeatist
      propaganda!’
      …2016 better see multiple zen/gpu discreet designs that are amazing. I’m just hoping amd are saving their fire for then…

        • Meadows
        • 4 years ago

        I don’t think NVidia have been discreet at all.

          • K-L-Waster
          • 4 years ago

          “Nv have released three overly flamoyant designs this year”?

          • TopHatKiller
          • 4 years ago

          That’s amusing.

            • TopHatKiller
            • 4 years ago

            You people are downvoting me for saying someone else’s joke is amusing?
            That’s very sad and pathetic.

            reasons for downvoting me
            [1] You didn’t like it
            [2] You didn’t think it was funny
            [3] You didn’t understand the joke
            [3.1] You didn’t understand it was a joke
            [4] You disagreed with my opinion
            and———————
            [5] You’ve agreed with others of your kind to exclude me
            [5.1] You think this is: highschool/secondary school/uni/college
            and ‘trials-by-fire’ or ‘initiation rituals’ are you’re thing, children.
            [5.2] You’ve found that “d-v’ing” and insulting newbies
            produced prestige to the clique of posters you’re trying to find favour with.
            [6] My non-de-plume irritates capitalists
            [7] You’re a banker
            [8] You’re a lawyer
            [8.1] You’re a lawyer soliciting for work
            [9] You’re a banker who used to study law
            [9.1] you’re a lawyer but changed ‘profession’ from a banker
            [10] You work in the editorial team of PN Review,
            and have somehow found out my real identity
            —————————-
            [11] None of the above
            [12] All of the above… and then some.

            • Meadows
            • 4 years ago

            Calm your tits, friend.

            • TopHatKiller
            • 4 years ago

            Uhm. [1] I am male… so no tits, here abouts. Sadly. Very sadly. [2] ,,, I’ve become so depressed about my life I can’t be bothered to add a second point.

            • Meadows
            • 4 years ago

            Worry not, friend. You’ll find a nice boyfriend and things will look up again.

        • NoOne ButMe
        • 4 years ago

        Kaveri wasn’t designing with HBM in mind. Kaveri was delayed over 6 months from the already 12mo+ delay. Kaveri DOES have GDDR5 support. Same with XV I believe.

        There was an AMD product that was earlier than Fiji that would have had HBM though, yes.

          • TopHatKiller
          • 4 years ago

          Sorry. But yes it bloody was. ‘Carrizo’, too. I except the, what they call it?, seventh generation apu, up for next year will be ‘carrixdfgdfkdkloooo’ [whatever] on am4 socket with hbm.
          Unless I’m wrong. Which, occasionally, has been know to happen.

            • NoOne ButMe
            • 4 years ago

            AMD knew that HBM wouldn’t be pref/cost sensitive enough to work with those parts. They have GDDR5 memory controllers that if enabled would allow you to easily feed the whole CPU+GPU.

            AMD’s counting of “generations” is terrible. I *think* it goes:
            Bobcat, Llano, Trinity, Jaguar, Richland, Kaveri, Carrizo. That, or it starts with Llano and doesn’t count bobcat.

            • TopHatKiller
            • 4 years ago

            Worser then that… rumours that had genuine believability:
            kaveri hex core/512sp w/hbm tertiary cache… quad core/768sp w and w/o hbm – both killed
            kaveri with ‘sideport’ gddr5 on mobo as a cost compromise – killed [possibly by board partners]
            There is a grave somewhere, the size [insert massive field] of killed amd designs over the last few years.

            • NoOne ButMe
            • 4 years ago

            Those Kaveri’s may have been on a product roadmap, but, they died the moment that Bulldozer launched… If not earlier.

            The parts that were actually continued to be worked on feature GDDR5 controllers. You could have GDDR5 dimms. No need for sideport. Not sure if would work if you tried GDDR5 and DDR3. My understanding is only one set of controllers could function at once.

            • TopHatKiller
            • 4 years ago

            Bulldozer, Vishera 2012?, Kaveri 2014?… I was gonna argue, then I thought…..AArghhh!!
            —“Sir!, our men! They are all dead. Quite dead, Sir! I’m sorry. None of them survived the battle, sir… Sir! Don’t cry! The war is perhaps, not yet lost…..”

            • NoOne ButMe
            • 4 years ago

            Kaveri was originally scheduled as a 1h2013 product. Maybe 1Q even.

      • bfar
      • 4 years ago

      They come up with great ideas, and then they struggle to make money out of them. Seems to be an execution issue, rather than an engineering one.

      I personally think this is an excellent product, there’s a lot of interest in ITX style gaming PCs out there. It’s just a bit expensive unfortunately, but then I’ve been saying that about GPUs all year.

        • HisDivineOrder
        • 4 years ago

        The lack of HDMI 2.0 really limits the appeal of any R9 Fury-based card in the ITX/HTPC/SFF arena.

      • ronch
      • 4 years ago

      AMD marketers are top notch. /s

    • swaaye
    • 4 years ago

    It’s going to be interesting to see if these stay in stock. The Fiji chip seems to be very low volume. If there’s no inventory why bother futzing with aggressive pricing? Project a weird unavailable prestige instead!

    • Demetri
    • 4 years ago

    This is not far off from what I expected. These are full Fiji chips right? The only significant cost savings to manufacture this vs the Fury X would be from the air cooler, and maybe less robust power delivery components.

    So really, the complaints aren’t about the Nano in particular, it’s about the entire Fury lineup and Fiji not living up to performance expectations for the price. Price cuts and new drivers are about all we can hope for until 14nm hopefully takes a dump on this stuff next year.

    • Vergil
    • 4 years ago

    [quote<]The big news here is the fact that the Nano packs a fully enabled Fiji GPU, with the 4096 stream processors and 1GHz peak clock speed as the [s<]Radeon R9 Fury[/s<][/quote<] [b<]Radeon R9 Fury X[/b<]

      • Damage
      • 4 years ago

      Fury X is 1050MHz. Fury is 1GHz.

        • Ninjitsu
        • 4 years ago

        Yeah but the X has 4096 stream processors, so I can see why that sentence may have been confusing.

          • Damage
          • 4 years ago

          Ah, true. Modified the wording a bit.

    • Damage
    • 4 years ago

    Radeon pricing team follows the R9 Nano’s launch: [url<]https://vine.co/v/ejQd0iZdBBU[/url<]

      • Mr Bill
      • 4 years ago

      ++

    • brothergc
    • 4 years ago

    $650 yea right , is that with or with out vasoline ?

    • anotherengineer
    • 4 years ago

    Well since most of these cards have dual-bios today (1- legacy & 1-UEFI) it would be interesting to see if someone can/will flash a FuryX with Nano bios.
    [url<]http://www.techpowerup.com/vgabios/index.php?architecture=ATI&manufacturer=&model=R9+Fury+X&interface=&memType=&memSize=[/url<] Then you could have a hi-perf mode or a power efficient mode with the flip of a switch. (if it works) It's what AMD should have done in the first place with the fury X to make it more marketable.

    • mad_one
    • 4 years ago

    Nvidia would have an easy time matching the power efficiency, but the space efficiency might be a real advantage if you want a tiny form factor. They should be able to use an 980 with a bit of TDP reduction on those 970 designs though.

    As for power efficiency, I agree that you pay a heavy price for the last few percent. The 390X is an extreme example of this, but even the fairly modest GTX970 in overclocked vendor designs exhibits this behaviour. I dropped the power limit by 20% on the MSI 970 Gaming 4G for noise reasons (poorly ventilated case and I consider abovementioned 42dBA to be very loud) and performance dropped by less than 5%.

    • xeridea
    • 4 years ago

    At 4k resolution, I am not sure how much AA really matters, given that the pixels are so tiny. Except for things like the shown ropes, it would be hard to tell. The ropes may or may not be as stark a difference as shown because of high res, were those screenshots from 4k or 1080p? Perhaps they should have done 2xAA as a middle ground.

    I would say a more important feature would be AF, which would be noticeable at any resolution. I could see them not doing 16x, but 4x would be a good compromise.

    Anyone know if it would be possible to have different clocks for the texture hardware and the shaders? If this could be tweaked it would help mitigate the die size limitation from using HBM, not sure how complicated it would be though.

    Either way it will still be a beastly card for the size, even if the 0.4% of those who play @ 4k need to tweak settings, which is basically required for smooth 4k gameplay regardless of what video card you use. 1080p and 1440p (the rest of us) should be easy for this card at any settings.

      • morphine
      • 4 years ago

      [quote<]At 4k resolution, I am not sure how much AA really matters, given that the pixels are so tiny.[/quote<] Hahaha... no. At least until we have 500+ PPI monitors, and even then, just maybe.

      • mad_one
      • 4 years ago

      It still matters, especially the crawling effect will be very noticable. But the reduction in detail caused by FXAA-like algorithms should be far less noticable, so there’s less reason to lose performance to MSAA (not that you can afford it 4K on any current card).

    • Ninjitsu
    • 4 years ago

    $650. Lol. Price vs performance charts should be fun.

      • wimpishsundew
      • 4 years ago

      A new metric will arise.

      Behold! Size vs performance!

      “Wow Jimmy!, You got 10k points per inch!”

        • Ninjitsu
        • 4 years ago

        Whole new meaning to “it’s over 9000!”, eh? 😀

          • Mr Bill
          • 4 years ago

          152,400,000 nanometers in 6 inches

    • chuckula
    • 4 years ago

    BUT EVERYTHING IN THE MITX MARKET IS OVERPRICED!!

    Not entirely untrue, but also not to the degree that’s being used as an excuse for the Nano.

    [url<]http://www.newegg.com/Product/Product.aspx?Item=N82E16814121912[/url<] About $50 overpriced*? Sure. About $250 - $300 overpriced? Not really. * Or about $30 after the MIR, and there's a bundled game.

      • Phishy714
      • 4 years ago

      Check the link you posted… takes you to a different website then newegg. Blocked by my anti-virus.

      *edit*

      And now its gone. Weird.

        • chuckula
        • 4 years ago

        Check your web browser, my link is perfectly fine, but if this is the level that AMD fanboys have to sink to in order to deny the existence of reality then thanks for the laugh.

          • Ninjitsu
          • 4 years ago

          I’m pretty sure they’re talking about the automatic redirect through kqzyfj.com that all newegg links on TR go through (hover the mouse pointer above your hyperlink).

      • NoOne ButMe
      • 4 years ago

      Are you trying to say the Nano is $250 overpriced? It’s ~980 level performance and about the same power. The form factor is worth maybe 50-100.

      As I’ve said, overpriced, sure, but, $250 overpriced? Can I have some of what you’re smoking =]

        • Ninjitsu
        • 4 years ago

        They’re comparing it to a 970 ITX, that means it’ll about match the 970 ITX, going by what we saw happen to the 980 Ti and Fury X comparison.

          • NoOne ButMe
          • 4 years ago

          They compared it to the current fastest mITX card. It’s performance in a segment.

          The reason why AMD hasn’t talked about performance with similar costed cards is the Nano IS overpriced (75-125 dollars given the form factor, otherwise 100-150) and would lose every single comparison.

            • Ninjitsu
            • 4 years ago

            I’d believe this theory (and it makes sense, no doubt) if they:

            A) Weren’t testing at a resolution irrelevant to 99% people.

            B) Weren’t fiddling around with AA/AF in the same way they were when they were trying to make the Fury X look better vs the Ti.

            Seeing that most HDTVs are likely to be ~720p or 1080p, the final outcome is going to be very different to what they’re representing.

        • chuckula
        • 4 years ago

        [quote<] It's ~980 level performance and about the same power.[/quote<] Yeah, we'll see about the performance part. Power can't be that far off because of the limitations in the power connectors to the card, but I seem to remember that the inside-the-second frametimes on even the Furry-X weren't much better than a GTX-980. I implicitly trust AMD's pre-fabbed performance numbers... to be complete fantasies.. and I implicitly trust TR to tell us the *real* story.

          • NoOne ButMe
          • 4 years ago

          The 980 is about 10-15% faster than the 970. I don’t think that cutting AMD’s numbers in half will get an unrealistic result. Especially when we know that the Fury performs about 5-10% over a stock 980. At 1440/2160. Looking at a wide range of a games.

          So, take 5-10% away from a stock Fury, and, you get ~980 level of performance.

          And, that’s assuming the Nano will run at about 800Mhz for games.

          [i<]downvoted for using math. Okay. Cool. I'll remember not to use math and logical reasoning next time[/i<]

          • K-L-Waster
          • 4 years ago

          To be fair, something that doesn’t quite match a 980 isn’t that bad, considering it would still outperform everything else available in that form factor (half length + not much taller than the PCI tiedown). In a MITX case, getting the GPU to fit has it’s perks.

            • chuckula
            • 4 years ago

            The Nano would be awesome… at about $450 or maybe a little less.
            At $650 it’s more of an insult than a product.

            • K-L-Waster
            • 4 years ago

            That is a fair point. Yes, it should handily outperform that MITX 970 in the comparison graphs, but price wise it’s not too far from the cost of 2 of those cards (2 x 970 would cost more, but not enormously so). 2 x 970s in SLI would outperfom the Nano if the numbers we are seeing are correct.

            ‘Course, SLI in a MITX case sounds insane, so this is prolly an academic suggestion….

            • NoOne ButMe
            • 4 years ago

            And, using *math* it should end up around as fast as a 980. With the same power (About) and the same amount of ram. Except it’s in a much smaller form factor. $0-25 over the 980 is the pricing it should be hitting.

            $50 if it is a bit faster.

      • Ninjitsu
      • 4 years ago

      [s<]970s seem to sell in the $320-340 range anyway (without rebates), so I'd say overpriced by $10...[/s<] EDIT: Ignore me, I had thought you were saying that the 970 ITX is overpriced by $50. Brainfart.

    • PrincipalSkinner
    • 4 years ago

    Comparing to a $300 cheaper card. Are they MAD?

      • chuckula
      • 4 years ago

      I’ll remember that the next time somebody screams that the i7 5775C is $200 more than a Kaveri part. [and there are a bunch of workloads where the 5775C actually wins on a price/performance basis].

        • NoOne ButMe
        • 4 years ago

        Given you’re looking for the maximum GPU performance in a CPU only/ultra SFF, yes. It’s quite valid. For regular users, it isn’t. For CPU performance, why would you be looking at AMD anyway. you might as well compare the A57 to Broadwell when your goal is to run x86 programs.

        This product (Nano) clearly isn’t worth $300 more for regular users for a noticable performance increase over the 970 mITX. For people who demand the maximum performance in smallest build, it will be worth it.

        Just like the Kaveri v. eDRAM Intel parts comparison. For a niche market it makes loads of sense. For everyone else, it’s stupid.

      • Nevermind
      • 4 years ago

      They didn’t launch at the same time.

        • chuckula
        • 4 years ago

        Yes, AMD is late to the market.

        Meaning they don’t have first-mover advantage.

        Meaning they should be charging LESS MONEY to remain competitive.

        • JustAnEngineer
        • 4 years ago

        They’re [i<]selling[/i<] at the same time, aren't they?

          • Nevermind
          • 4 years ago

          They’re selling after one has been on the market over 6 months. Apples:apples?

          No, derp.

        • Meadows
        • 4 years ago

        That’s the worst excuse I’ve ever heard.

          • Nevermind
          • 4 years ago

          You’re apparently assuming it’s an excuse for something, derp.

          The observation stands.

            • Meadows
            • 4 years ago

            All right then, it’s the most irrelevant observation I’ve ever heard.

            • Nevermind
            • 4 years ago

            Only because you made no attempt to understand why I made it…

            • Meadows
            • 4 years ago

            I’m pretty sure neither did you.

            • Nevermind
            • 4 years ago

            I’m pretty sure you jumped on my d1ck before I got past mentioning the fact…

            “That’s the worst excuse I’ve ever heard.” Lol, well to be a fly on your wall…

            • Nevermind
            • 4 years ago

            ” the most irrelevant observation I’ve ever heard. ” -Your turf, I guess?

      • NoOne ButMe
      • 4 years ago

      Comparing to the highest card in it’s class.

      If you’re aimed at mITX market, wouldn’t the obvious comparison be with the fastest mITX card there is around besides/previous to that product?

      Now, looking at the overall GPU market. AMD needed to make another comparison. Due to their pricing structure of their parts, they’re unable to make a very good one. Hence, they use the 290x for some inane reason.

        • PixelArmy
        • 4 years ago

        It would make sense if it wasn’t 2x the price. At that point, destroying the 970 is meaningless and should be a given. Basically, they tried wowing the less informed with a huge differential in the marketing material charts to justify the big price tag.

        If they priced it right, I think better marketing would be to find a comparable card and go “we perform like card X, but in mini-itx size!”. Same message, different presentation. But they didn’t price it correctly, so presenting it as I described would drawn attention to this.

        [url=https://techreport.com/news/28481/amd-fiji-gpu-rides-in-on-four-radeon-r9-fury-graphics-cards?post=915194<]I agree with the assessment it should have ~980 performance[/url<]. 980's MSRP at $480. So +$48 premium for the form factor = $528, maybe $550 max. [url=https://techreport.com/news/27443/asus-shrinks-the-geforce-gtx-970-for-mini-itx<]Mini-itx doesn't [i<]really[/i<] carry that much premium[/url<] Anyways, speaking as someone who always preferred/required shorter PCBs in the past, mini-ITX cases with risers for full cards are becoming more prevalent, so I just don't care anymore...

          • Nevermind
          • 4 years ago

          Well it’s an additional “form factor” consideration, like PSU wattage or anything else.

          If the card don’t fit you must a-quit.

    • Zizy
    • 4 years ago

    If this gives near Fury performance, why would you expect lower than Fury price? Those 100W difference are one jump for the price, mITX size is the other.
    The pricing of the card is in line with its performance, power consumption and premium paid for the small form factor, assuming AMD performance numbers at least somewhat represent reality.
    Anyway, there is no competition for the card. Another reason for its high price. If the card is really 30% faster than the only other mITX competitor, it can easily command double price.

      • Mr Bill
      • 4 years ago

      +++ Agreed, and when the nano is eventually DX12 benched it should be compared to the mini-ITX version of the GeForce GTX 970 and any other mini-ITX competition in that “slot” as it were.

      • travbrad
      • 4 years ago

      [quote<][b<]If this gives near Fury performance[/b<], why would you expect lower than Fury price? [/quote<] It doesn't. It has lower performance than the Fury and the same price as the Fury X.

    • NoOne ButMe
    • 4 years ago

    Okay, need a new comment about the price and issues surrounding it.

    Basically, AMD has shot itself in the foot when they priced the Fury at $550 and the Fury X at $650. Or, rather, Nvidia launching the 980ti at $650 before the Fury X came out shot AMD in the foot. Pick your poison.

    At this point, AMD cannot price the Fury X higher (it’s just barely fairly priced, and only thanks to the cooling), they really don’t want to drop the Fury price, AND most importantly, they cannot just price the Nano at the Fury price.

    As, it uses a full GPU. Why bother making the Nano if they could put it into a Fury X instead?

    So, they choose $650. Is it to high? Almost certainly. The price given that it’s performance is supposedly around the Fury, and it has a full Fiji, and for the ultra-small form factor really should be something inbetween the Fury and Fury X.

    [b<]I believe[/b<] that AMD refuses to price it at the same price as the Fury (for obvious reasons) but also looked into pricing it inbetween the Fury(X). They found that they didn't get more sales pricing at at $600. So, they put it at $650. Because, if you are in the market for the smallest possible mITX card with maximum performance. Well, that $50 probably doesn't matter that much. On a side note, I looked back at some numbers, the original Titan card launched (as prosumer) with about 30% performance over other single-GPU cards at 1080p, about 35% at 1440p. It costed over twice as much. In the same form factor, with much higher compute. So, if you want to look at this in purely the mITX realm of things, the Nano isn't that insane, just, AMD never pulls moves like this. Now, when you look outside the mITX world... Ouch. Terrible, Terrible Damage.

      • wimpishsundew
      • 4 years ago

      Here’s what AMD should’ve priced it at.

      Fury X $599
      Nano $550
      Fury $475

      Seriously, if the pricing of the Fury is too close to the Nano, it makes no sense to buy the Fury over the Nano. But selling the Nano too much over the GTX 970 mini version, isn’t a good idea. The Nano also need to have a decent price gap to the Fury.

      If AMD sold them at these prices, they won’t see the drop demand shortly after launch. But I’m sure AMD considered that since they have a supply shortage. Demand will surpass supply regardless if it’s slightly overpriced because they’re barely cranking out enough HBM to satisfy the demand.

      Unfortunately for AMD, they took the gamble with HBM which cause their launch to be late. Their marketing is crap, their execution is crap, their pricing in regards to performance is crap. The Nano is probably their best best product right now but they still have their massive supply shortage and it’s priced $100 higher than what people would reasonably expect.

        • NoOne ButMe
        • 4 years ago

        I would go with a flat 500/550/600, but, either way they’re all overpriced. +1

        • w76
        • 4 years ago

        [quote<]Unfortunately for AMD, they took the gamble with HBM which cause their launch to be late. Their marketing is crap, their execution is crap, their pricing in regards to performance is crap. [/quote<] The blame for this, IMO, is squarely on the executive team, particularly since it was an across the board failure; engineering, logistics, marketing. Ruiz wasn't THIS bad. Jerry Sanders, what's he up to these days, asides from being 78? Couldn't be any worse!

          • NoOne ButMe
          • 4 years ago

          It’s a shame that the CEO who really seemed to have a long term vision for AMD had to leave to ensure AMD didn’t go bankrupt.

          Well, and of course Sanders as you said, but, he retired iirc.

      • Mr Bill
      • 4 years ago

      AMD makes a little more money off the Nano…
      (1) no water cooler
      (2) lower TDP design; product binning

    • chuckula
    • 4 years ago

    OK, I’ll give AMD the benefit of the doubt: Considering that today’s “launch” is of the wood-fibrous variety, I’m going to assume there are going to be some pretty steep price cuts coming not long after the real launch occurs.

    Even a $100 price cut to put these things on the same price level as the Furry could remove a large chunk of the ludicrous factor going on right now.

      • NoOne ButMe
      • 4 years ago

      I think at $100 the Fury would never sell! Maybe Fury to $500 and this to 550 or so? And the Fury X to $600.

    • Pantsu
    • 4 years ago

    My guess is AMD is simply prepping people to accept that Fury lineup will be 500-1000$ in the future, and we just have to deal with it. 16nm isn’t going to be cheap and AMD needs have their own premium category. Besides, it’s not like it would make any sense for them financially to sell massive highly binned Fiji chips as some sort of value offering. This card will only end up in a very small niche (lol) where you simply can’t fit a 980 Ti.

    • southrncomfortjm
    • 4 years ago

    Why not also post pictures comparing lower and higher post processing and shader effects that the Nano does well? Yeah, most people will like better texture filters and higher AA but better shading isn’t a bad thing.

    • NoOne ButMe
    • 4 years ago

    I was expecting $600. Meh. $650 isn’t to bad.

    Think for a moment, if Nvidia had ~10% below 980ti 175W mITX card what would it be priced at? Probably 600-750 dollars.

    AMD did price it a little high, but, it’s nothing to crazy.

    I should note, given Nvidia can get the board to fit into a mITX size card, the 980ti would suffer well over a 10% performance hit. Memory would probably need to be clocked in the 5-5.5GBps range which appears to be about the limit before pref/watt for GDDR5 goes a bit crazy. I think that would be closer to a 20% performance hit. Still probably about equal to the R9 Nano. =]

      • K-L-Waster
      • 4 years ago

      If they can deliver on the power use and dBA numbers, and the performance is within 5 FPS of a Fury (non-X model) it will still be by far the best performing MITX card out there. Anyone looking for a console killer to attach to a 4K TV should be at least kicking the tires on this one.

        • NoOne ButMe
        • 4 years ago

        Yeah. This also should be what the next generations of high end GPUs hit. In terms of form factor. It’ll be really nice! 😀

        And, they will deliver on the power usage, Powertune caps it at 175W, probably allowing for minor spikes as long as the average is ~175W over X length of time. I don’t know the specifics of how it works that detailed.

        • shank15217
        • 4 years ago

        Except fury doesn’t support hdmi 2.0!

          • K-L-Waster
          • 4 years ago

          0_o — Ooops! Would kinda help…

      • Tristan
      • 4 years ago

      “it’s nothing to crazy” – 2x higher price for 30% more (in reality may be 20% or less) is crazy enough.

        • NoOne ButMe
        • 4 years ago

        What have highest end premium GPU prices have been for months at a time before? Titan, 30%+ for over double the price, Titan X for a few months.

        The problem is, this card will cross-over into the non-ITX market. And, it shares similar performance with AMD’s Fury cards. So, if they price this at $500 (seems a very fair price all things considered) they would have to drop their Fury/FuryX by 100 each, maybe only $50 for the Fury X.

        So, yeah. $650 isn’t to unreasonable when you consider market conditions and business decisions. Now, it still is unreasonable, as long as this card is over $50 on top of the Fury I think it’s going to always be unreasonable to some degree.

    • R2P2
    • 4 years ago

    I actually expected the Nano to be $650… but in Canadian dollars. The Canadian prices for AMD cards go roughly

    $450 – 390
    $550 – 390X
    $650 – nothing?
    $750 – Fury
    $850 – Fury X

    I figured the Nano would fit in the hole at $650, albeit with performance closer to the 290X than the Fury.

    Is it just me, or is the selection of Fury cards available really limited? Seems like there’s just Asus and Sapphire, and most places in Canada seem to only have the Sapphire, if anything.

      • NoOne ButMe
      • 4 years ago

      I’m hoping/expecting the Nano to drop down at least $50, maybe $100, and the Fury dropping about $50-75 also.

    • tviceman
    • 4 years ago

    1. It’s really annoying AMD doesn’t post minimum clock speed numbers.

    2. They are playing funny with internal benchmarks again. No one in their right mind runs AF at anything other than 16x.

    3. As with Fury X, I’m expecting great performance that will be overshadowed by AMD’s exaggerated claims in most aspects.

      • NoOne ButMe
      • 4 years ago

      1. Powertune doesn’t give you a minimum clockspeed because it’s power locked. The minimum clockspeed will be what it idles at. While I would prefer that it doesn’t cap the top clockspeed, I do like the powertune implimentation of a hard power wall compared to what NVidia has.

      2. Probably. Still, it’ll beat it’s competition in the space solidly. I’m sure someone is going to throw out some silly overclocking thing…. Buying a low power card to get quite performance in a very SFF and overclocks. Yes, that makes sense.

      3. Yeah, the performance is to low to justify the price even given the form factor I feel. $600 at the most, $550 much more reasonable.

      Cannot wait to see someone waterblock one of these!

    • TheRealSintel
    • 4 years ago

    Could be interesting to compare impact of anisotropic filtering & texture quality settings on performance of the Fury vis-a-vis Maxwell.

    Feels like we’re going backward, I thought anisotropic filtering was “a solved problem”. Next you’ll tell me they’re going to introduce the “performance/quality” or “balanced/aggressive” modes, and we’re back to angle dependent filtering (like so 10 years ago: [url<]http://www.anandtech.com/show/1205/6[/url<] )

      • Ninjitsu
      • 4 years ago

      Well, at least on Nvidia cards anisotropic filtering is never really an issue. I’ve been keeping it to 16x since 2009, when I had a 9600GT.

      Heck, even Nvidia’s own optimisation guides have shown a negligible performance impact, and they recommend that users just set it to full.

      I don’t know why AMD’s having issues with it, to be honest. It could also be the case that they’re only losing 1 fps and are like NOOOOO.

      • wimpishsundew
      • 4 years ago

      I’ve always used 16X AF on my Radeons and it barely affect performance if you have high end 7000 series or later. I really have no idea why AMD does this for PR.

      At least we all know one thing is consistent with AMD. Their PR/Marketing people are idiots for the past 40+ years.

      • chuckula
      • 4 years ago

      What’s weird about it is that they were only comparing the Nano to another AMD GPU. Has the new version of the hardware in the new Fiji GPU regressed from Hawaii?

      You would think they could have left AF on in both cards and shown the same relative performance delta.

    • chuckula
    • 4 years ago

    Looking forward to the review but: $650? Really AMD? Is this a bad joke that you want to put in the next F1X3R video?

    I thought the big schtick about the Fury-X was that it wasn’t overpriced because WATER COOLER. So now the Nano isn’t overpriced because… NO WATER COOLER??

      • Shobai
      • 4 years ago

      I neg’d, I felt like you wanted to be offended.

      Like Scott says, maybe it’s supply and demand. Maybe it’s that in combination with a premium for a smaller footprint [c.f mini ITX, et al.]. Who knows? Let’s see if the price is in any way justified when we get independent performance numbers.

      At the end of the day, I wasn’t in the market for this card. If AMD can get people to pay what it’s asking, more power to them.

        • chuckula
        • 4 years ago

        I neged because I’m pretty sure you’ve never downvoted anyone who’s called a Titan or high-end Intel CPU overpriced.

        There are these things called “standards” and the AMD fanbase seems to like to make impossibly high ones for anything not made by AMD while exempting themselves from any standards whatsoever.

        Oh and one more thing: All of those reasons are [b<]excuses[/b<] for why the Nano is overpriced, but they don't change the fundamental fact that the Nano [b<]is[/b<] overpriced.. and I'm talking about compared to other AMD products, much less Nvidia.

          • NoOne ButMe
          • 4 years ago

          [TotalFanboy]But the Nano has so much more FP64 compute! It has like, 3x the Titan x! for only 2/3rds of the price!!! It’s so great, that’s like, 2x better FP64/dollar!!!!! I compare gaming GPUs on compute !!!! I are smart!!![/totalfanboy]
          (upvoted)

          This is overpriced product. Anything over $600 is blatantly. $50 more over regular Fury due to form factor and slightly less performance in most cases seems fair. I think it needs to be $550 to really be a “great” product.

          Comparing it to what NVidia would price such a card at, it’s not to crazy. However, as Chuckula says, compared to AMD cards… Well. It’s pretty bad.

          (it took me three tries to spell blatantly right. My edits are fail)

      • NoOne ButMe
      • 4 years ago

      [troll]Well, now you can get a fully waterblocked Fiji card!!! In a smaller form factor with less power allowed at the same cost as what you could before! Great Deal!![/troll]

      • DoomGuy64
      • 4 years ago

      Everything M-ITX is overpriced, and this is a fully enabled Fury. The problem isn’t the price, it’s that people have already accepted this pricing as normal, and you can thank them and Nvidia for that. The only way to lower prices now, is if people start refusing to pay $650+ for high end cards across the board.

        • K-L-Waster
        • 4 years ago

        Well, you could say the same thing about exotic cars and beach front property…

          • DoomGuy64
          • 4 years ago

          I could, and my answer to the pricing complaints would be the same.

      • Terra_Nocuus
      • 4 years ago

      lol, WCCFTech was throwing around “$400!!!” yesterday, those hacks

      • ImSpartacus
      • 4 years ago

      The Nano isn’t overpriced because it’s the best mitx-class gpu on the market by a long shot.

        • chuckula
        • 4 years ago

        “mitx-class” is a made-up word.

        But if you insist, I would posit that the GTX-980Ti… which can be used in a “mitx case” is the best “mitx-class gpu” on the market by a longshot.

        Here’s a video of one in operation: [url<]https://www.youtube.com/watch?v=JbeGBfJEwfw[/url<]

        • EndlessWaves
        • 4 years ago

        Except it’s dual slot and full height, while cases the size Mini-ITX was designed for typically only take single slot, half height cards.

        And as Chuckula mentioned, if you are going for an El Gigantico case then 10% more volume will get you a normal length card anyway.

Pin It on Pinterest

Share This