Fury Nano Discussion Thread

[RussianSensation]

The one thought that keeps running through my head is- if AMD has such highly binned GPUs then why isn't the Fury Nano the Fury Mobile instead (aka a laptop GPU)? Seems like where they are weakest is in laptops, they don't have any way of milking that "daddy is buying me a gaming laptop for college" market.

1. Look at the small supply of Fiji XT cards. It's going to be hard to get any major OEM selling high-end gaming notebooks accept a mobile Fiji XT GPU with $400-$650 upgrade cost in a pre-built gaming laptop without having a steady supply.

2. 175W TDP is still way too high for a gaming laptop. They'd need to get it down to at least 125W.

What they really needed is a 350-425mm2 HBM2 GPU to go into laptops but I guess they have no money to design that alongside with Fiji. After all, if they did, they wouldn't have released R9 390/390X either as this 350-425mm2 mid-tier HBM1 chip would have replaced R9 390 cards.

With their limited budget, it looks like they are putting all their eggs into the 16nm basket. It is a risky strategy because they will be giving up a lot of market share over the next 3-4 quarters until 16nm HBM2 GPUs show up. Another risk is that their 16nm GPUs may not be as good as Pascal so this strategy would backfire and they will have lost 4 quarters of potential sales.

If this thing was green and called a Titan Nano, ppl would call it the next best thing to sliced bread.

It's coming - and it'll be a Pascal product.

But anything AMD does first, is always downplayed. GDDR5 first -> downplayed, HBM1 first -> downplayed, whenever AMD cards have more VRAM (HD6950/6970/7970/290X) -> downplayed.

Remember how GTX970 SLI/980 SLI had sufficient VRAM for 9 months but once 980TI came out, 4GB HBM1 is a bottleneck.

The 290X was never 300$ here, specially not in september 2014.

Most of us don't live in Denmark. I could quote prices in Russia, UK or Canada every day but it's of no use to AT's large userbase. You should know better given who most of the posters are on this forum. Not to mention, even when posters in the US would be building rigs, not once did I see you recommend R9 290/290X/R9 295X2, not once.

I have no problem recommending the superior NV card when it is.

The point is you have ignored R9 290/290X/R9 295X2's awesome price/performance in the US/Canada for their entire generation, even when R9 290 was on sale for $400 vs. GTX780 for $650 and after 780's price drop to $500. So really, you can wiggle your point all you want but even if R9 290 was $99, you wouldn't have recommended it.

Just like now you still do not ever recommend $220 R9 290 cards vs. GTX960/970. You simply do not recommend any AMD products and I bet almost anything you've never owned any AMD/ATI videocard.

I guarantee it if the Nano cost $329, you'd say "Oh it's too late since miniITX 970 has been available for a year." If the Nano was $229, you'd say "Oh it doesn't have HDMI 2.0" so it's worthless as an HTPC. If the Nano was $129, you'd say "wow AMD is desperate, too dangerous to buy their products as I don't believe they'll survive to 2017 and there will be no driver support." I guarantee it no matter what AMD makes, you'll always find something wrong with it.

I bet if tomorrow a Fury Nano fell out of the truck on the street and you had a choice of taking it for free but you could never resell it, gift it or otherwise, you wouldn't put it in your system. I bet you are one of those GeForce 5 and GeForce 7 owners, am I right?

For ITX you want a blower card or something similar that puts the heat outside the case.

Nano's cooler should allow 40-50% of the heat being exhausted out of the case. Are you implying 105W of the other heat that is dumped into the case would be too much for a miniITX rig to handle? Do you have proof of that?

As I already said, all it takes is 1-2 miniITX rigs with a Nano to prove you wrong that a miniITX rig needs a blower.

 

[golem]

But anything AMD does first, is always downplayed. GDDR5 first -> downplayed, HBM1 first -> downplayed, whenever AMD cards have more VRAM (HD6950/6970/7970/290X) -> downplayed.

Remember how GTX970 SLI/980 SLI had sufficient VRAM for 9 months but once 980TI came out, 4GB HBM1 is a bottleneck.

Isn't this taken out of context though? GTX980/970 came out what, a year ago? Fast forward to now, shouldn't AMD's highest end card have more than 4gb?

 

[cbn]

R9 Nano is a great concept, but there also needs to be a R7 Nano as well.

40W TDP Performance per watt optimized Dektop low profile card for the very large installation base of Pre-built SFF desktops (as well as future bus powered applications involving Thunderbolt III laptops and UCFF desktops).\

 

[RussianSensation]

Isn't this taken out of context though? GTX980/970 came out what, a year ago? Fast forward to now, shouldn't AMD's highest end card have more than 4gb?

It would be better if it did, but the point is none of the same people who keep discussing Fury having a 4GB VRAM limit said a word about:

1) 680 2GB, 770 2GB vs. HD7970/7970Ghz/R9 280X 3GB and their respective SLI vs. CF configurations;

2) 690 vs. HD7990;

3) Recommended GTX970SLI/980 SLI despite both of those setups faster than the Fury XT chips. So if 4GB of VRAM is a problem for a Fury, then are 970 SLI/980SLI obsolete then?

What makes their point even more laughable is that scientific testing shows Fury X beating 980TI SLI. The only reason 980Ti SLI wins is via overclocking, not because of VRAM.

The point is anything AMD does first, whether they succeed or fail, is clumped as irrelevant by people who only buy NV. For example, bitcoin mining made $ but because NV cards couldn't do it, this was downplayed as irrelevant, risky, etc. What that means the type of consumers who only buy NV wouldn't buy a Nano if it cost $299; it doesn't matter. Their only interested in seeing what AMD does is 2-fold: (1) The want AMD to make some products so that it forces NV to lower prices/release faster cards; (2) But they'll always downplay any advantages AMD has to justify their own loyalty/purchase. This article explains this phenomenon.

If you read some of my posts or other gamers, we ripped the Titan X apart and claimed that (1) NV's 980Ti after-market cards will be faster than the Titan X out of the box; (2) AMD's 2nd best card in the foot-steps of HD7950/R9 290 -> Fury Non-X could offer 87-89% of the flagship NV card's performance.

But by human nature people don't want to admit they overspend or wasted $. They just don't because it would be akin to admitting that they have failed to assess data objectively and let emotions dictate their decision.

The Nano doesn't make sense for most us at $649 and we all know it but AMD knows it doesn't have a lot of Nanos and the market for miniITX products is small so they decided to pull a Titan strategy which is to price the product as high as possible for people who aren't crunching price/performance, much like NV went out to lunch with Titan Z's $3000 price.

 

[Glo.]

Its pretty interesting that this type of GPU made on 14/16 nm process can have around 100W TDP.

Nice times are coming.

About the thread. The price is IMO a bit too high. I agree, perf/watt, small form factor and overall - full Fiji chip are worth the premium. But consumers would like it a bit cheaper .

I think this GPU was made in OEMs in mind first.

Also its quite funny, but how small TDP could have cut down Fury if Full Fiji chip can have 175W TDP?

 

[SimianR]

The price is disappointing but I'm still very intrigued and I'm excited to see some reviews I honestly think that when production of the Fiji based cards finally ramps up (hopefully by the fall) that AMD takes advantage and comes back down to earth with some more sane pricing and tries to take some market share back.

 

[tential]

R9 Nano is a great concept, but there also needs to be a R7 Nano as well.

40W TDP Performance per watt optimized Dektop low profile card for the very large installation base of Pre-built SFF desktops (as well as future bus powered applications involving Thunderbolt III laptops and UCFF desktops).\

This would have gotten people excited and all, but there is NO WHERE NEAR the volume to make this happen lol.

Might happen in January though if AMD is lucky and gets design wins.

 

[tential]

The price is disappointing but I'm still very intrigued and I'm excited to see some reviews I honestly think that when production of the Fiji based cards finally ramps up (hopefully by the fall) that AMD takes advantage and comes back down to earth with some more sane pricing and tries to take some market share back.

Same here.
And yes, I'm excited about the review irregardless of the price. Any new GPU review that comes out above the $200 mark excites me lol.
Low end gpus are boring but I'm sure some people care about those parts obviously =D.

 

[cbn]

This would have gotten people excited and all, but there is NO WHERE NEAR the volume to make this happen lol.

I bring up a hypothetical R7 Nano because the current 40W and below offering (R7 240) is pretty underwhelming.

And there are so many SFF Pre-built machines (Dell, HP,, Lenovo, etc)......so many of them (not even counting future bus powered Thunderbolt III applications).....that could use something much better than a R7 240-at 40W and below.

Instead of 320sp @ 720 Mhz/780 Mhz (R7 240 spec @ 30W), I was thinking having at least 896sp @ 500 Mhz would be much better for all these very capable machines.

Might happen in January though if AMD is lucky and gets design wins.

What is happening in January?

 

[tential]

January is just my own guesstimate of when AMD would have enough volume to actually even begin to consider a low end SKU.
It's honestly not coming though lol.

 

[frozentundra123456]

It should be AMD's most appealing product, but the price is way to high.

It is OK to pay a premium for small size and energy efficiency, but they went too far with this card. I also want to see some real world performance, power consumption, and temperatures to see if it can really deliver the performance and performance per watt that AMD is promising in a wide variety of games and settings.

But even if it delivers all AMD is promising, it should be 550 to 600 dollars. I think basically it will be in short supply, so AMD is figuring they still can sell all they can produce even at an inflated price.

 

[DustinBrowder]

I mean its the full chip, its just clocked lower, so this should be able to bring about 90% of the performance of the Fury X.

I would have still liked to see it released at $600 since its the full chip, but it is premium quality build, its is very small, very low consumption, it beats ALL Nvidia cards in performance per watt, so Nvidia shills can't just claim that Nvidia is better in that department.

AMD's Fury X, Fury and Nano are better in performance per watt than Nvidia, are premium quality build, have HBM memory so future proof, 4GB HBM is like 8GB GDDR5. It just savaged Gddr5, it has almost double the bandwidth than most gddr5 memory.

 

[crisium]

Fury X and Fury do not beat Maxwell 2 in PPW. Fury Nano might... we shall see.

$650 is too much, but AMD probably cannot afford to sell it for less. We don't know the margins on this thing at $650. If it were $450 would it be a huge consumer winner? Sure. But that could mean they have to sell 3x as many to make as much money, or who knows however many more. And when they will probably sell most anyway, it would be foolish to do this. They have little choice but to sell it at a high price. If inventory doesn't move much, then they will be forced to drop it of course. AMD simply cannot afford to do the 4870 revolution here.

 

[thesmokingman]

The margins on the Fury line are most likely dismal given the development cost of HBM1. It is what it is... shrugs. I would assume they would have priced it lower if they could.

 

[Pariah]

​

If this thing was green and called a Titan Nano, ppl would call it the next best thing to sliced bread.

Why? When has Nvidia ever targeted a niche market besides rich people? Nvidia has not demonstrated any desire over the years to produce products with such a limited target audience as to be pretty much irrelevant from a revenue generating perspective the day it is released.

AMD needs to make money. They aren't going to do it with this card. They need to be targeting a larger market.

 

[Pariah]

It's coming - and it'll be a Pascal product.

But anything AMD does first, is always downplayed. GDDR5 first -> downplayed, HBM1 first -> downplayed, whenever AMD cards have more VRAM (HD6950/6970/7970/290X) -> downplayed.

Remember how GTX970 SLI/980 SLI had sufficient VRAM for 9 months but once 980TI came out, 4GB HBM1 is a bottleneck.

Is this your attempt to be funny or just intentionally deceive people? The picture of the Titan mini you posted is from an Nvidia April fool's joke. It was not intended as foreshadowing of any upcoming product and has no basis in reality. Why would you even post that?

 

[NTMBK]

​

Why? When has Nvidia ever targeted a niche market besides rich people? Nvidia has not demonstrated any desire over the years to produce products with such a limited target audience as to be pretty much irrelevant from a revenue generating perspective the day it is released.

Titan targeted a niche market. Shield portable, Shield tablet and Shield TV all targeted niche markets. GSync targets a niche market.

 

[SolMiester]

It would be better if it did, but the point is none of the same people who keep discussing Fury having a 4GB VRAM limit said a word about:

1) 680 2GB, 770 2GB vs. HD7970/7970Ghz/R9 280X 3GB and their respective SLI vs. CF configurations;

2) 690 vs. HD7990;

3) Recommended GTX970SLI/980 SLI despite both of those setups faster than the Fury XT chips. So if 4GB of VRAM is a problem for a Fury, then are 970 SLI/980SLI obsolete then?

What makes their point even more laughable is that scientific testing shows Fury X beating 980TI SLI. The only reason 980Ti SLI wins is via overclocking, not because of VRAM.

The point is anything AMD does first, whether they succeed or fail, is clumped as irrelevant by people who only buy NV. For example, bitcoin mining made $ but because NV cards couldn't do it, this was downplayed as irrelevant, risky, etc. What that means the type of consumers who only buy NV wouldn't buy a Nano if it cost $299; it doesn't matter. Their only interested in seeing what AMD does is 2-fold: (1) The want AMD to make some products so that it forces NV to lower prices/release faster cards; (2) But they'll always downplay any advantages AMD has to justify their own loyalty/purchase. This article explains this phenomenon.

<snip WOT>


.

You just answered your own point. VRAM - in no case with 680\770 did the 3GB of the 7xxx series play any advantage against NV, as NV have better texture compression, and you pointed it out with the OC 980TI being quicker with OC as against extra VRAM.
VRAM however does play a bigger part with AMD GPUs, and is the reason AMD Fury company slide all show graphic with no AA, as it kill the fury at 4K

As for your bitcoin comment, how you think that played out for AMD gamers wanting to purchase cards?...They couldnt, because bitcoin merchants bought them all...LOL.

You just cant face it, whatever AMD do, they drop the ball. They have to be the most inept technology company still in business, and probably half the reason most NV purchasers stay away from them.

 

[AtenRa]

You just answered your own point. VRAM - in no case with 680\770 did the 3GB of the 7xxx series play any advantage against NV, as NV have better texture compression, and you pointed it out with the OC 980TI being quicker with OC as against extra VRAM.
VRAM however does play a bigger part with AMD GPUs, and is the reason AMD Fury company slide all show graphic with no AA, as it kill the fury at 4K

Fury Nano and Fury X has the same specs expect of the TDP. Vram, TMUs, ROPs etc are the same.
So Fury Nano doesnt have a problem with AA because Fury X has no problem with AA.

4K Fury X with AA

http://www.guru3d.com/articles_pages/amd_radeon_r9_fury_x_review,26.html

http://www.guru3d.com/articles_pages/amd_radeon_r9_fury_x_review,27.html

 

[Kuiva maa]

Most of us don't live in Denmark. I could quote prices in Russia, UK or Canada every day but it's of no use to AT's large userbase. You should know better given who most of the posters are on this forum. Not to mention, even when posters in the US would be building rigs, not once did I see you recommend R9 290/290X/R9 295X2, not once.

I live in Finland (another nordic country, near Denmark) and I bought an Asus DCUII R9 290X, brand new for 309 euros 4 months ago. It was the standard price for this model for quite a while, until 390X arrived actually. GTX980 only recently had model dropping to 450 which is still overpriced vs a 300 euro 290X.

 

[ShintaiDK]

*walltext snip*

I gave you the US price chart. You on purpose ignored that plus the GTX970 and performance chart as well.

 

[Glo.]

For me current AMD lineup shows, that AMD finally has competent management team, and started to pull right triggers from business point of view. Milking the customers to death.

Nvidia does it for a long time now. Time for AMD to get some revenue. From this point of view its good that they priced Nano that high.

 

[positivedoppler]

Is this your attempt to be funny or just intentionally deceive people? The picture of the Titan mini you posted is from an Nvidia April fool's joke. It was not intended as foreshadowing of any upcoming product and has no basis in reality. Why would you even post that?

It is???? Color me disappointed, I actually got excited when I saw that

 

[tential]

It should be AMD's most appealing product, but the price is way to high.

It is OK to pay a premium for small size and energy efficiency, but they went too far with this card. I also want to see some real world performance, power consumption, and temperatures to see if it can really deliver the performance and performance per watt that AMD is promising in a wide variety of games and settings.

But even if it delivers all AMD is promising, it should be 550 to 600 dollars. I think basically it will be in short supply, so AMD is figuring they still can sell all they can produce even at an inflated price.

It's not that R9 Nano isn't a good product.

It's that for some reason, AMD disassociated it from Fury X.

Just like the fake Titan X Mini ad,
Call this R9 Fury X Nano, and make only references to how this is Fury X at 175W TDP for a mini Fury X. And people would think differently.

Instead, it was R9 Nano, 30% faster than a GTX 970 Mini! 200% the price!!!!!!!!

Rather than R9 Fury X Nano, 95% of the Fury X performance. Same Price!

Most people believed this would be a cheaper product due to the positioning and marketing of R9 Nano. It was a far more premium product than originally anticipated due to stupid comparisons to vastly cheaper products.

 

[tkrushing]

I mean its the full chip, its just clocked lower, so this should be able to bring about 90% of the performance of the Fury X.

I would have still liked to see it released at $600 since its the full chip, but it is premium quality build, its is very small, very low consumption, it beats ALL Nvidia cards in performance per watt, so Nvidia shills can't just claim that Nvidia is better in that department.

AMD's Fury X, Fury and Nano are better in performance per watt than Nvidia, are premium quality build, have HBM memory so future proof, 4GB HBM is like 8GB GDDR5. It just savaged Gddr5, it has almost double the bandwidth than most gddr5 memory.

I don't understand the performance/watt thing in a desktop PC. I mean if I'm going to drop the cash on a sports car for racing, why would gas mileage be at the top of my priority list and not performance? The only time I see wattage being a big deal is when it gets out of hand like it does when over-volting a Fury for little gain.