• Save 15% on ALL SFF Network merch, until Dec 31st! Use code SFF2024 at checkout. Click here!

GPU Now this could be big: R9 Nano!

iFreilicht

FlexATX Authority
Feb 28, 2015
3,243
2,361
freilite.com
Well yeah, that's exactly what it is: A very short Fury X with lower clock speed and therefore lower power consumption for the same price. As you say, there's literally no reason to go for the Nano unless you need a short card, and that's exactly the market its placed in.

Now that I think about it, it may actually be pretty smart of AMD to use the regular Fiji chip instead of having a different one just for the Nano, because it lets them adapt the supply of Nanos and Fury X cards to the demand. If it turns out that the Nano is more commonly used in a few years, they don't have to change anything about the chip manufacturing volumes, they just have to put the chips on different PCBs.
 

Vittra

Airflow Optimizer
May 11, 2015
359
90
If I were doing a custom loop, I'd get the Nano, not the Fury X. Same Fiji XT chip - possibly better binned, even. I'd want the air cooler to fall back on in times of troubleshooting or if I was moving away from WC at a later date, and it's resale value will be better due to most peoples preference for aircooled solutions.
 

iFreilicht

FlexATX Authority
Feb 28, 2015
3,243
2,361
freilite.com
Also the board is shorter, so if you get a matching waterblock, you've got more space for the res, pumps, fans and rads. And you will be able to push the TDP up in software, though maybe not as high as on the Fury X, and it may be that you run into problems with a single 8pin connector.
 

rawr

SFF Lingo Aficionado
Mar 1, 2015
137
10
I speculate that it will be higher bin. If it really is undervolted that much, I'd imagine that hard to do on a regular Fury.
 

BirdofPrey

Standards Guru
Sep 3, 2015
797
493
Maybe I wasn't paying attention, but I had thought the Nano would be a cut down Fury rather than a cut down Fury X, so wasn't expecting the price. It does seem a bit high. While I do get that SFF is a specialized market and thus, could demand a premium, with the Fury X pretty much being SFF as well, it's kind of a hard sell in the size vs cost department.

What may yet be relevant though, might actually be the power usage. The Nano uses less wattage, and you can find a larger variety of small PSUs the lower you go in wattage not to mention less heat to dissipate.
Makes me think, though, it might be neat to have a GPU with a dial your power feature, choose how many 6 or 8 Pin power connectors to use, and the GPU sets itself appropriately.
 

BirdofPrey

Standards Guru
Sep 3, 2015
797
493
There is a Fury, it's the one that's air cooled and OEMs can modify the PCB. It's binned lower than the Fury X and Nano
 

iFreilicht

FlexATX Authority
Feb 28, 2015
3,243
2,361
freilite.com
Oh, whoops. Well what's the difference between the Nano being a cut down Fury or Fury X? Is the Fury that much cheaper?

(Except for the binning of course)
 

BirdofPrey

Standards Guru
Sep 3, 2015
797
493
$550 for the Fury if I am not mistaken vs the $650 for the Fury X.
Given that I was anticipating the Nano would be $600 (having assumed it would have fewer compute units than the Fury X but be marked up as a niche product). Not a huge difference from what they announced, but I take what I can get.

I'm scared to think what the Fury X2 will retail for.
 

Vittra

Airflow Optimizer
May 11, 2015
359
90
The Fury is cut down. The Nano isn't - it's the same fully enabled Fiji XT core as the Fury X.
 

EdZ

Virtual Realist
May 11, 2015
1,578
2,107
However, it may well be more aggressively power gated. As can be seen from Intel's -S and -T series CPUs having improved binning (to deal with lower voltages), and comparable boost clocks on-paper at much lower TDPs to the regualar line, but whose performance scales down with power due to more agressive throttling to remain within the specified TDP.
Best to wait to see how the Nano performs in benchmarks compared to the Fury X.
 

Vittra

Airflow Optimizer
May 11, 2015
359
90
That is an important consideration. It is likely their best binned chips, but as it's a 175w part vs the Fury X at 275w, so it's definitely going to see some serious throttling - the "up to" 1000mhz clock figure is worded that way for a reason.

If I remember correctly though, the CCC allows you to push power targets up about 20%? Beyond that, those with no fear in using aftermarket software and even BIOS mods, it's probably perfectly capable of matching the Fury X - albeit at increased noise levels - unless you go WC.
 

BirdofPrey

Standards Guru
Sep 3, 2015
797
493
The Fury is cut down. The Nano isn't - it's the same fully enabled Fiji XT core as the Fury X.
I know.
I was just mentioning that surprised me since I expected the Nano WOULD be a cut down chip. I never imagined they'd use their best wafers for a niche product.

That is an important consideration. It is likely their best binned chips, but as it's a 175w part vs the Fury X at 275w, so it's definitely going to see some serious throttling - the "up to" 1000mhz clock figure is worded that way for a reason.

If I remember correctly though, the CCC allows you to push power targets up about 20%? Beyond that, those with no fear in using aftermarket software and even BIOS mods, it's probably perfectly capable of matching the Fury X - albeit at increased noise levels - unless you go WC.
I'll have to find it again, but I seem to remember one of the articles that recently came out about the Nano mentioned it isn't likely to actually hit max clock since it's going to be power limited, and AMD only put that high of a clock so heat and power are the limiting factors rather than the clock.

We are definitely going to have to see some real benchmarks to see what the reduce power availability does to performance.
 

Phuncz

Lord of the Boards
SFFn Staff
May 9, 2015
5,948
4,953
AMD Radeon R9 Nano Can Be CrossFired With Radeon R9 Fury X – Fiji Powered Cards With Air and Liquid Cooling Combo

Read more: http://wccftech.com/amd-radeon-r9-n...cards-air-liquid-cooling-combo/#ixzz3l4qjjz00

AMD’s Radeon R9 Nano is going to hit the market this week on 10th September 2015. We know that the Nano is powered by AMD’s flagship Fiji graphics core but what is more interesting is that AMD will be allowing Cross Fire compatibility of the Radeon R9 Nano with their Radeon R9 Fury X graphics card that has been available in the market since June 2015.

I hadn't thought about this but this kind of setup basically allows you to dual-wield a couple of "Fury X's" and a single 120mm radiator CLC for the CPU in the Kimera, while leaving the bottom cleared.
 

Vittra

Airflow Optimizer
May 11, 2015
359
90
Why is wccftech reporting these cards as full DX12 support? AMD themselves have refuted such claims. Nothing on the market currently can make such a claim.

Anyways, such functionality would be expected, otherwise it would deviate from past practices of cards of the same core being able to be crossfired together.

I don't see why you would do this though. The cards when paired are limited by the weaker one, which can potentially be drastic for a 175w part paired with a 275w. I believe the AMD CCC allows for about a ~20% increase in power target for the most part, so aftermarket tools or BIOS modding will be needed to get them even closer.

If you need clearance on the bottom but need dual-gpu, the best solution may end up being the Fury X2. I'm curious what pricing on it will be. Will they simply price it at $1300 USD MSRP, the equivalent of 2x Fury X cards, or will it be slightly cheaper than the single cards as has been AMDs previous positioning?
 

iFreilicht

FlexATX Authority
Feb 28, 2015
3,243
2,361
freilite.com
If you have two cards directly next to each other, the top one will get worse airflow to its cooler than the bottom one. So you can get a watercooled top card and an aircooled bottom card while still saving a good amount of space. It's definitely a niche use, though.

I agree that a Fury X2 could be the better choice.

Not really surprised by this either, but nice to see.