I was interested for awhile but now I'm out in considering this gpu for purchase. This raises questions for me though like if this is really as fast as a 980 or even in the realm of the fury x then how does a max 250 watt part really keep up with one that is 400-500 at max load? If so then I question the engineering and marketing of amd of what is the purpose of releasing a more power hungry part when if the lower wattage one is the close in comparison? So yeah about $150 more than i'm interested in paying. 980ti anyone?
Erm, what? The FuryX stays under 300W for the most part, where do you get that 500W figure from? The fury nano will probably be binned chips that work with very low voltages, so you'll stay at that 200-ish W and just live with lower clock speeds due to heat and power constraints. It's a very niche card, it is specifically meant for SFF cases, and that's exactly what it'll do.
The nano is still basically a Fury X, that's probably why it's priced that way in addition to it's most likely limited supply
The Fury is still overall the card to go with unless you really want the best which would be the 980ti, the nano is just made for people who want to build a powerful rig in something like an elite 110 case, also it may be useful for helping out AMD's mobile GPU department which is rather lacking
Still woulda been nice to see AMD drop the price of the Fury X to something like 600 and price the nano there as well, but that'll probably happen shortly
What's worrying me is the one 8-pin connector, that makes me believe that the whole power delivery might be a bit weaker as well. But since it is meant to hit 1000MHz (as long as it stays cool enough) I assume that it'll still work. Fury Nano + a waterblock would be the way to go if it was cheaper than the FuryX honestly.
If they launch it at $649, i strongly believe AMD would be shooting themselves in the foot. For a few reasons,
We've had a few OEMs produce ITX versions of GPUs. ASUS has the Direct CU mini, Gigabyte has the Windforce ITX, EVGA has the superclocked ITX 960s. They don't charge extra because it's iTX. AMD produced the Fury X.. By design it's a short card, the way HBM is it can shorten the length of the card, cause you're not putting the chips on the PCB anymore, now you have the nano, which is an ITX version of Fury? And it will raise a few questions on AMD's end...
Why the Water cooler if it's efficient? If the Nano performs like the Fury X, then why buy the Fury X? Why didn't the R9 Fury, have an ITX variant? And what's the point of the R9 Fury? If the Nano performs like the Fury, then why wasn't the Nano originally the Fury X?
Personally too many questions will be asked, and I think AMD will recieve backlash if they sell it for more than $500.
Exactly the 8 pin connector won't even allow you to reach the average 300 watts of a fury x. You're paying a premum per inch of 108 dollars and some change. Not to mention the performance won't be the same. No thanks.
It'll thermal throttle, that has basically been confirmed. It has a frequency of up to 1000MHz, in reality it'll probably be lower. Also, not all FuryX's could become this GPU, it'll probably be binned chips that work very well with lower voltages, similar to the -e FX CPUs that AMD released some time ago. The Fury is cut down, that product needs to exist, whereas the Fury Nano will be a power efficient and tiny version of the FuryX.
Well, both the PCIe interface and an 8-pin power connector can deliver more power, quite easily actually. AMD has ignored the specs before, with the 295x2. It's more down to the power delivery than the connector alone. And yes, you're paying extra for that formfactor, that's the entire point of using such a small cooler.
Of course the supply is limited. AMD and Hynix decided that the most important piece of the card has to go through customs several times before it ready to be put on a pcb. Epic fail there.
Wait, what, It's limited by HBM a bit yes but it's more so that not every fiji chip could pull off running how the nano is going to run at such low wattage for it's performance. They probably had to build up a decent supply which is probably why it took so long for it to come out.
It begs the quesiton why undercut your own cards then. the same way they did the 285 replacing the 280. just replace the 390x. and put the nano there. there was really no need for the R7-265 to be in that line-up.
The Product Line should of been like this
R7-360X - Let this be an R9-270x -- Let this compete with the 950. R9-370X - Let this be the R9-285 - Let this compete with the 960 R9-380X Let this be R9-290x - Let this compete with the 970 R9-390X let this be the Nano - Let this compete with the 980 R9-Fury X - Let this compete with the 980ti. (even though it doesn't beat it)
AMD should shorten the line-up. just get rid of the Non-X cards all together, I mean it was good for us, but it's not good for AMD, they are undercutting their own cards.
I mean seriously, If the R9-Nano performs about the same as the Fury X / Fury , I would never grab the Fury X or the Fury
I think the Nano (and all the Fury cards for that matter) weren't supposed to be part of the normal line-up for a good reason. Limited availability would make you look terrible and would annoy people (especially after what happened with the 200 series). Not to mention that they can distinguish between traditional and HBM cards for this generation. It's probably not a great idea but I can see why they'd do it.
I don't want to and I certainly won't. But in the end this GPU is not meant for me, so I accept that and see it for what it is, a tiny form factor GPU that'll fit in many mini-ITX systems and still (hopefully) perform similarly to a 980 or 390x while being even more efficient.
$649.99 is probably what the economists at AMD came up with the supply they have. That price for the Fury X was controversial, but obviously they are practically sold out. So if they sold it for less, they'd directly lose money.
There are no third-party review out yet. I wouldn't speculate with such certainty. It could be more than just core clocks, like other very technical performance effecting values that are very low level. Besides, the core clock isn't really know either.