Why does the Fury beat the Nano?

[custombuild]

Has someone worked this out? According to the stats the difference between them is that the Nano does 256 GTexels vs the Fury's 224 and are otherwise the same in GPixels and memory bandwidth, yet in the benches I've seen the Fury beats the Nano.. wa?

 

[William Gaatjes]

Did you check the clock speeds ?
One has watercooling, the other a fan. That means the fury runs a lot hotter for a reason.

 

[custombuild]

GTexels, GPixels and memory bandwidth stats are the end results after applying clock speeds, and I'm comparing the nano with the Fury, not the Fury X(the one with water cooling,) so it's fanned card vs fanned card.

 

[Crono]

There's some throttling going on with the Nano to keep it from exceeding a set temp (75C, from what I've seen). Clock speed of mine out of box and under load was 750MHz, not the 1000MHz it's rated at (though I increased the power limit to keep it at 1000MHz now), which is more like the max speed. Reviews that I've read seem to confirm this.

Makes sense given that the Nano is more likely to go into HTPCs and other small cases where limiting heat and power usage can be as important as raw performance.

EDIT: the AnandTech review notes the power throttling

 

[custombuild]

I was going to ask about if there were reviews going over the phenomenon but your ninja edit did the trick

 

[RussianSensation]

I was going to ask about if there were reviews going over the phenomenon but your ninja edit did the trick

Nano is power limited. That means on average its clock speed never reaches 1ghz, and roughly hovers at 850-900mhz clocks. To maximize the potential of the Nano, you have to move the PowerTune to +50% and probably crank the fan higher to 50%+. Once you do that, you can run the Nano at 1000-1050mhz and it will easily beat the Fury.

http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/70246-amd-r9-nano-review-17.html

Of course the Fury can also be overclocked so it's somewhat of a circular argument. Another side effect of maxing out the Nano is power usage will go up from 175W.

 

[Rvenger]

+50 power target nearly doubles the Nano power consumption. 445w total system power readout on par with Fury X. 5820k @ 4ghz. It has to be pulling power from the PCIE slot once power target increased.

 

[Madpacket]

This is why the Fury X is a better deal IMO. Fury Nano is great if you're really space constrained but that's about it. Fury-X can do 1125-1175mhz with a bit of voltage and never power throttles as the water cooler is overkill.

Plus with the Nano you risk getting a card with major coil whine.

 

[Shmee]

I would only consider a Fury with Nitro tri X cooler, if it is decent deal that is. Sometimes they unlock too.

 

[custombuild]

I would only consider a Fury with Nitro tri X cooler, if it is decent deal that is. Sometimes they unlock too.

I wonder how much cooling AMD's next gen chip is going to need with its half power requirement, that tidbit hasn't been revealed yet.

 

[Madpacket]

I wonder how much cooling AMD's next gen chip is going to need with its half power requirement, that tidbit hasn't been revealed yet.

Look at what's cooling a Radeon 7850. About that much.

 

[james1701]

I wonder how much cooling AMD's next gen chip is going to need with its half power requirement, that tidbit hasn't been revealed yet.

Rumor has it there may be some passive cooling options, but nothing has been confirmed yet.

 

[RussianSensation]

I would only consider a Fury with Nitro tri X cooler, if it is decent deal that is. Sometimes they unlock too.

With new cards around the corner, especially AMD's competition likely to beat Fury X by 25-30% at 1440p, I wouldn't buy any of the current gen cards now. I also wouldn't buy 390 level class anymore with Polaris 10 around the corner. This is one of those times where it's better to sit it out until 1st week of June and go from there. I have a feeling AMD will have to pull a 290/290X move and lower price on the entire Fiji line since the successor to that line isn't launching until 2017 with Vega. The only thing AMD can do is cut prices.

 

[Thala]

This is why the Fury X is a better deal IMO. Fury Nano is great if you're really space constrained but that's about it. Fury-X can do 1125-1175mhz with a bit of voltage and never power throttles as the water cooler is overkill.

Depends. The Nano is quite a bit cheaper than the FuryX, about on Fury level and it consistently beats an even overclocked Fury when you increase the power target to +50%. When power target is increased it starts throttling at about 80 C, which it rarely reaches when you tune the fan setting slightly.
So yes, i think a Nano currently is a good deal.

 

[Shmee]

With new cards around the corner, especially AMD's competition likely to beat Fury X by 25-30% at 1440p, I wouldn't buy any of the current gen cards now. I also wouldn't buy 390 level class anymore with Polaris 10 around the corner. This is one of those times where it's better to sit it out until 1st week of June and go from there. I have a feeling AMD will have to pull a 290/290X move and lower price on the entire Fiji line since the successor to that line isn't launching until 2017 with Vega. The only thing AMD can do is cut prices.

I agree with what you are saying, I am just saying I would not consider the Fury X or the Nano, assuming relative pricing is the same. We should wait for price drops/new cards, I am just making the point that assuming pricing drops proportionally, the Fury with nitro cooler is the only one I would consider over some Hawaii setup. The cooler/board setup of the nano and the Fury X just don't appeal to me.