AMD Radeon R9 Nano coming next week

FoxEye · 3 Sep 2015 at 17:42

humbug said:
Not "The same performance as Fury-X" i never said that but with a full fat Fiji chip it shouldn't be far off.

Then why the massive difference in TDP. Why release a FuryX with much higher power consumption, if you have the tech to reduce it by such an amount without impacting performance?

What is the purpose of the FuryX release? Too **** off people who bought it before seeing what the Nano brought?

Gregster · 3 Sep 2015 at 17:45

Hopefully they will have more stock than they did with the FX and F. Then again, I am not sure the demand will be there for this? Not seeing much hype being shown.

loftie · 3 Sep 2015 at 17:47

The Nano should be the best Fiji chips, able to run stably at lower voltages/power. Fury X will use chips that can't accomplish that so they'll draw more power and pump out more heat so they use the AIO. Fury will be the the chips that fit into neither of the other two, but can be run with parts disabled.

humbug · 3 Sep 2015 at 17:48

FoxEye said:
Then why the massive difference in TDP. Why release a FuryX with much higher power consumption, if you have the tech to reduce it by such an amount without impacting performance?

I don't know, you quoted a post with my speculations on that.

The only other thing that sprigs to mind is power scaling, take your CPU which is running at a clock rate in what one could call a "Goldilocks Zone" a happy medium where the power required for its performance is just so. now overclock it by 20%... suddenly its out of its comfort zone and to hold that +20% its using +80% more power.

GPU's are the same, Fury-X could already be outside that happy medium, Nano may simply by running right in where its most happy.

Just another theory ^^^^

Orangey · 3 Sep 2015 at 18:42

It seems nobody in this thread understands power gating binning or profiling.

pmc25 · 3 Sep 2015 at 18:57

I really wish mods would start taking a hardline. The level of trolling and astroturfing is so severe in most AMD threads now. There are multiple people constantly claiming that this will have the same performance as a 970 ... which is obviously total crap. In non-GW titles it should very significantly exceed the 970 at 1920x1080, and absolutely slaughter it at 2560x1080, 3440x1440 and 4K.

The 970 is badly beaten by the 390(non X) above 1920x1080, let alone a TDP or clock limited Fiji die.

eddyr · 3 Sep 2015 at 19:01

FoxEye said:
Then why the massive difference in TDP. Why release a FuryX with much higher power consumption, if you have the tech to reduce it by such an amount without impacting performance?

What is the purpose of the FuryX release? Too **** off people who bought it before seeing what the Nano brought?

To hit another performance level using all available power and TDP.
The Fury cards are pushing clocks and volts as far as reasonably able.

When they are not brute forcing it their adaptive voltage tech will be able to to do its job to a tighter margin getting you the drop in power at ~900mhz or wherever it hangs out to achieve it.

pmc25 · 3 Sep 2015 at 19:17

eddyr said:
To hit another performance level using all available power and TDP.
The Fury cards are pushing clocks and volts as far as reasonably able.

When they are not brute forcing it their adaptive voltage tech will be able to to do its job to a tighter margin getting you the drop in power at ~900mhz or wherever it hangs out to achieve it.

Plus the PCB and its components are specifically designed to economise on power.

I don't think there'll necessarily be any differentiation in bin between Fury X and Nano ... for the ridiculous number of GCN units in Fiji, at full power the Fury X is reasonably economical too. I don't think there is any 'magic' going on to make the Nano, besides HBM and the interposer allowing for the physical size.

Stanners · 3 Sep 2015 at 19:44

pmc25 said:
I really wish mods would start taking a hardline. The level of trolling and astroturfing is so severe in most AMD threads now. There are multiple people constantly claiming that this will have the same performance as a 970 ... which is obviously total crap. In non-GW titles it should very significantly exceed the 970 at 1920x1080, and absolutely slaughter it at 2560x1080, 3440x1440 and 4K.

The 970 is badly beaten by the 390(non X) above 1920x1080, let alone a TDP or clock limited Fiji die.

AH right. So everyone has to have the same opinion as you or they are trolling. That makes sense, you should pm a mod and point that out.

LoadsaMoney · 3 Sep 2015 at 19:59

pmc25 said:
I really wish mods would start taking a hardline. The level of trolling and astroturfing is so severe in most AMD threads now. There are multiple people constantly claiming that this will have the same performance as a 970 ... which is obviously total crap. In non-GW titles it should very significantly exceed the 970 at 1920x1080, and absolutely slaughter it at 2560x1080, 3440x1440 and 4K.

The 970 is badly beaten by the 390(non X) above 1920x1080, let alone a TDP or clock limited Fiji die.

Its not total crap, the 970 isn't much slower than the 980, the FX is around the 980s performance, the Nano will not be much slower than the FX, therefore, it will be around the 970.

980 - FX
970 - Nano

pmc25 · 3 Sep 2015 at 20:33

LoadsaMoney said:
Its not total crap, the 970 isn't much slower than the 980, the FX is around the 980s performance, the Nano will not be much slower than the FX, therefore, it will be around the 970.

980 - FX
970 - Nano

It's blatantly obvious you don't actually believe that.

LoadsaMoney · 3 Sep 2015 at 20:53

pmc25 said:
It's blatantly obvious you don't actually believe that.

nashathedog · 3 Sep 2015 at 23:38

pmc25 said:
It's blatantly obvious you don't actually believe that.

Obviously he's pushing your buttons, The point is people are allowed to voice there opinion regardless of what it is as long as it's not done rudely or repeated over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over. :confused:

LoadsaMoney · 3 Sep 2015 at 23:41

Gregster said:
Hopefully they will have more stock than they did with the FX and F. Then again, I am not sure the demand will be there for this? Not seeing much hype being shown.

Supposed to be plenty of these

CAT-THE-FIFTH · 4 Sep 2015 at 01:30

Orangey said:
It seems nobody in this thread understands power gating binning or profiling.

Obviously not!

maonayze · 4 Sep 2015 at 10:53

Why Oh Why, do Nvidia users feel they have to constantly troll in the AMD threads. It doesnt seem to happen the other way around.

Are Nvidia users THAT insecure?

At least keep your comments 'on topic' and reasonably sane, please.

Any comments of "The thread would only be 2 pages long" will be countered by "So would yours if there weren't so many of you belly aching about your windows 10 drivers"

mmj_uk · 4 Sep 2015 at 10:58

Orangey said:
It seems nobody in this thread understands power gating binning or profiling.

Fury X already has power gating so that only leaves binning which will mean £££££.

I think the majority of the power reduction will be coming from the fact that it's not clocked to its very limit like Fury X and a likely aggressive TDP cap, either way I think people are going to be disappointed by the price.

fs123 · 4 Sep 2015 at 11:20

When is the NDA expiring for the Nano? The reviews will settle the arguments one way or the other.

Avenged7Fold · 4 Sep 2015 at 11:37

Rather than low voltage chips being binned, i believe these cards are actually the dregs of the Furyx chips that couldn't reach stability at their specified voltage and the nano is a way to sell off these chips at a more stable lower clock. I also think the TDP figure is probably at a conservative clock and that the maximum clock is dependant on temperatures. TDP has always been a conservative estimate with the definition being vague, i would treat it no different to the static pressure stats on fans (some are way off, some are accurate some are plain wrong and so comparisons between them are irrelevant).

If the yields on teh Fx chips are as low as the stock levels, then it would make sense that AMD put the 'not quite there' chips to good use somewhere.

For all we know the TDP figure is at 675MHz under 'reasonable load' but the card can go up to 1000MHz if the application is requires it and temperature allows.

bru · 4 Sep 2015 at 11:50

fs123 said:
When is the NDA expiring for the Nano? The reviews will settle the arguments one way or the other.

I believe it is the 10th.