Fury Nano: First results in!

[boozzer]

but it is ok to push that job towards the hardware vendors you guys are something else.........

don't buy games from devs who gives crap support if that worries you.

 

[beginner99]

In Europe there are available from 3-4 different AIBs. I can currently buy Fury X from ASUS, MSI, VTX3D and XFX.

Not where I live. Only 2 and I actually have never seen one in stock. And it's $700. (GTX 980 still go for $550-$600 so thats just the prices here)

 

[crisium]

I never said anything about 980 performance. I only spoke about power consumption.

Oh, ok. That makes perfect sense, who cares about performance? Well, I have good news for you. AMD has a card called the Radeon 370 that uses 50w less than the Nano. Are you satisfied, since performance is irrelevant?

Let me know when they knock 50W off the TDP, that's a card that might spur my interest.

It appears that way. Or will you dismiss the 370 for x, y, and z reason that aren't in the above quote?

 

[omek]

This will be an ongoing problem with all explicit APIs like D3D12 and Vulkan. If a new GPU comes (especially with new microarch) it may not have better performance than 980 Ti and Fury X because the games won't have the optimal memory allocation scheme for it. The devs may need to patch the application.

If that is the case, then DX12/Vulkan will damage the industry and DX11.3 will come out as a winner.

I think that may be a bit of a blanketed assumption because I doubt Dice made use of any of the asynchronous abilities GCN offers in BF4 instead using it as a vehicle for demonstration and adoption maybe, Dice being very excited about anything which could extract more performance and potential. Another reason being that if the ACE's were used properly there should had been a small but substantial bump at the rendering end which there really wasn't.

I believe that Mantle's implementation within BF4 was mostly about the CPU efficiency and multi-threading capabilities which show in the consistency of the frame timings. With BF4 being the epitome of PC gaming it was probably a mutually joyful endeavor for both Dice and AMD, AMD for marketing the API and for Dice to geek out and extract performance from their engine - I could be wrong.

BF4 was a little early to the next gen party also (Frostbyte 3.0 not being a large departure from 2.0 and continuing to aim at the last gen consoles and their architecture) there's no way of telling how much Dice would have to manipulate or deconstruct the engine to use the asynchronous shading - it may not had been feasible to do. But I think that we can safely assume that most console crossovers will make use of the asynchronous shading if they are doing anything worth the while and one could see that that could easily translate to the PC. If the game engine is designed to natively take advantage of 12 at it's core and asynchronous shading breaking compatibility may not happen as easily.
TLDR; I don't think BF4 is a fully fledged example of DX12 or Mantle and the breaking of compatability may be because these games are trying to retain compatibility between two completely opposed API's - one being engine level and the other driver orientated
Food for thought

 

[AtenRa]

Not where I live. Only 2 and I actually have never seen one in stock. And it's $700. (GTX 980 still go for $550-$600 so thats just the prices here)

Yea they are not available in every country but in Germany they are in stock.

example
https://www.caseking.de/en/search?sSearch=fury+x

 

[XFXSupport]

FYI- I haven't heard anything about the Nano in weeks. still no updates on a launch date or product details.

 

[Phynaz]

Oh, ok. That makes perfect sense, who cares about performance? Well, I have good news for you. AMD has a card called the Radeon 370 that uses 50w less than the Nano. Are you satisfied, since performance is irrelevant?


It appears that way. Or will you dismiss the 370 for x, y, and z reason that aren't in the above quote?

I didn't say anything about performance. I didn't say I expected 980 performance, I didn't say performance was irrelevant.

What I did say was I'll be interested in a card like the Fury Nano that has 50 watts lower power consumption. You know, like maybe a binned version that has some units disabled? They do that with video cards nowadays you know.

Holy crap people need reading comprehension around here.

 

[crisium]

Holy crap people need reading comprehension around here.

You're as vague as ancient Greek directions on how to find Atlantis.

Let me know when they knock 50W off the TDP, that's a card that might spur my interest.

That is easy to interpret as meaning: you are disappoined with what they are offering for 175W TDP; this card should be 125W TDP. Heck, I wasn't even the first person to think this is what you meant.

What I did say was I'll be interested in a card like the Fury Nano that has 50 watts lower power consumption. You know, like maybe a binned version that has some units disabled?

Ooooh. You want a slower performing Fiji card with even less TDP?

I agree, it would be nice to have a fuller lineup of Fiji from top to bottom.

 

[AtenRa]

You're as vague as ancient Greek directions on how to find Atlantis.

They didnt have GPS those days :biggrin:

 

[TheProgrammer]

Or devs will patch the games

There's really very few game engines that need to be maintained. On games with home brewed engines, as long as you have DX12 support on your card it'll continue to work. Valve will keep a game like CSGO updated. Sure you won't get performance updates for most games, but you don't get those today either.
No one updates games like Sacrifice from 2001 for performance gains either, nor does anyone release a driver shim for games like that. Same concept will continue.

Turning back from DX12/Vulkan won't happen now, so embrace the change from Nvidia to AMD superiority. And all the money and resources NV built up for the DX11 era. Basically, all the hijacking and recoding of shaders, power that NV had mastered is gone. The nads are quickly being cut off NV's most potent asset.

Sad for some here probably. Not for the gaming public though, it's an improvement for the industry to remove Nvidia from the equation.

Speaking as a AAA game dev, in fact, the more forward looking hardware designs like GCN, will gain more payoff going forward. Everyone knows GCN is better than NV for VR at this point. So getting games optimized for GCN will pay off down the road for GCN owners, and those who buy descendent cards.
Typically AMD had the better hardware so expect a shift in how things work around here.

 

[wand3r3r]

Curious to see what this brings. The market has been pretty stagnant lately and it's hard to be excited about the current options at the current prices imo.

 

[tential]

Curious to see what this brings. The market has been pretty stagnant lately and it's hard to be excited about the current options at the current prices imo.

Holiday season I'm sure prices will drop so vendors can sell all they can before new gpus hit so they can double dip on people who will upgrade to 2016 gpus once they see the perf

 

[RussianSensation]

Well, 850mhz Fury X would in theory be ~20% slower than Fury X. That's a big gap.

Particularly since Fury X isn't much faster than 980 at 1080/1440p, it really only shines at 4K.

The statements you made above are contradictory. Fury X/ 980Ti are nearly neck and neck at 1440P and are both > 23% faster than a 980. On one hand you say that 20% slower is a big gap but then you claim that Fury X isn't much faster.

Also, the Fury is extremely close to the Fury X so I don't see how the Nano has any chance to beat the Sapphire Fury Tri-X unless it has the capability of boosting > 900mhz and exceed 175W of real world power usage.

The Nano would need 900mhz clocks to even match the Sapphire Tri-X Fury, but then it's operating in a 175W power envelope? That would imply Fury/Fury X are very leaky/factory overvolted chips if dropping the clocks to just 850-900mhz on the Nano drops power usage nearly 100W!

That also implies AMD had a tremendous opportunity to release an awesome perf/watt R9 290/290X/390/390X cards but they failed to do so. I do look forward to the Nano but I think the performance metrics of it being nearly as good as the Fury at 175w power usage are insanely optimistic.

It has higher core clock. Much higher with four numbers not just three in MHz. I can't say more.

WOW, so you are implying the Nano is the best binned Fury X chips that have a 175W TDP and > 1000mhz clocks, and are fully unlocked? I am not saying I don't believe you but it sounds too good to be true on paper.

I never said anything about 980 performance. I only spoke about power consumption.

You don't need to say anything about 980 level of performance because it's already implied when AMD said Nano > 290X and 290X is extremely close to the 980, unless we give 980 a 15-17% boost via after-market versions. What that means is everyone in this thread except for you has understood that 50W lower TDP wrt to the Nano automatically implies you wanted a card nearly as fast as a 980/390X with a 125W TDP. Everyone knows that's completely unrealistic on 28nm node.

IMO, the term TDP should be banned from correlating it to power usage, especially on Team NV. A typical after-market 980 uses 190-200W of power in games, a far cry from its marketing 165W TDP.

For Nano, what we want to know is its real world power usage, not TDP. Marketing TDP is meaningless for people who know what TDP stands for and NV/AMD have purposely misrepresented TDP for years, especially NV (Most 970/980/GTX480 cards, etc. perform nowhere near their TDP ratings).

 

[RampantAndroid]

FYI- I haven't heard anything about the Nano in weeks. still no updates on a launch date or product details.

Seeing as how no one else has quoted this...I'm guessing this is actually an official XFX account? Shouldn't XFX be in the know about Nano?

Why would XFX (or XFXSupport) not know anything about the Nano?

 

[tential]

If you use tdp in your decision to purchase a gpu you are stupid. It's not a mean thing it's just a fact.
If you want to know power consumption test it, don't guess.

 

[happy medium]

Seeing as how no one else has quoted this...I'm guessing this is actually an official XFX account? Shouldn't XFX be in the know about Nano?

Why would XFX (or XFXSupport) not know anything about the Nano?

They can't handle the truth!

 

[RussianSensation]

If you use tdp in your decision to purchase a gpu you are stupid. It's not a mean thing it's just a fact.
If you want to know power consumption test it, don't guess.

Ya, but a lot of PC gamers are ignorant as to what TDP actually stands/means. TDP has never stood for real world power usage of NV/AMD GPUs. At best it's just a guidance for a cooling system/board's power. If you drop 2x8-pin power connectors on the card, might as well assign 375W TDP to it, and technically such a card can pull > 350w of power.

GTX680 = 192W TDP, real world power peak in games 186W
HD7970 = 250W TDP, real world power peak in games 189W
GTX580 = 244W TDP, real world power peak in games 229W
GTX480 = 250W TDP, real world power peak in games 272W

GTX970 supposedly has 145W TDP. Ya, good luck with that.

Conversely, R9 270X has a TDP of 180W but in the real world peaks at 130W in games.

NV claims 90W TDP for the GTX950.

MSI Gaming 950 = 103W peak
EVGA 950 = 110W
Asus 950 = 103W
Gigabyte 950 = 105W
Zotac 950 = 133W (yes 133W)

TDP today is mostly just a marketing gimmick for GPUs. The only thing that matters is the real world power usage of the specific GPU you buy, as measured scientifically.

Usually online if you see someone state something stupid how a 950 uses 90W of power vs. 180W for an R9 270X and a GTX970 uses 145W of power vs. 290-300W for an R9 290, it tells you a lot about the person making these claims: (1) They are either ignorant about how real world power usage is actually measured (2) They are a ***boy trying to troll/make a point to exaggerate real power usage differences to crap on the competitor's product.

The most misleading stuff you'll online related to:

#1) People who use a reference card's power usage and use an after-market card's performance!

Like claiming 980Ti uses 240W of power and but adding 15-17% more performance vs. Fury X.

#2) People who use FurMark to claim real world power usage of a graphics card (in real world applications like games). When in reality, FurMark is just a power virus that loads up all the components on the board.

Gigabyte GTX980 = 342W peak
Gigabyte GTX980Ti = 359W peak

Since the Nano has an 8-pin connector and it can draw at least 75W from the PCIe slot, it can theoretically draw at least 225W of power in games.

What's crazy here is let's say the Nano draws 175-180W of power in games, well I already linked that a 925mhz HD7970 reference card drew 189W. So why in the world did AMD use these garbage reference blowers on HD5870/6970 and 7970 if the cooler fitted on the Nano can deal with 175-180W of power?

It's really said AMD has been that incompetent about reference cooler selection since September 2009.

 

[RussianSensation]

They can't handle the truth!

What truth?

"AMD Radeon R9 Nano Might Launch on 27th August – 175W Fiji and HBM"

Powered Graphics Card"The report comes from Zolkorn (an industry veteran based in the Thai market) who received details from industry based sources that AMD will be unveiling their 16cm graphics card on 27th August that is just a week away.

It's pretty obvious that thus far it's all rumours and they related to AMD unveiling the Nano on August 27th. We don't know if that date has even been set in stone by AMD, and even if it were, we don't know if August 27th is a hard launch or just an official unveiling. AMD could easily unveil the Nano then but launch it in 2-3 weeks from now. It's not the first time they have done that (i.e., HD7970 was unveiled Dec 22, 2011 but hard launched only on January 9, 2012).

"For those of you trying to decide whether to get a 7970, you will have some time to decide. This is a soft launch; AMD will not make the 7970 available until January 9th (the day before the Consumer Electronics Show), nearly 3 weeks from now."~ AT

Anyway, the card is coming soon, supposedly only 6 inches (~15.25 cms)

Fury X2

 

[Glo.]

http://videocardz.com/57409/amd-radeon-r9-nano-confirmed-to-feature-4096-stream-cores
"Up to 1000 MHz core clock".

 

[tential]

Again RS, it comes to AMD not knowing how to position themselves. They slap on TDPs without thinking it seems. Clearly the market likes a low TDP, and CLEARLY the market never checks beyond that, so just slap a low TDP on and call it a day! No, AMD insists on shooting itself in the foot. Why AMD would massively overstate their TDP on some chips, I do not know. No one does.

 

[RussianSensation]

Again RS, it comes to AMD not knowing how to position themselves. They slap on TDPs without thinking it seems. Clearly the market likes a low TDP, and CLEARLY the market never checks beyond that, so just slap a low TDP on and call it a day! No, AMD insists on shooting itself in the foot. Why AMD would massively overstate their TDP on some chips, I do not know. No one does.

Yup. NV has been slapping marketing TDP for years. It's about time AMD plays the same game. I would just put a 150-175W TDP on a 225-250W card. The same people who think TDP = power usage wouldn't know any better since it's not like they read reviews/pay attention to proper power consumption testing.

With nearly the same specs as the Fury X and 100W less TDP, AMD is getting smarter when it comes to the marketing game.

 

[happy medium]

What truth?

The truth that your full of numbers and information but you hardly ever have nothing good to say about .............. like a walking AMD billboared and that mabe the Nano is delayed.

 

[Spanners]

I didn't say anything about performance. I didn't say I expected 980 performance, I didn't say performance was irrelevant.

What I did say was I'll be interested in a card like the Fury Nano that has 50 watts lower power consumption. You know, like maybe a binned version that has some units disabled? They do that with video cards nowadays you know.

Holy crap people need reading comprehension around here.

When you say "like the Fury Nano" then what could you have possibly meant if not performance? If performance is not relevant then it's trivial to get a card that size with a sub 125W TDP.

The card is already binned and if you disable units then it's a different card altogether. Furthermore "Let me know when they knock 50W off the TDP, that's a card that might spur my interest." doesn't mention or even vaguely imply either of those points, so I think the fault is with your ability to communicate not anyone's ability to comprehend.

Lets be honest you were trying to make a dismissive rejoinder and now people are calling you out for how unrealistic it was.

 

[happy medium]

Fury Nano is the fastest SFF card

link to review?

 

[Abwx]

Yup. NV has been slapping marketing TDP for years. It's about time AMD plays the same game. I would just put a 150-175W TDP on a 225-250W card. The same people who think TDP = power usage wouldn't know any better since it's not like they read reviews/pay attention to proper power consumption testing.

AMD wont follow this exemple as they have a different offering in that they sell reference cards and it wouldnt be cautious to underestimate TDPs. Nvidia know this well that s why there s no reference cards at all, so they cant be held responsible as they use the OEMs as proxy.