AMD R9 Fury reviews!

[littleg]

Seems a good card to me. Decent performance and pitched at a nice price point for AMD. For my money this should be the value buy of this generation.

 

[tviceman]

I could easily say my 2500k and r9 290 draws 330 watts at the wall especially with no proof, not saying what game or benchmark, and not proving its stable.

You very well could be right, but your post is high on claims and lacking in proof.

Lol y r u sounding surprised and well, insulted?

It is insulting. He might as well have just called me a liar and/or just said "prove it" instead of beating around the bush and laying claim that I don't know what I am saying.

 

[Subyman]

I don't get it, why they decided to price it $50 higher, sure it's a little faster, but they have to be realistic, the 980 is a Geforce and it's better in some regards (power usage is a big win for example)... $499 or even less would make this a big win, for the price they choose it's just... OK... better than fury X vs 980 Ti, but still, not quite there...


in any case, this card makes me think they should allow aircooled Fury X ASAP, specially considering the pump noise drama.

I agree. Its an odd choice. Those that wanted 980 performance at $549 have had 9 months to purchase that. AMD needed to come in with a bit more performance at a bit lower price to move a decent number of people into a purchase. Slotting into the current price/performance ladder doesn't do much to upset the market, it just means they were late to the game.

I'm thinking the prices will shuffle soon as stock increases.

 

[tential]

It is insulting. He might as well have just called me a liar and/or just said "prove it" instead of beating around the bush and laying claim that I don't know what I am saying.

Then I'll say it.
You made the claim, prove it.

 

[rgallant]

AMD biggest problem is the 980 been out for 9+ months , so there real late to the table.

Many have bought cards in this performance area, even on AMD side (290x).
So were talking a lot more money for little performance , it going to be very hard for them to gain market share now IMO.

yea but the slower 980 cards in the reviewers hands were most likely a $550.00 card if they had it before june 2 2015.
http://www.kitguru.net/components/g...eforce-gtx-980-gtx-980-ti-but-not-for-europe/
not many peeps with a 980 in their sigs saying they paided $550 + for their slower card re their fury postings, yea price dropped 5 weeks ago and amd will do a price drop in the future. imo

 

[IEC]

A 50 watt explosion! O.M.G. My 1475/7900mhz GTX980 + mildly OC'd 4770k draws 330 watts max at the wall.

Strix Fury is $100 more than Zotac 980. 980 OC performance >= Fury OC performance.

That's unlikely because of how power usage scales with overclocking. Hint: it's not linear.

OC Wattage = TDP * (OC MHz / Stock MHz) * (OC Vcore / Stock Vcore)^2

Using the following as baseline figures:
1126 MHz stock base clock
165W "Graphics Card Power"
Assuming a conservative +0.1V vcore for the overclock
(TechReport has voltage baselines for reference 980 vcore vs a less overclocked Zotac Amp 980 vcore @ 1.118 vs 1.212V, or ~+0.1V)

OC Wattage = 165W * (1475MHz/1126MHz) * (1.212/1.118)^2

~254W just for the video card. Even with conservative numbers, you're burning as much power as a Fury (and more than the ASUS model).

 

[AnandThenMan]

Fury is going to end up faster than big Maxwell too. Mark my words.

I have a strong feeling as well this will happen.

 

[tviceman]

Then I'll say it.
You made the claim, prove it.

http://forums.anandtech.com/showthread.php?p=36799848&highlight=#post36799848

Where were you and Vulgar when I posted this? Some people in this forum are nothing but bad jokes.

 

[tential]

http://forums.anandtech.com/showthread.php?p=36799848&highlight=#post36799848

Where were you and Vulgar when I posted this? Some people in this forum are nothing but bad jokes.

Pics?
I mean you said you had all the gear and could prove it. Since you can. Let's just put this to bed

 

[Enigmoid]

That's unlikely because of how power usage scales with overclocking. Hint: it's not linear.

OC Wattage = TDP * (OC MHz / Stock MHz) * (OC Vcore / Stock Vcore)^2

Using the following as baseline figures:
1126 MHz stock base clock
165W "Graphics Card Power"
Assuming a conservative +0.1V vcore for the overclock
(TechReport has voltage baselines for reference 980 vcore vs a less overclocked Zotac Amp 980 vcore @ 1.118 vs 1.212V, or ~+0.1V)

OC Wattage = 165W * (1475MHz/1126MHz) * (1.212/1.118)^2

~254W just for the video card. Even with conservative numbers, you're burning as much power as a Fury (and more than the ASUS model).

Maxwell does very well without touching voltage.

And + 0.1 V is not be any means conservative on maxwell.

 

[monstercameron]

I have a strong feeling as well this will happen.

This is Assuming Artic islands is yet another evolution on gcn rather than something else. This situation only exists only because they tweak drivers for gcn and every enervation gets a boost and I don't think terascale has been getting better.

 

[AnandThenMan]

I don't think drivers are the only reason, GCN was/is a very forward looking architecture.

 

[Erenhardt]

Maxwell does very well without touching voltage.

And + 0.1 V is not be any means conservative on maxwell.

Maxwell touches voltages without your knowledge by itself:

980 sucks [full stop] more than 290X:

I don't think drivers are the only reason, GCN was/is a very forward looking architecture.

Consoles are.

 

[tviceman]

Pics?
I mean you said you had all the gear and could prove it. Since you can. Let's just put this to bed

I will do it tomorrow when I am home. I have a great idea though - let's put some money on it (or a steam game). If you and Vulgar are going to be so indecent to call me out only because it craps all over your theories, then it needs to be worth my time to do what I already know to be true.

You put up a steam game of my choice if my rig tops out under 330 watts with vsync disabled Metro Last Light benchmark looping, and if it goes above 340 watts I'll buy one of you two clowns a steam game.



That will be enough with the name calling.

-Rvenger

 

[el etro]

Problem with Maxwell cards decontrol of power consumption with overclocking is the very high clock the cards is sitting at. Even the transistor magics(IC improvements that allowed Maxwell electrically perform better) can't surpass the power consumption increase by operating at such high clocks. Watercooler may help a little with the situation.

 

[TheProgrammer]

The Fury X reviews were muted in enthusiasm for all the wrong reasons.

At the $650 price point it's the only card I would buy on the market today. It's got the best architecture for VR.. if I'm spending that much on a card, I want the best for VR in 6-12months. For me, 980Ti is dead in the water. Essentially same performance, but NV won't have a VR-quality design like GCN until at least Pascal.. maybe even later.

 

[nitromullet]

The Fury X reviews were muted in enthusiasm for all the wrong reasons.

At the $650 price point it's the only card I would buy on the market today. It's got the best architecture for VR.. if I'm spending that much on a card, I want the best for VR in 6-12months. For me, 980Ti is dead in the water. Essentially same performance, but NV won't have a VR-quality design like GCN until at least Pascal.. maybe even later.

I'm sort of on the opposite side of this. I play games at 1440p, so the GTX 980 Ti is a no-brainer for me. I buy cards for the performance they can give me today, not (possibly) 6-12 months from now.

 

[Sabrewings]

The Fury X reviews were muted in enthusiasm for all the wrong reasons.

At the $650 price point it's the only card I would buy on the market today. It's got the best architecture for VR.. if I'm spending that much on a card, I want the best for VR in 6-12months. For me, 980Ti is dead in the water. Essentially same performance, but NV won't have a VR-quality design like GCN until at least Pascal.. maybe even later.

If that's the case, why have a lot of Crescent Bay and SteamVR demos been run on a Titan X? You make it sound like a Maxwell card is incapable of running VR, yet Oculus lists the 970 alongside the 290 in its minimum specs list.

Not sure if you've been drinking the Koolaid or you're pouring, but Maxwells will run VR (and have been since late last year) just fine. I've used a DK2 on a 980 and it worked great.

 

[flopper]

If that's the case, why have a lot of Crescent Bay and SteamVR demos been run on a Titan X? You make it sound like a Maxwell card is incapable of running VR, yet Oculus lists the 970 alongside the 290 in its minimum specs list.

Not sure if you've been drinking the Koolaid or you're pouring, but Maxwells will run VR (and have been since late last year) just fine. I've used a DK2 on a 980 and it worked great.

Nvidia themselves says not so much.
fury is the better tech for > VR

 

[Enigmoid]

Maxwell touches voltages without your knowledge by itself:

You are reading far too much into that chart.

Its a plot of the frequency and voltage of the games in the review. Obviously this varies between cards.

Lets look at aftermarket cards.

ASUS STRIX

MSI Gaming

Obviously its very possible to substantially increase the clockspeed without touching the voltage.

 

[YBS1]

I will do it tomorrow when I am home. I have a great idea though - let's put some money on it (or a steam game). If you and Vulgar are going to be so indecent to call me out only because it craps all over your theories, then it needs to be worth my time to do what I already know to be true.

You put up a steam game of my choice if my rig tops out under 330 watts with vsync disabled Metro Last Light benchmark looping, and if it goes above 340 watts I'll buy one of you two clowns a steam game.

Oh, snap! It just got real....

 

[tviceman]

Oh, snap! It just got real....

It's probably too real for them. Neither will likely pony up.

 

[Silverforce11]

You are reading far too much into that chart.

Its a plot of the frequency and voltage of the games in the review. Obviously this varies between cards.

Lets look at aftermarket cards.

ASUS STRIX

MSI Gaming

Obviously its very possible to substantially increase the clockspeed without touching the voltage.

Dude did you look at the graphs you linked? It shows increased vcore (X-Axis) with higher clocks (Y-Axis).

That's how NV's BOOST works. If you up the power limit and set a target clock, it will try to reach that by automatically adjust voltages and clocks.

It's OC made easy and a very good technology.

With AMD OC, you need to manually up vcore and test for stability, its a trial & error to arrive at a clock you want, so it requires more effort. Hopefully next-gen, AMD will implement easy-OC-Boost like NV.

 

[Rvenger]

It's probably too real for them. Neither will likely pony up.

No and I am not amused that this challenge one bit. You can stop the nonsense.


-Rvenger

 

[Enigmoid]

Dude did you look at the graphs you linked? It shows increased vcore (X-Axis) with higher clocks (Y-Axis).

That's how NV's BOOST works. If you up the power limit and set a target clock, it will try to reach that by automatically adjust voltages and clocks.

It's OC made easy and a very good technology.

With AMD OC, you need to manually up vcore and test for stability, its a trial & error to arrive at a clock you want, so it requires more effort. Hopefully next-gen, AMD will implement easy-OC-Boost like NV.

Of course. But look at what happens at a voltage just over 1.2V. Clockspeed increases greatly at that given voltage.

Reference - 1.22V - 1255 Mhz
STRIX - 1.215V - 1305 Mhz
Gaming - 1.205V - 1355 Mhz

Gaming at the top bin runs less voltage and gets higher clocks. This basically goes with what many have said: Manual voltage increases are not necessary on nvidia cards when overclocking (unless the cards are under water or liquid nitrogen).

My point is that overclocking Nvidia cards will force the cards into the top voltage bin. However, it will not raise voltage above the voltage that normally applies at that bin. My guess is that if you overclocked the STRIX or GAMING another 10% (should be possible for 1450 mhz) you would still see that cards operating at the same ~ 1.2 - 1.23V.

This is true on AMD cards as well. As long as the voltage is not changed, power consumption scales well with frequency. Anyway this is the fury thread.