The Fury(X) Fiji Owners Thread

[Ferrari100]

Here's an interesting post on a German forum where an AMD whiner got so furious that he dismantled his card.

It does seem that the noise is caused by a combination of coil resonance, badly applied glue and component contact with the pump cover.

Google Translation

He also upload a video which demonstrates how the pump cover actually acts as an acoustic amplifier.

A simple solution might be to insert some soft dampening material between the pump cover and pump.

Eaxactly what AMD said it was then

 

[humbug]

https://www.youtube.com/watch?v=fFpy2L3B8lk

anyone seen this?

lol

 

[TheRealDeal]

I see the crap amd driver support is letting them down again. Oh wait thought Amd were the green bars

 

[JediFragger]

I'll trade those scores no problem for the battering the FX gets in a single card scenario

 

[Lokken86]

Why are the ti cards scaling so much better than the tx combo.
?

 

[IT Troll]

Toms hardware and other users tried that already makes almost no difference. The main problem as has been said is inside and is to do with the adhesive.

Toms only tried applying dampening to the main card cover not the pump cover.

Eaxactly what AMD said it was then

Not quite. They only mentioned the glue and it would seem there are other factors at play. The guy removed the patchy glue and re-glued the coils but this did not eliminate the noise (perhaps explaining AMD's "largely reduced" phrase).

He found that some components were too tall and so made contact with the underside of the pump cover. This is likely to be a manufacturing variance and will affect some cards and not others. Better clearance and/or some dampening to this cover should help.

 

[Mauller]

I see the crap amd driver support is letting them down again. Oh wait thought Amd were the green bars

Amd just upped their game with the 15.7 WHQL driver on the 15.200 branch. So that video is a little irellevant now. Some guy on guru 3d also went from 740k on 15.15 drivers to over 1 million draw calls on 15.7

These are my 3dmark api overhead results.

14.12 omega drivers had me at 440,161 ST
15.6 had me at 485,280 ST +10% over 14.12
15.200.1040 leaked had me at 590,211 + 21% over 15.6

15.7 pushed me all the way up to 712,263 + 20% over 1040 leaked.

from 15.6 to 15.7 i gained + 46% in draw call performance.

MADNESS!!

On an FX8350 by the way.

 

[TheRealDeal]

Amd just upped their game with the 15.7 WHQL driver on the 15.200 branch. So that video is a little irellevant now. Some guy on guru 3d also went from 740k on 15.15 drivers to over 1 million draw calls on 15.7

These are my 3dmark api overhead results.

14.12 omega drivers had me at 440,161 ST
15.6 had me at 485,280 ST +10% over 14.12
15.200.1040 leaked had me at 590,211 + 21% over 15.6

15.7 pushed me all the way up to 712,263 + 20% over 1040 leaked.

from 15.6 to 15.7 i gained + 46% in draw call performance.

MADNESS!!

On an FX8350 by the way.

Very nice gains. Lets hope all this transfers into game performance.

 

[Gerard]

Toms only tried applying dampening to the main card cover not the pump cover.

It presses onto the pump cover though minimising any vibration that could be transmitted externally. The main and most noticeable issue was the pump whine which seems to have been the adhesive.

 

[Kaapstad]

lol

Some people can not bench squat lol

For use @2160p with ultra settings 4 Titan Xs are the fastest and it is not even close.

 

[D.P.]

Why are the ti cards scaling so much better than the tx combo.
?

Because the benchmarks are seriously flawed.

 

[D.P.]

Some people can not bench squat lol

For use @2160p with ultra settings 4 Titan Xs are the fastest and it is not even close.

Interesting, got any numbers to post?

 

[nashathedog]

I see the crap amd driver support is letting them down again. Oh wait thought Amd were the green bars

At the end of the day how many people will actually be running 4 way anything?
Makes it a moot point and personally I'd rather see AMD do better work on 2 way implementation rather than a set up that will only benefit a few people. Once they actually give us reliable driver releases with profiles I'll consider getting a second card to try crossfire again but until then I'll stick to one card and along with the other 99.9% of users I'll have no real interest in what they can achieve with 4 way when they have got a profile working.

 

[Silent_Scone]

MainGear are an AMD partner this is their time to shine and shift units., I wouldn't put faith in a purchase just on those benchmarks alone. Not to mention no TressFX in Tomb Raider is an odd bench to run. The only people that should take note of those video slides are trolls lol

 

[Kaapstad]

Interesting, got any numbers to post?

Not with those settings lol.

With 4 cards you only use max settings or you risk a very big CPU bottleneck.

4 up the TXs will win against 4gb and 6gb cards @2160p more often than not because the other cards run out of VRAM.

 

[Final8y]

At the end of the day how many people will actually be running 4 way anything?
Makes it a moot point and personally I'd rather see AMD do better work on 2 way implementation rather than a set up that will only benefit a few people. Once they actually give us reliable driver releases with profiles I'll consider getting a second card to try crossfire again but until then I'll stick to one card and along with the other 99.9% of users I'll have no real interest in what they can achieve with 4 way when they have got a profile working.

There is no such thing as good 4 way unless the 2 and 3 way is good, so no the 4 way is not hindering 2 way development.

 

[Final8y]

Not with those settings lol.

With 4 cards you only use max settings or you risk a very big CPU bottleneck.

4 up the TXs will win against 4gb and 6gb cards @2160p more often than not because the other cards run out of VRAM.

Indeed, unless you dont like some of the max settings.

 

[LtMatt]

Not with those settings lol.

With 4 cards you only use max settings or you risk a very big CPU bottleneck.

4 up the TXs will win against 4gb and 6gb cards @2160p more often than not because the other cards run out of VRAM.

Kaap has a point, but remember that the cpu bottleneck will also apply to the Fury X too. It comes down to cpu grunt and 3/4 driver SLI/XDMA performance/scaling. Theres a lot of different factors at play here for sure.

 

[N19h7m4r3]

Folks got any updates on how the Fury X is performing with the new 15.7 drivers?

 

[Zeed]

Does not look like it.

I'm very interested how good are they