AMD Fury X Reviews

[Silverforce11]

It's rather mind-boggling actually. The 390X and 980 are within a few percent of each other and the 4096-shader Fury-X is only beating the 980 by 5%. I expected much better results.

1080p the gap for everything is closer. Games are simply running into CPU bottlenecks.

At 1440p its 51% faster than R290X (which the 390X improved upon a lot)

As it shifts to GPU bottleneck, the gap widens, as expected:

The problem as seen on this chart, are custom 980Ti with high out of box boost clocks. They demolish Titan X even.

Fury X would need to pull off a 25% OC to match that.. chances of it doing that? No way. Not even with vcore mods. I'm guessing 1.25ghz tops with vcore. Still not enough.

 

[.vodka]

Hm, good observation. W10+DX12 to drop the CPU bottleneck going forward, then?

 

[Head1985]

1080p the gap for everything is closer. Games are simply running into CPU bottlenecks.

At 1440p its 51% faster than R290X (which the 390X improved upon a lot)

As it shifts to GPU bottleneck, the gap widens, as expected:

The problem as seen on this chart, are custom 980Ti with high out of box boost clocks. They demolish Titan X even.

Again its 830-870Mhz 290x
http://www.computerbase.de/2015-05/grafikkarten-testverfahren-testsystem/2/
if you want coparing compere it to 290x oc witch is 1030Mhz

 

[Stuka87]

There has got to be some driver issues with Fury. It should be destroying the 290X/390X, but it isnt in some cases (1080P). beat it yes, but just barely. Although at 4K it pushes ahead by a good margin.

 

[AtenRa]

Pathetic,

Same mm2 vs GM200, HBM advantage, WC advantage and still slower than GTX980ti and at the same price ???

They better release the Nano at $400 or they dont get a cent from me.

 

[Skurge]

Not impressed.

 

[Silverforce11]

Again its 830-870Mhz 290x
http://www.computerbase.de/2015-05/grafikkarten-testverfahren-testsystem/2/
if you want coparing compere it to 290x oc witch is 1030Mhz

That's the reference R290X. The OC R290X is listed on that chart also. Sizeable lead either way. Less against the 390X though.

 

[SK10H]

Fury need to have 3840 not 3584 and price at $500 to grab any market share. If another month past by and oc situation remain at $549 with 3584, it's double meh.

 

[AnandThenMan]

Not sure if this is a repeat of past AMD launches, underwhelming performance at first but the card came into its own and over the lifespan stretched its legs and became a great long term value.

OR Fury just plain sucks. Not sure which. One thing for sure is no way it should be hardly better than the 290X at lower resolutions something is off there.

 

[Headfoot]

Fury X has some bad frame time spikes in GTA V though:

Tom's review found nothing like that at all. I'd chalk it up to that particular run.

 

[monstercameron]

No opengl, mantle, dx12, 4k video with hevc, compute, encoding via vce tests? Showing me a bunch of old dx11 games for a forward looking card isn't enough.

Did anyone even test power consumption with frtc enabled?

 

[Head1985]

There has got to be some driver issues with Fury. It should be destroying the 290X/390X, but it isnt in some cases (1080P). beat it yes, but just barely. Although at 4K it pushes ahead by a good margin.

Its ROP limited + slow tessellation.
IF fury x have 128Rops it will be way faster than 390x

 

[Headfoot]

Maxwell has had 9 months of driver optimization under its belt. Fiji has 0. I imagine Fiji will come up a few percent on average over the coming months.

If Fury is ~10% slower than Fury X ala 290x vs 290 we could have another situation where the top end chip isn't worth getting but the 2nd tier chip is a great deal.

 

[MrTeal]

Well, I think my crossfired 290s are safe for this generation. Skylake + Fury might just wait until SL-E and 14-nm GPUs.

 

[Innokentij]

http://www.hardocp.com/article/2015/06/24/amd_radeon_r9_fury_x_video_card_review/11

Well the fury x ran out of vram at mere 1440p in dying light, heres hoping it will go the way all AMD cards latley has gone and get it's price cut by alot to get a cheap downclocked "980TI". I would pick it up then if nvidia dont release pascal in Q1 2016 .

 

[Azix]

I think they really needed the nano and pro out with fury x with the 390x cheaper. Lower price ($50 or so) on each. This isn't just about having new cards, its about taking market share. Getting that money for 2016

AMD would do well being transparent on this and doing interviews etc to shed some light on the situation. But my guess is they will think being hush hush is good. ETA on drivers, OC, boost details etc. The cooling is solid, so wheres the performance? When's the performance? What will dx12 and windows 10 bring?

eg someone mentioned the Teraflop advantage the AMD cards have, will dx12 expose this? Will that 2.6TFLOP advantage the fury x has over the titan and 980ti be exposed with something like asynchronous shaders? etc. etc. etc. etc. etc. They need to get out there.

 

[Silverforce11]

Vram has nothing to do with why Dying Light runs crap on all AMD GPUs. Get with the GameWorks program.

980 beats Fury X in that game:

http://www.hardocp.com/article/2015...x_980_platinum_vs_amd_r9_295x2/2#.VYq8fPmqpBc

http://www.hardocp.com/article/2015..._gtx_titan_x_video_card_review/6#.VYq8W_mqpBc

@Azix
All that Win10 and DX12 stuff is fluff cos so few games this year will be DX12 ready. It's a pointless footnote.

 

[.vodka]

I suppose they're focusing all their driver efforts for W10 going forward on, 8.1 and its intricacies about to be obsoleted by microsoft in a little time. Fiji on DX12 should be revisited then. Overvolting support should be there by the time W10 launches.

I mean AMD cards are already showing improvements under W10 preview. Time will tell. I still feel AMD needed that performance they're probably and surely getting in the future today :\

If nV hadn't launched the 980TI it would have been one hell of a launch, like the 290x did to the Titan.


All in all, it's a much better product than the 290x/390x and at least closes the gap somewhat with GM200. Going forward if past history is any indication, it should do much better. Let's see what aircooled Fiji and what Nano can do next.

 

[AtenRa]

One thing, I will wait to see what the new 15.20 driver on Win 10 will bring. Hope it will get a better scaling otherwise they really need to cut down the price.

 

[Azix]

Why on earth are these sites even using PCars in their benchmarks.

 

[monstercameron]

I think they really needed the nano and pro out with fury x with the 390x cheaper. Lower price ($50 or so) on each. This isn't just about having new cards, its about taking market share. Getting that money for 2016

AMD would do well being transparent on this and doing interviews etc to shed some light on the situation. But my guess is they will think being hush hush is good. ETA on drivers, OC, boost details etc. The cooling is solid, so wheres the performance? When's the performance? What will dx12 and windows 10 bring?

eg someone mentioned the Teraflop advantage the AMD cards have, will dx12 expose this? Will that 2.6TFLOP advantage the fury x has over the titan and 980ti be exposed with something like asynchronous shaders? etc. etc. etc. etc. etc. They need to get out there.

Can't play all your cards at once...fury still has a few unanswered questions.

 

[DiogoDX]

It does look that way. It seems while they got HBM into the card they cut too many corners with the work done to the gpu one of which was not scaling up the number of rops as they increased core count and memory bandwidth.

Look at this: http://www.computerbase.de/2015-06/.../6/#diagramm-metro-last-light-redux-2560-1440

Games where nvidia has advantange FuryX had good scaling like ACU, Bioshock an Metro.

Games that a 290X do good against a 980 like Mordor and Ryse that a though that FuryX would be great the scaling is suprisingly low.

In other games like COD, Watch Dogs and Asseto the performance is realy bad.

 

[at80eighty]

While the [H] article is brutal, it raises a good question. Who does the card target? I sincerely hope for their sake that new drivers will bring out improvements, which should happen. It's a long way off for the next cycle.

 

[Elfear]

1080p the gap for everything is closer. Games are simply running into CPU bottlenecks.

At 1440p its 51% faster than R290X (which the 390X improved upon a lot)

As it shifts to GPU bottleneck, the gap widens, as expected:

.

True, Fury does make significant progress over the 290X but look at how it does compared to the 390X since they have the same core clocks. At 1440p it's only 20% faster. With 43% more shaders and the added efficiency gains from GCN 1.1 to 1.3, I expected a lot better.

 

[utahraptor]

Did Anandtech receive a card to evaluate?