AMD R9 Fury X reviews!

[destrekor]

destrekor, I have 2 Sapphire Tri-X R9 290s in CF and under water (EK blocks) and hoped the Fury X would match their performance. Simply put it doesn't.

I'm probably going to wait until the second generation HBM2 memory comes out next year.

If you have 290X Lightnings in CF no real upgrade unless you buy 2 FuryX's

What cpu are you running them on?

2600K @ 4.4GHz.

I was hoping my 290X purchase (February of this year) would be temporary but I'm now more inclined to wait until the 14/16nm lineup (along with HBM2).

I might, might just do a system rebuild toward the end of this year with Skylake, haven't felt pressured to move away from SandyBridge, not the 2600K at least, but the performance of the latest chips of the same class are finally starting to make a difference in the latest games.

 

[JDG1980]

I think full tonga would be hit a little with memory bandwidth penalties.

I don't. The 384-bit bus on Tahiti was probably overkill to begin with, and especially so if faster RAM is used. (The 7970 debuted with GDDR5 running at 1375 MHz.) AMD erred on the side of higher bandwidth than needed, because Tahiti was a dual-use card, intended as much for professional HPC/GPGPU applications as for gaming.

Note that when tested at stock core clocks (1000 MHz in both cases), the R9 280X shows absolutely no benefit over the 7970 GHz Edition from the additional memory bandwidth given by its faster GDDR5. This indicates that whatever was holding Tahiti back, it wasn't memory bandwidth. With Tonga's delta color compression, a 256-bit bus should be plenty. And if they need a bit more bandwidth, it would be easier to do so by increasing the RAM speed from 1375 to 1500 MHz than to increase the bus width. Maxwell got good results by using narrower buses and faster RAM.

Tahiti, Tonga, and Fiji stick out as having fewer ROPs than they should.

Cape Verde - 640 shaders, 16 ROPs
Pitcairn - 1280 shaders, 32 ROPs
Tahiti/Tonga - 2048 shaders, 32 ROPs
Hawaii - 2816 shaders, 64 ROPs
Fiji - 4096 shaders, 64 ROPs

It's clear that the highlighted parts are being shorted in ROP count, and with Fiji we can immediately tell how it affects performance. I think if Tonga had 64 ROPs, its performance would be quite a bit closer to GM204 at 1080p. If Fiji had 128 ROPs, it would likely not show the embarrassing performance deficit it exhibits at that same resolution.

AMD seems to like strapping huge shader arrays to their chips. It helped with tahiti/pitcarin with tahiti pulling away as games became more shader intensive.

Pitcairn is actually one of their better balanced chips. Take the same number of shaders, TMUs, and ROPs, engineer it with GCN 1.2, do some tweaks for better clocks (pipeline length increase?), update the fixed-function blocksm and put it on the 14nm FinFET process, and you'd have something reasonably competitive at the low end.

 

[biostud]

I think fury needs dx12 to stretch its legs.

http://www.hardwareluxx.com/index.p...5798-reviewed-amd-r9-fury-x-4gb.html?start=21

 

[MentalIlness]

http://www.tweaktown.com/reviews/72...width-memory-hbm-video-card-review/index.html

http://www.legitreviews.com/amd-radeon-r9-fury-x-video-card-review_167134/5

 

[ZGR]

When it's on a game that runs well, Fury X is near 295x2 performance. When it isn't, it is barely faster than 290X. I wonder what's wrong?

[H] is saying that the Fury X is VRAM bottlenecked on Dying Light at 1440p, yet on 4k, it is close to 980 ti performance.

I'll be interested to see how upcoming drivers change performance.

 

[someEEguy]

I think fury needs dx12 to stretch its legs.

http://www.hardwareluxx.com/index.p...5798-reviewed-amd-r9-fury-x-4gb.html?start=21

Seems like it...

Moving to Mantle versus DirectX 11 performance we can see that our Fury X picks up almost 20 percent more performance by switching to AMD’s own API. This bodes well for performance in upcoming DirectX 12 and Vulcan games.

https://semiaccurate.com/2015/06/24/amds-radeon-goes-premium-with-the-fury-x/

 

[MeldarthX]

When it's on a game that runs well, Fury X is near 295x2 performance. When it isn't, it is barely faster than 290X. I wonder what's wrong?

[H] is saying that the Fury X is VRAM bottlenecked on Dying Light at 1440p, yet on 4k, it is close to 980 ti performance.

I'll be interested to see how upcoming drivers change performance.

H review got to be one of the worst reviews I've read from them in a long time. Its like the guy already came to his conclusion before the review and made his review to fit it.

Its down to drivers - whole new memory subset; there is a lot of performance still there to tap. Learning how to do it

 

[3DVagabond]

I've only read the 1st 2 pages but I thought the reason for the separate forums was to eliminate the AMD vs. nVidia back and forth? We have more people talking about why people should buy nVidia than talking about Fury.

If this has already been spoken on then just ignore this.

 

[2is]

I've only read the 1st 2 pages but I thought the reason for the separate forums was to eliminate the AMD vs. nVidia back and forth? We have more people talking about why people should buy nVidia than talking about Fury.

If this has already been spoken on then just ignore this.

This is a highly anticipated flagship release and probably the last one we will have for the next year plus. You're dreaming if you hunk simply segregating the GPU forum is going to keep AMD talk in the AMD forum and nvidia talk in the nvidia forum.

Not to mention, it's been pretty civil, not really sure what you're complaining about.

 

[Kallogan]

I think fury needs dx12 to stretch its legs.

http://www.hardwareluxx.com/index.p...5798-reviewed-amd-r9-fury-x-4gb.html?start=21

wow !

guess i'll wait for dx12 perf/ new drivers before taking a decision

 

[krumme]

wow !

guess i'll wait for dx12 perf/ new drivers before taking a decision

Yeaa. Damn. That is an extreme difference vs 980ti at about 50%.
Lets hope prices fall for fury before dx12 hits. Lol.

 

[Glo.]

Well, the reviews are a pretty big blow in the head, IMO for someone who was hoping that Fury will amaze us.

It is not amazing, however its not lacking anything. The problem I have with it its that it was designed, IMO from the ground up to be high resolution part.

Second thing is the inevitable jump in performance when we will see DX12 games. And here's the main thing.

For a person who is buying GPU right now, I think overall is better to buy a GTX 980 Ti. However, for someone who would like to have some future proof for gaming, Fury is overall better option, thanks to water cooling, thanks to high resolution performance. I think Drivers, and DX12 will shake things a lot. But for DX11 and lower resoltuins than 4K better option is GTX 980 Ti.

I will have extremely hard time to decide what to pick...

 

[KingFatty]

The issue with the "better for the future" speculation is that the Fury X has 4 GB of ram. So that kind of messes with the line of thinking that you should speculate and buy it for future enjoyment.

 

[maniacalpha1-1]

Have there been any reviews for the Fury/Fiji Pro? When did/will that release?

 

[Stuka87]

Have there been any reviews for the Fury/Fiji Pro? When did/will that release?

No official date for the Pro last I saw, or the air cooled X.

 

[flopper]

No official date for the Pro last I saw, or the air cooled X.

14 july

 

[Despoiler]

Yeaa. Damn. That is an extreme difference vs 980ti at about 50%.
Lets hope prices fall for fury before dx12 hits. Lol.

Fury is good in the overhead tests yes, but it's the GCN architecture in general that is responsible for the score. My Gigabyte R9 290 Windforce only gets 1 million less in the overhead test. (results below) So it's still massively ahead of anything Nvidia has despite being last gen.

http://www.3dmark.com/aot/33649

 

[pj-]

Yeaa. Damn. That is an extreme difference vs 980ti at about 50%.
Lets hope prices fall for fury before dx12 hits. Lol.

Isn't this just the draw calls test shown in a different way?

 

[flopper]

Isn't this just the draw calls test shown in a different way?

yea Mantle shows it also
http://semiaccurate.com/2015/06/24/amds-radeon-goes-premium-with-the-fury-x/

 

[exar333]

I think fury needs dx12 to stretch its legs.

http://www.hardwareluxx.com/index.p...5798-reviewed-amd-r9-fury-x-4gb.html?start=21

What's the point then? DX12 titles will not even start to come in until next year, and probably not until late in 2016 to fully start taking advantage of it, if even then. By then we have 16/14nm.

NV has shown great improvements on DX12 too, so I doubt AMD will make up a lot of headway here. Both companies have had years to optimize...

This reminds me a lot of the Win8 scheduler stuff from BD that never materialized...

 

[maniacalpha1-1]

Fury is good in the overhead tests yes, but it's the GCN architecture in general that is responsible for the score. My Gigabyte R9 290 Windforce only gets 1 million less in the overhead test. (results below) So it's still massively ahead of anything Nvidia has despite being last gen.

http://www.3dmark.com/aot/33649

I don't really know much about GPUs and DX11/12/mantle/etc on a technological level, but I'll ask anyway, what kind of FPS improvements could be expected solely from DX12? I mean, if going from 290X to Fury X takes FPS from 47.6 to 65.2 (using Anandtech Bench's Battlefield 4 1440p http://www.anandtech.com/bench/product/1439?vs=1513), what kind of improvement might DX12 alone on a 290X bring?

Also, I was reading an article from Stardock's CEO, saying that DX12 will improve multi-core CPU usage. (http://wccftech.com/wardell-explains-companies-dont-promote-dx12-benefits/)

Given that MMOs, or single player games with lots of units are very heavy on CPU, how much could DX12 impact those kinds of games?

And, more importantly, will a 290X be able to advantage of DX12 fully, or will that really need a Fury/X?

 

[flopper]

I don't really know much about GPUs and DX11/12/mantle/etc on a technological level, but I'll ask anyway, what kind of FPS improvements could be expected solely from DX12? I mean, if going from 290X to Fury X takes FPS from 47.6 to 65.2 (using Anandtech Bench's Battlefield 4 1440p http://www.anandtech.com/bench/product/1439?vs=1513), what kind of improvement might DX12 alone on a 290X bring?

Also, I was reading an article from Stardock's CEO, saying that DX12 will improve multi-core CPU usage. (http://wccftech.com/wardell-explains-companies-dont-promote-dx12-benefits/)

Given that MMOs, or single player games with lots of units are very heavy on CPU, how much could DX12 impact those kinds of games?

And, more importantly, will a 290X be able to advantage of DX12 fully, or will that really need a Fury/X?

engine built for dx12, 100 to 600% depending on game.
if a game does 40fps today it would mean more fps than you need.
stardock brad wardell been vocal about it.
he run tests which showed a lot of difference in the ballpark of dx11 13fps and dx12 120fps.

http://www.dsogaming.com/news/dx12-...sed-gpu-in-new-test-way-beyond-console-stuff/

It will differ on type of games, MMO, rpg, rts etc...

 

[maniacalpha1-1]

engine built for dx12, 100 to 600% depending on game.
if a game does 40fps today it would mean more fps than you need.
stardock brad wardell been vocal about it.
he run tests which showed a lot of difference in the ballpark of dx11 13fps and dx12 120fps.

http://www.dsogaming.com/news/dx12-...sed-gpu-in-new-test-way-beyond-console-stuff/

It will differ on type of games, MMO, rpg, rts etc...

Could this do anything to further the cause of MMOFPS games? Right now there's really only Planetside 2, and pretenders/dinosaurs like Heroes and Generals/WW2Online.

 

[Despoiler]

Could this do anything to further the cause of MMOFPS games? Right now there's really only Planetside 2, and pretenders/dinosaurs like Heroes and Generals/WW2Online.

It's going allow your computer to draw a lot more people/vehicles on the screen before choking or you can get the same amount of people at full graphics without graphics lag. PvP MMOs that have large battles use a lot of graphic tricks to allow them to even be playable. Usually they reduce the texture quality dynamically. DX12 is going to be great for all games, but it will be HUGE for MMOs and RTS. Check out the video below. These battles are not even possible with DX11.

https://www.youtube.com/watch?v=t9UACXikdR0

On your first question. In BF4 Mantle gave AMD 10-15% better max FPS, but the min FPS shot up 50-80%. Frametimes are 40% lower than DX11. Gameplay feels so smooth. That's on a mid to high end computer. DX12 is actually looking like it's going to be better than Mantle.

 

[flopper]

Could this do anything to further the cause of MMOFPS games? Right now there's really only Planetside 2, and pretenders/dinosaurs like Heroes and Generals/WW2Online.

absolutly but as anything dx12 engines are needed and atm dice been in the forefront of games but the rate of engines and games coming and when is at best guesswork atm.