Unofficial R9 Fury X reviews master post.

[HalGameGuru]

I really really really doubt that. This isn't a new architecture or anything like that and even if it was, I don't think we've seen anyone say "okay let's sacrifice DirectX 9 performance to get better DirectX 11 performance". Is that even possible to do?

Again, don't expect some magical thing in the future to happen and all of a sudden the card performs like a monster. You will just end up being disappointed again.

 

With a lot of the new implementation within DX12, I think it may well be possible, especially in those techs that more effectively leverage VRAM. Between AMD's own efforts to better compress and DX12's new texture techniques my thought is that Fury X may be situated to leverage these things to benefit it under DX12 ESPECIALLY in 4k. Now I'm not so sure what kind of an effect, intended or un, DX12's other features like ganged VRAM pools and asymmetrical GPU pairings will have, but I do doubt that AMD put a reliance on THAT for Fury, but the specific new techs within DX12, which could ameliorate issues of lacking VRAM capacity, and multithreading GPU processes allowing the HBM transfer speed and wider bus a more prominent spot in performance especially at higher resolutions makes me want to see further performance testing with the next gen APIs and standards implemented.

[MegaDave91]

This is all fine and dandy but where's the 4k surround benchmarks?

Now that it has finally reached the hands of the reviewers and is shown (overall) to trail slightly behind the 980ti--I don't think we will be seeing 4K surround benchmarks. 

 

We are not even remotely close to single gpu cards being capable of gaming at 11520x2160p.

[VagabondWraith]

I really really really doubt that. This isn't a new architecture or anything like that and even if it was, I don't think we've seen anyone say "okay let's sacrifice DirectX 9 performance to get better DirectX 11 performance". Is that even possible to do?

Again, don't expect some magical thing in the future to happen and all of a sudden the card performs like a monster. You will just end up being disappointed again.

I don't get people saying "wait for Windows 10!" Or "drivers will make our cards 50% faster than Maxwell!" Or "HBM will be unleashed upon the world with DirectX 12!" AMD has been working on this what, 5 years? Are you telling me that future updates are gonna fix performance that much after a 5 year development cycle? That's pretty bad. I'd expect something twice as powerful as this in the first place.

 

Sure, drivers will certainly help optimize architecture. There will be improvements. But I believe this is the best we can expect Fiji to be +/- a few percent or so.

 

And just because AMD releases driver optimizations/improvements, doesn't mean Nvidia will sit on their ass. They'll make sure that Maxwell remains on top.

[Noctus]

I'm gonna be THAT guy, but i think currently, and this has a lot to do with games having an nvidia bias and whatnot more than the hardware, you'll have a better experience on the NVIDIA side of the fence.

Alternatively, 295x2 or a couple of 290xs are not a bad solution at that pricce point. Crossfire can be a massive headache, but even stuck with one card it's pretty damn good.

no exact details, yet, but the Fury Nano is under Fury and unde Fury X - in that order

Hmm looks like I'll skip Fiji and blag a 980ti this coming black Friday then.

Shame cos I was hoping AMD to blow us away this time (not saying they haven't produced a decent card, just with the hype I expected alot more) but I guess I'll be team green this time around again.

[Briggsy]

Personally a lot of the results so far are less than I was hoping for, but I have a theory as to why. Could the Fury X, with its 4GB VRAM, a presumed limitation, and other changes in bias from earlier chips indicate its focused on future performance in DX12 and on Windows 10? What with all the new tech being implemented in DX12 especially in regards to textures will HBM be more heavily and effectively leveraged in future games using DX12 than most of what is currently being benched? Have we seen any numbers on the difference between DX11 and Mantle for the Fury yet? I just wonder if AMD may have focused their efforts on performance dividends over the next handful of years rather than just at launch...

 

I get the feeling that the current mixed bag numbers may turn into a bigger badder beast with the more advanced API's making use of its new set of hardware techs and changes in how VRAM is being used in engine under DX12. 

 

Historically AMD cards are always designed for the long haul, and in this case DX12 features appear to be the focal point of their marketing, as well having high memory bandwidth for motion prediction in VR. The kicker is that AMD made the 390/x cards 8GB, making the Fury/x cards look short term.

 

Also, AMD drivers typically have more cpu overhead than Nvidia drivers due to a variety of factors that are hardware related (GCN cores need to be programmed for properly, while CUDA cores don't need to be), which makes DX12 much more ideal for AMD than DX11.

[Takata]

This is all fine and dandy but where's the 4k surround benchmarks?

Go look for the NCIX video where they concluded that 4 Titan Xs couldn't get you a good experience (too much stuttering) with a 4k surround configuration.

[zMeul]

that VRM cooling thingy isn't doing a proper job:

 

[LAwLz]

Historically AMD cards are always designed for the long haul, and in this case DX12 features appear to be the focal point of their marketing, as well having high memory bandwidth for motion prediction in VR. The kicker is that AMD made the 390/x cards 8GB, making the Fury/x cards look short term.

 

Also, AMD drivers typically have more cpu overhead than Nvidia drivers due to a variety of factors that are hardware related (GCN cores need to be programmed for properly, while CUDA cores don't need to be), which makes DX12 much more ideal for AMD than DX11.

[Citation Needed] on both claims (AMD cards being designed for the long haul, as well as GCN requiring better written programs).

[GrimNeo]

2 of these Fury X's will look good in my second pc. Sadly it will be a couple months before I can get them.

[MEC-777]

that VRM cooling thingy isn't doing a proper job:

 

 

Can't see the pic...

[Sunako]

[Citation Needed] on both claims (AMD cards being designed for the long haul, as well as GCN requiring better written programs).

no citation needed. just because you disagree.

cuda has been a standard across nvidia for the longest time now  making it easy to fix for any generation not so much so on the amd  when a  new problem pops up

[zMeul]

Can't see the pic...

check now

[GrimNeo]

check now

NVM lol. 104C wow

[zMeul]

link?  Work blocked this pic

http://www.hardware.fr/medias/photos_news/00/47/IMG0047700.png

 

and link to article: http://www.hardware.fr/articles/937-8/temperatures-nuisances-sonores.html

[GrimNeo]

http://www.hardware.fr/medias/photos_news/00/47/IMG0047700.png

ty sorry. updated post. It was blocked on ie 8 but then I used firefox to view it.

[VagabondWraith]

that VRM cooling thingy isn't doing a proper job:

 

IMG00477001.png

Isn't that worse than Titan X backside? Ouch.

[zMeul]

Isn't that worse than Titan X backside? Ouch.

yep:

 

[Briggsy]

that VRM cooling thingy isn't doing a proper job:

Attachment IMG00477001.png

 

^^ that must be with Furmark or something, and without any air flow.

 

Edit* Nevermind, just realized your pic is from the bottom side

 

toms Hardware gaming loop

 

 

toms hardware Stress test

[Notional]

Isn't that worse than Titan X backside? Ouch.

 

Nope:

 

The Titan X was on the actual ram module, which are not made to be that hot. The Fury X is on the VRM part, which is made to handle those temps.

[MEC-777]

that VRM cooling thingy isn't doing a proper job:

 

 

 

Wow, that is pretty toasty. That's probably why it's voltage locked and they don't want you overclocking it beyond +100mhz. Once you start adding voltage is when the VRM and GPU temps really start to climb.

 

I'm wondering if the Fury (non-X) with the aftermarket cooling solutions will have better VRM cooling and actual overclocking potential...

[zMeul]

Wow, that is pretty toasty. That's probably why it's voltage locked and they don't want you overclocking it beyond +100mhz. Once you start adding voltage is when the VRM and GPU temps really start to climb.

 

I'm wondering if the Fury (non-X) with the aftermarket cooling solutions will have better VRM cooling and actual overclocking potential...

my guess is either the copper pipe thinky is pretty much useless or something else is going on

and by something else, I mean the card is factory overclocked  - those temps don't look right either way

[Linusinium]

Good job OP. One central thread with all relevant reviews and the discussions contained within. Much better than having a thread for each review or scattered discussions.

Think all of us would have benefited if the Fury fared better.

I keep going back to that old saying I heard on Anandtech, to paraphrase " nobody knows more about upcoming hardware than the most fierce competitor". Nvidia saw Fury coming to kill the Titan so they decided to kill it themselves with the TI.

[zMeul]

regarding the Fury (non X):

AMD says one of the biggest differences between the Fury X and the rest of the Fury line is that its water-cooled card shouldn't throttle under load (a big issue we identified with the Radeon R9 290X when it first launched). Instead, this board should always run at its highest clock rate. The air-cooled cards, by comparison, are expected to scale back as their heat sinks and fans are saturated with thermal energy.

http://www.tomshardware.com/news/amd-fury-x-fiji-preview,29400.html

 

expect monsters 

[Jord1990]

Good job OP. One central thread with all relevant reviews and the discussions contained within. Much better than having a thread for each review or scattered discussions.

Think all of us would have benefited if the Fury fared better.

I keep going back to that old saying I heard on Anandtech, to paraphrase " nobody knows more about upcoming hardware than the most fierce competitor". Nvidia saw Fury coming to kill the Titan so they decided to kill it themselves with the TI.

Finally some one that sees it. We should all thank AMD for releasing a beast of a card for the price. It was meant to be a titan killer and if you look at the price/performance it did. The problem is that Nvidia saw it coming and indeed killed the Titan themselves to not give the pleasure or gpu sales to AMD. Without the fury X there would not be a 980ti at this price point.

[SpaghettiCarbonara]

Not sure if anyone else has posted it yet, but I reckon Hardware Canucks did the best review, as always.

 

 

All of this is fairly disappointing, but remember - the 290x launched at something like $550 and the 7970 was the same and now look at their prices! Later down the line there'll be price cuts for the Fury X and it'll definitely be worth the money then. Also driver updates and what-not.

 

Until then, we'll have the Fury and the Nano to keep us company! Remember - the 290 was AMD's best card for the price out of the 200 series, the Fury/Nano will be the same with the 300/Fury series. 980 killer anyone?

 

 

please please please get some traction AMD, I'm sick to death of Nvidia and Intel and their army of shills