AMD Radeon R9 Fury X 4 GB

[haswrong]

...
if it line up with the 980Ti in price and in performance (2% under??? a big gap indeed) then all is technically right no? if it was priced like a Titan X it would be different
...

tit x has 3x more vram and is faster while consuming less energy.

 

[mroofie]

every reviewer i talked to mentioned the pump noise. the coil noise is a separate noise depending on gpu load.

slower than the 980 ti

Amd needs to step up fast

 

[the54thvoid]

Bear in mind 2 things of course:
1: NVidia cards with boost 2.0 all differ and some boost at stock speeds well beyond the clocks (not saying this one was just pointing it out).

Yeah but some boost stupidly high. Maxwell's perf lies in it's higher clocks. The 1400+Mhz boosts really push 980ti's to 15% improvements.

2: AMD cards always (At least in the last 2 generations) have gotten much better with time while NVidia get worse (Just look at reviews from the past till now, they eventually start beating out GTX 780ti)

The 780ti was (tin foil hat required) hobbled to push the perceived perf of the 980ti. The 780ti was 'allowed' to get worse because Nv had a faster card to sell. In other words, Nvidia had a lot of room to work with to make their products look even better (against their own brand)

@W1zzard - On topic of the scores - why does it falter so much at <4K res? I've never understood that. Obviously the 980ti must take a hit from 1080p>1440p>4k. But how does the Fury X do so badly at lower res?

 

[Vulpesveritas]

a disaster ... funny


a fail ... oh ... yes it's a fail ... and a big one indeed

Compared to the hype I felt? Yes, yest it is.

I wanted to wake up this morning and squee about how amazing the Fury X was after I read the reviews. I wanted it to be fast enough to justify my paying more for it despite my 1080p-ness.

I wanted a Titan X Killer, and instead I saw a 980ti match.

So yes, it was a fail from the hype-stab side of things.

 

[FrustratedGarrett]

As many others, extremely disappointed by this release.

Also having a lot of trouble justifying the relatively high score in the review. 9.2 for this card? It is underwhelming across the board:

- price/performance ratio is shit, only card doing worse is Titan X.
- extremely high power draw
- performance/watt is last-gen at best
- bad - virtually non-existant overclocking potential

Price point is OK-ish, but really only makes sense if you love AMD because the card performs sub-par across many different resolutions and games. For 4K it may be a sensible choice, but that is about it, and then it still is a coin toss between Nvidia/AMD offerings.

So much for the big marketing push and incredible performance/watt jump of HBM. To be honest, I had expected a slight jump in performance, even if only 5% above Titan X, but even that is nowhere to be seen. Basically AMD put a lot of effort in new tech that does not perform better than GDDR5, and the fundamental GCN issues such as high power draw while watching Blu-Ray and at maximum output have not been tackled, if anything they have gotten slightly more pronounced.

The most surprising thing however, is the whopping 4096 shaders used and the virtually unlimited bandwidth and how little performance it all adds. It's almost like the card doesn't get to stretch its legs. Weird.

It performs quite well in most games and it's a lot quieter and cooler than 980 TI. Your post screams fanboy 0-

 

[mroofie]

The massive variation in the Fury's relative performance, especially how badly it performs in GameWorks titles is a real problem.
The card seems to perform as well the TitanX in the neutral titles, but in titles like the Witcher 3 and Project Cars, it falls behind badly. In GTAV Nvidia does better but me thinks its due to their crappy HBAO+ implementation.

http://www.hardocp.com/article/2015...mage_quality_comparison_part_5/7#.VYqmJLYpnIU

ALien Isolation, Battlefield 4 all amd titles come on now

 

[mirakul]

Yeah but some boost stupidly high. Maxwell's perf lies in it's higher clocks. The 1400+Mhz boosts really push 980ti's to 15% improvements.



The 780ti was (tin foil hat required) hobbled to push the perceived perf of the 980ti. The 780ti was 'allowed' to get worse because Nv had a faster card to sell. In other words, Nvidia had a lot of room to work with to make theiur products look even better (against their own brand)

@W1zzard - On topic of the scores - why does it falter so much at <4K res? I've never understood that. Obviously the 980ti must take a hit from 1080p>1440p>4k. But how does the Fury X do so badly at lower res?

It's the notorious draw call problem AMD having with DX11. That issue is bigger in lower res.

 

[buggalugs]

I'm happy with the card's performance but if the only option is the closed loop water cooler I'm not going to be buying it. I was hoping Asus and MSI would be making their own versions. The single 120mm radiators perform worse than a decent air cooler, then you have to put up with pump noise, and more noise than a decent highend air cooler. I don't know why AMD would limit their sales by releasing a water cooling only option.

I don't understand why people are so impressed with 60 degrees gaming. I get that on my Asus 290X DCUII

It looks like HBM is the future though for highres 4K displays and multi-monitor. With a few multi-monitor resolutions in the review the FuryX would appear much better. It looks like the FuryX stretches its legs at high resolutions. Theres enough performance and new tech to keep AMD relevant, especially for 4K users which will be a lot of people now that 4K monitors are cheap, but I'm not getting one with a closed loop cooler.

Most people willing to spend $600 on a graphics card will probably go 4K in the near future, making this card a serious option.

 

[DarkOCean]

So we waited so long just for this ? ... it's slower, more power hungry, it has way less memory, sucks at overclocking (on water!) while costing the same as a 980ti ... so why would anyone buy this?

 

[GhostRyder]

It's a disaster. There's no sugar coating it. These will be sitting on shelfs as paper weights when volume ramps. The driver must be written by a 5th grader and gameworks' cheat....optimization finishes it off.

How, enlighten us how this card is a disaster.

1: It works
2: Runs Cool
3: Priced at the area it matches
4: Smaller than the competition (If that is your thing)

Yeah but some boost stupidly high. Maxwell's perf lies in it's higher clocks. The 1400+Mhz boosts really push 980ti's to 15% improvements.



The 780ti was (tin foil hat required) hobbled to push the perceived perf of the 980ti. The 780ti was 'allowed' to get worse because Nv had a faster card to sell. In other words, Nvidia had a lot of room to work with to make their products look even better (against their own brand)

@W1zzard - On topic of the scores - why does it falter so much at <4K res? I've never understood that. Obviously the 980ti must take a hit from 1080p>1440p>4k. But how does the Fury X do so badly at lower res?

Indeed, and that is what I was referencing as some boost without people noticing which leads to variations in review across the web but not everyone's does that.

Well allowing it to get worse still is getting worse in my book but that up for a whole different discussion. But either way its still worth noting especially depending on how often you upgrade.

 

[hero1]

Thanks for another great review. I am simply shocked by the performance levels for anything less than 4K. This card has the potential to beat the 980Ti cleanly if they unlock the voltage and memory overclock. It's crazy that it can keep up with the Titan X in 4K. Looks like I'm sticking with my 290Xs in CF under water for the foreseeable future.

 

[Vayra86]

It performs quite well in most games and it's a lot quieter and cooler than 980 TI. Your post screams fanboy 0-

Three examples in a benchmark suite of over 20 games is nothing to write home about. And this is only at 4K, the ONLY res at which it only marginally excels, while performing notably worse on anything lower than that.

Relative performance is where its at. I don't care about individual titles and neither should you. You dont play three games on an enthusiast card, and you don't pick games based on their best performance on AMD or Nvidia hardware either.

 

[TheGuruStud]

I'm seriously not even arguing this BS. Look at the gd numbers!

 

[mirakul]

@W1zzard Given that Gameworks has become a well known issue, could you please make another chart without Gameworks titles? (Project cars, Watch dogs, Witcher, etc..)

 

[the54thvoid]

At least I know what my next purchase is now. It's a shame, I was looking forward to going Red again.

I'm seriously not even arguing this BS. Look at the gd numbers!

Irony is - I've seen you defend AMD a lot. Your words are open and honest.

 

[BiggieShady]

These will be sitting on shelfs as paper weights when volume ramps

I sure hope that AMD is not ramping any volumes, they should tread carefully and make a single batch for now.

 

[FrustratedGarrett]

Three examples in a benchmark suite of over 20 games is nothing to write home about. And this is only at 4K, the ONLY res at which it only marginally excels, while performing notably worse on anything lower than that.

Relative performance is where its at. I don't care about individual titles and neither should you.

That's true, but what you forget is that we have both Witcher 3 and Project cars in this test suite, and AMD performs way worse in these games. GTAV is another title were AMD is doing worse mostly due to NVidia's cheaper HBAO+ implementation in this game.

http://www.hardocp.com/article/2015...mage_quality_comparison_part_5/7#.VYq3hbYpmUk

 

[Assimilator]

I'm laughing at all the people who believed AMD's marketing hype.

@W1zzard Given that Gameworks has become a well known issue, could you please make another chart without Gameworks titles? (Project cars, Watch dogs, Witcher, etc..)

Oh get over your butthurt, the only "well known issue" is that AMD can't write drivers for s**t. W1zz uses a decent sample of popular game titles that isn't biased for or against any company. I guess you'd prefer it if he only tested AMD cards with Mantle games?

 

[HD64G]

My humble opinion on the "Fury" subject is below

In 4K resolution (all other resolutions are too low for these monsters after all to have any problem playing in ultra settings):

Taking ALL games of W1z's review into consideration we have 10-10 for Fury X vs 980Ti (2 are draws)

By throwing the 3 totaly inbalanced games (PC, WoW and Wolfenstein are too green) into the equation we have 10 wins for red team and 7 for green team.

If we oblige the DR3 game from the competition (seems red balanced this) we still have 9-7 for Fury X.

So, having a same priced, lower temp, more quiet GPU which is somewhat better in 4K with not matured drivers yet is not good enough to buy it? Since when logic is crazyness and the opposite also?

And Nano will be a GREAT product if price is correct...

 

[GreiverBlade]

Compared to the hype I felt? Yes, yest it is.

I wanted to wake up this morning and squee about how amazing the Fury X was after I read the reviews. I wanted it to be fast enough to justify my paying more for it despite my 1080p-ness.

I wanted a Titan X Killer, and instead I saw a 980ti match.

So yes, it was a fail from the hype-stab side of things.

you can't have a titan X killer ... since it cost the same price of a 980Ti ...

 

[mroofie]

how about this ?
http://www.techpowerup.com/forums/threads/error-with-tpus-wolfenstein-benchmarks.213767/
http://tpucdn.com/reviews/AMD/R9_Fury_X/images/wolfenstein_1920_1080.gif

It just one game the rest are legit and go look at other sites same conclusion!! (Gtx 980 ti > Fury X)

 

[Vayra86]

That's true, but what you forget is that we have both Witcher 3 and Project cars in this test suite, and AMD performs way worse in these games. GTAV is another title were AMD is doing worse mostly due to NVidia's cheaper HBAO+ implementation in this game.

http://www.hardocp.com/article/2015...mage_quality_comparison_part_5/7#.VYq3hbYpmUk

The Witcher 3 is a non-issue because TPU tests without Hairworks and the performance falls right in line with expectations. The fact that AMD has trouble with their drivers is not CDProjekt RED's fault. Without Hairworks and HBAO+ the AMD cards still underperform. Project Cars, yes, true. GTA V: another non-issue, if Nvidia makes do with lighter HBAO+, what is stopping AMD for bringing a similar solution?

Drivers are an important part of the purchase and AMD has a history of being late to the party or unable to tackle issues. This is not Nvidia's fault. Frame pacing is a notorious example, it was only until the shitstorm was unleashed that AMD moved to driver optimizations. Nvidia was ahead of the game, and generally still is. This is what customers recognize.

(By the way, recent Nvidia driver issues are also on my radar, don't get me wrong)

I simply call it like I see it. If AMD brings a killer product, I'll be the first to applaud them for it. They just don't, and they keep saying they do.

 

[haswrong]

It performs quite well in most games and it's a lot quieter and cooler than 980 TI. Your post screams fanboy 0-
...

ehm... who exactly plays games at 30fps? only consolists do.. we are not them though.. 4k perf is totally irrelevant these days.. (expensive, unplayable, unenjoyable = a horrible combo)

 

[Joss]

See (hear) here for noise, minute 8 onwards.

 

[mroofie]

It performs quite well in most games and it's a lot quieter and cooler than 980 TI. Your post screams fanboy 0-

1.Only at 4K (It performs quite well in most games)
2.Water vs Air cooling

Only fanboy here is you