AMD Radeon R9 Fury X 4GB Review

Page 12 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.
I do not see any 4K being tested here, only UHD. In other comments I see 4K mentioned, but again only UHD has been tested in this article. Double 1920x1080 is 3840x2160 which is UHD.
 


What is this 4K you speak of, is it DCI 4K Native, DCI 4K CinemaScope, DCI 4K Flat. There is no such thing as just 4K.
There are also other competing standards other then DCI.
 




I've got 980 sli and don't feel the least bit "ripped off". Great performance for the money and constantly getting new updated drivers for new game releases. In contrast I owned a 290x for an afternoon and quickly returned it due to stale unstable 9 month old drivers and performance that couldn't match my highly overclocked and well cooled 680 GTX it was meant to replace.

 


hardocp has great tech and overclocking articles. Reading their review of my MSI 980 gaming 4g card showed me how to overclock my 980 to get best results. Their benchmark reviews seem fair and I like their 1440 and 4k apples to apples charts.
 


I doubt just heat sink will help for this kind of card... it a hot card...
if pump fail u will see steady temp increase, unlike on air u can see it rapidly increased
water is better transfer medium than air.. u know...
by it's nature will circulated by it's own to some extend..

 


If you read the conclusion of their Fury X review, it's ridiculous.There are contradictions and they seem imply that they know more about HBM than AMD does. I don't doubt their 4k test numbers being accurate, but the conclusion was...off.
 


AMD gave me the (used) card in rotation with other media one day before launch (afternoon!) and I had to send it back on launch day! No time for accurate and additional testing. This was so frustrating and shows us perfect AMDs thinking this time. But don't worry - I'll get today my own card, but not from AMD. It was bought and is 100% retail. I will made two things as follow-up: in-depth power consumption analysis with different games and workloads and a day with Fury and me in my audio-lab to detect the difference between pump noise and micro-bubble noise 😉

 


The cooler makes up the difference in performance and the design. 5% is in the margin of error imho. On toms review there are titles that give the Fury a 10% lead.
 

Water may have higher specific heat but passive circulation is abysmally low unless you have a large temperature gradient between a cold inlet located below the hot outlet. On a GPU, the inlet and outlet would be horizontal with the board and have effectively no vertical difference. If you put the same temperature liquid the bottom of both sides of a cooling loop, net circulation through the loop will be zero. Depending on the type of pump used, the pump may also prevent natural convection. Yes, there may be some convection vertically through each tube but the contribution would be negligible.

Modern high-end air-cooled GPUs have three fans these days. They might not be able to hit their top performance after losing a fan or two but they should still be able to keep going for much longer than a liquid-cooled GPU after losing its one and only pump / liquid loop.
 


At the time of purchase, pulling ATI out of bankruptcy. Now, Radeon's are keeping AMD out of bankruptcy... lol

At one point, there was talk about AMD owning Nvidia, however that fell through when Nvidia wanted it the other way around...
http://www.tomshardware.com/news/AMD-ATI-Nvidia-GPU-Tegra,14795.html
 

I have not seen anything that puts it at 10% above the 980Ti consistently. In some it is, in others it is behind.

The issue I see is that while it has that cooler it is only to keep the GPU that would otherwise run way too hot at that level of performance on air. It is obvious since the majority of overclocking reviews give it a 5% overclocking result. That is not enough to make up the difference with the 980Ti which can do around 25% on air and even more on water.



It was a smart choice, just at the worst time. They were not able to keep OEM channels supplied with CPUs due to only having one FAB, spent too much for ATI and even worse it was doomsday for K8 as Intel had released Core 2 that was creaming AMD at their own game (high IPC, low power).

While it was smart for the APU side of things, I doubt they would have had the R&D to design their own GPU, it was bad timing as shortly after that they started to lose market share and eventually had a flop of a CPU that caused them even worse loses in the lucrative server market.
 


I would love to see it benched consistent with the 390x review and the FuryX results added to the charts.

 

The Fury X is not "sold out" due to being a phenomenal commercial success. It is "sold out" simply because availability is so low that some fairly large retailers are still waiting for their first shipment.

There isn't much merit to being sold-out when it only means something like all of 10 units from the first shipment and no ETA on the next one.
What did I just say? Is there some kind of problem with reading comprehension going on? Doesn't matter what the reasons for it being sold out, the simple fact that supply is low and demand is high compared to supply so the price is about right or should be jacked up. If at some point the supply is increased where it's sitting on the shelf collecting dust then you lower the price.
 


http://www.legitreviews.com/amd-radeon-r9-fury-x-complete-teardown_166851

It is a Coolermaster design and it looks like most AiO radiators so I would assume a high SP fan would be better. Not sure why they have such a low RPM fan on it.



Careful on the wording, no need to push towards personal insults.

This, however, a artificial high demand. High demand is like when the bitcoin mining with GPUs was taking off and you couldn't get a R9 290X at all let alone for less than $1K. There was plenty of supply but the bitcoiners were buying them up and until GPUs became no longer viable they stayed artificially inflated in price.
 
I agree that if the Fury X continues to sell out quickly for a long period of time, then AMD should keep the price where it is. I don't think this will happen.
Here's some points I believe you missed.

1. Supply is low right now. Most big stores don't have many in stock. Of course they will sell out quick. There's so little supply that even a small demand will exhaust it.

2. The initial buyers are most likely fanboys. From reading the initial reviews, most people who don't have a strong loyalty won't buy it over the 980ti this quick. What I'm saying is that the Fury X doesn't stomp out the 980ti, thus there's no big incentive to buy it immediately. Only those who have a strong loyalty to AMD will want it so bad that they absolutely have to have it now will buy it immediately. That's where the first sales are coming from. (I don't mean to make a blanket statement about Fury X buyers. I do realize that there are some that bought the Fury X for other reasons, and I respect that, but I believe that for the most part, my above point is true.)

3. Unless you are using 4K, the 980ti is the best choice so far. Benchmarks clearly state the the 980ti has a slight edge over the Fury X, plus Nvidia has a clear advantage in the area of driver support. Add in that superior overclocking and game optimizations, and it's clear that the 980ti has an edge over the FuryX. The only way that the FuryX will have a chance at having decent long-term sales is for AMD to lower the price on it so that it's not the same as the 980ti. From a buyer's perspective, at this price level, it's worth paying the extra $20 to get the 980ti and thus some extra performance, but it probably isn't worth paying $75 extra for the slight performance increase. If AMD lowers the price of the Fury X so that it's $75 lower than the 980 ti, then it becomes a compelling product, but if it's stuck near or at the same price, then the 980ti is the better option of the two.
1. That a given.
2. The fury X is meant for fanboys or enthusiasts just like the 980 Ti and Titan X are meant for fanboys/enthusiasts. Also, there's a whole library of games that aren't tested in these reviews. Most reviews test popular or recent games which gives a sort of bias interpretation of results. If you want to see how it compares with a giant library of games then go to http://www.techpowerup.com/reviews/AMD/R9_Fury_X/
where it handily beats the 980 Ti and Titan X at several older games
3. Now is not the time to lower price on anything. It's time to milk as much money as possible while sales are hot. Also, since performance is relatively the same you can argue that some people prefer cooler gpu over heat, or noise levels. Also, a water cooler is a feature that would cost an extra $100 if you were to install it yourself. Let's be honest, people who are buying the Fury X, Titan X, and 980 Ti are fanboy/enthusiast. No normal people are going to shell out this type of money for any of these gpus so long term sales is irrelevant. These are extreme level. The main sales that will determine market shares are the high and mid-level cards like the GTX 960, 970, 980, R9 285, R9 290x, R9 290, R9 390X, R9 390, etc.
 

In regards to your third point, I totally agree. Any company that has executives with any sort of brains will try to milk as much as possible from a sales rush. I was just saying that the price will have to drop after a period of time in order to consistently draw people from the enthusiast group.
 
Please run benchmarks on Windows 10 with DX12. 3dmark draw call test, and any others you can find that are compatible. Also OPENCL test. Thanks
 


They paid way too much for ATI, that put them behind the 8-ball. Since that purchase they seem to have been a non-factor in CPU development. If they had bought the assets after ATI went default they may have gotten everything at a good discount and not have gotten sucked down that red ink black hole. Of course Nvidia may have bought some of the assets too.
 
I have a koolance erm-2k3u and pump-450 noise and fans can be loud.
What I did, was to remove back fans, leaving just the big fans, and disconnecting the pump, and the fans from the koolance controller, that has steps, instead I connected those to lab grade variable power supply's instek gps-1850d & 1830d, now I can fine tune the noise with the resonance, I thought that lower dc would give lower noise but it turns out it does not, around 9.25vdc the pump resonance is near silent, and fans a can go lower.
A quick easy mod I can think of reading this review, It's to replace the fan with dual antec tricool 120 that have a L-M-H 3 speed switch, push pull dual fan at Low setting would be much better, than single at high.
 
Lol AMD! 4097bit, HBM memory and this is it? Nvidia uses like what? 256bit, no HBM memory and still kicks AMD in nuts. AMD will only be good at dropping prices, nothing else.
 

The 980Ti uses a 384bits wide memory interface running at 7GT/s. HBM might be 4096bits wide but only runs at 1GT/s, which gives it about 50% more raw bandwidth.

The Fury's problem is not HBM. It is the rest of the architecture not being able to keep up with HBM's bandwidth - AMD messed up the resource balance somewhere, such as the ROPs. On the plus side, AMD now has hands-on numbers and will hopefully be able to avoid repeating the same mistakes with next year's models.

HBM2 next year will double the bandwidth per stack, putting it well beyond GDDR5's reach.
 
Status
Not open for further replies.

TRENDING THREADS