AMD Fury X Reviews

[Silverforce11]

I guess H wont be receiving any more review samples. That was a way harsh review I must say. Saying AMD's GPU progrss is now more closely resembling it's CPU progress was a bit too far.
Does anyone commend AMD for this higher performance bigger die with less power consumption than 290x? Or can this simply be attributed to HBM and a top end 56C water cooling temp? Cooler temps do reduce power consumption. Thoughts?

Well its definitely a combination of factors that help it achieve a big efficiency gain compared to their prior gen stuff.

What's interesting looking at computebase summary chart is it is very close to 50% faster than R290X.

But though I don't want to mention it, the recent inclusion of more GameWorks titles means that 50% faster than R290X doesn't cut it like it would have a few months ago. Some of those same sites had Titan X at 45-48% above R290X not long ago!

I would think at the price they are asking, it has no redeeming qualities besides that water cooler.

If it was air cooled custom variants at $549, yeah sure, definitely recommend.

Edit: Look at those charts and how well it stacks up according to that ChipHell leak from LAST YEAR. Spot on. But the difference now that it can't beat Titan X, more GW titles have come.

Also, [H] have discredited themselves, every review so far has show it actually scales better at 4K.

HWC:

"In UHD scenarios this card is easily able to trade blows with the GTX 980 Ti (if it wasn’t for a lack of optimizations in Dying Light, it would actually be ahead by 2%) and gives the uber expensive TITAN X a swift kick in the pants."

 

[Silverforce11]

It didn't embarrass itself, but it didn't move the performance bar one millimeter. Great option at least for those who like to stick with AMD, plus it's quiet, cool, and small. However, even assuming voltage tools become available I just don't see this thing being able to attempt hanging with the Classified, Lightning and HOF variants of the 980Ti, maybe not even the lesser custom models.

That's the problem. Enthusiasts love to OC.

That's a big win for custom 980Ti.

 

[LTC8K6]

All the water cooled hype, and it might be the air cooled version that shines?

 

[Noctifer616]

Does anyone commend AMD for this higher performance bigger die with less power consumption than 290x? Or can this simply be attributed to HBM and a top end 56C water cooling temp? Cooler temps do reduce power consumption. Thoughts?

Extremely difficult to say for sure. In order to figure that out we would need power consumption from both the memory chips and memory controller and this is a number only AMD could potentially give us.

If there are two version of the Fiji chip, one with HBM and one with GDDR5 then the picture would be clearer, but like this there really is no way to tell.

 

[YBS1]

Hardware Canucks claim big pump noise

https://www.youtube.com/watch?v=rfCb6oiJ6EI

claim AMD is aware and have made changes to production pump but if that doesn't get fixed it will KILL sales.

I'm going to go ahead and make a WAG that they simply received a noisy unit, you see those complaints occasionally with all the CLC CPU units but the vast majority are nearly silent. Probably not an issue with most Fury cards.

 

[Pariah]

So where are the massive over clocks that AMD boasted about in their press conference? If it's case of the tools not available, one has to wonder why not? Another failure to deliver on it's claims.

 

[IEC]

Hardware Canucks claim big pump noise

https://www.youtube.com/watch?v=rfCb6oiJ6EI

claim AMD is aware and have made changes to production pump but if that doesn't get fixed it will KILL sales.

The only time a CLC AIO makes a lot of noise is when you have a defective unit. They are seriously weak pumps otherwise.

Pretty sure other sites would have mentioned a problem of that magnitude. Sounds like they got a lemon.

 

[LTC8K6]

So where are the massive over clocks that AMD boasted about in their press conference? If it's case of the tools not available, one has to wonder why not? Another failure to deliver on it's claims.

I really don't understand boasting about overclocks and then having no overclocks to talk about on release.

 

[iiiankiii]

I guess H wont be receiving any more review samples. That was a way harsh review I must say. Saying AMD's GPU progrss is now more closely resembling it's CPU progress was a bit too far.
Does anyone commend AMD for this higher performance bigger die with less power consumption than 290x? Or can this simply be attributed to HBM and a top end 56C water cooling temp? Cooler temps do reduce power consumption. Thoughts?

Holy cow. H was freaking harsh. lol. But, you know what? They're right. At $650, it is second rate to the GTX 980 TI. AMD didn't expect the GTX 980 ti to spoil this much of the fun for AMD.

AMD could have priced the Fury X at $799 and that will actually look good against the Titan X. Of course, Nvidia pulled a fast one on AMD with the pricing of the GTX 980 TI. At $650, to me, there's no question about it. I would pick the GTX 980TI over the Fury X. More VRAM, and better overclockability means a win for the GTX 980TI. Well done, Nvidia.

The big question now, since we know what the Fury X offers, what can the Fury Pro do? Yes, I expect it to slower; maybe 10% than a Fury X?. But, how will it be priced? If it priced correctly, it MIGHT be compelling. I think the Fury Pro @ $500 would be a much better deal than than the Fury X.

 

[Head1985]

So fury is massive Rops bottleneck + slow tessellation.Thats why its only 20-30% faster than 390x
That was big mistake only 64rops

FURY X vs 390x
46% more SP
30% more memory bandwidth
45% more TMU
Same rops only 64

 

[Silverforce11]

HWC:

"In UHD scenarios this card is easily able to trade blows with the GTX 980 Ti (if it wasn’t for a lack of optimizations in Dying Light, it would actually be ahead by 2%) and gives the uber expensive TITAN X a swift kick in the pants."

Finally a reviewer mentions some problems with optimizations in certain NV games, as posted above, Fury X per CB.de is 51-56% above R290X. That would have been enough to beat Titan X a few months ago. But NV has gained a nice lead with more sites adding GW.

AMD needs to work harder to optimize those titles asap for sure.

 

[iiiankiii]

So fury is massive Rops bottleneck + slow tessellation.Thats why its only 20-30% faster than 390x
that was big mistake only 64rops

FURY X vs 390x
46% more SP
30% more memory bandwidth
45% more TMU
Same rops only 64

Seriously.. wth AMD. are they serious? lol. Way to shoot yourself in the foot.

 

[Silverforce11]

http://www.computerbase.de/2015-06/amd-radeon-r9-fury-x-test/12/#abschnitt_tessellationleistung

Tessellation is improved a lot. I don't trust Tech Report sorry, not after their shilling.

 

[cbrunny]

https://www.youtube.com/watch?v=-CDFNOTZy8o

Linus' card was DOA. I smell production problems....

 

[flopper]

AMD needs to work harder to optimize those titles asap for sure.

early drivers is obvious.

Voltage lock seems at this point strange even for me as those cards should been able to add voltage to them at the reviews.

 

[imaheadcase]

I find lots of people saying they simply won't get AMD cards because of driver issues never getting fixed.

It would not surprise me lots of people go nvidia simply for certain games that are better on it, i never saw the fuss over people crying foul how one developer makes a game better on one hardware than the other. That is a good thing.

 

[imaheadcase]

Why do these sites test BF4 or Hardline in DX11 mode when Mantle for those games are clearly superior in frame rate and smoother in frame times?

It never has been superior. I remember the day they released the mantle patch, and it was quickly remove because of graphic issues, then put back in with no better fps.

 

[ShintaiDK]

Why do these sites test BF4 or Hardline in DX11 mode when Mantle for those games are clearly superior in frame rate and smoother in frame times?

Same reason why Tonga users use DX11 over Mantle. It just works like crap with GCN 1.2.

 

[nvgpu]

https://www.youtube.com/watch?v=-CDFNOTZy8o

9:34, pump noise mentioned also and he also confirmed many other reviewers also have the same issues.

 

[DarkKnightDude]

The frametime reports over at guru3d is very impressive. Of course nobody cares about it now lol, was only relevant when certain parties wanted it measured.

 

[Alatar]

So aside from the SFF angle is there a reason to buy a Fury X when only reference models are allowed and for $20 more you can get a screaming fast custom 980Ti with more memory?

 

[Azix]

https://www.youtube.com/watch?v=-CDFNOTZy8o

Linus' card was DOA. I smell production problems....

lol. That will happen if he got second or third hand. Review samples jumping around isn't exactly ideal.

 

[Fire&Blood]

Didn't do as well as I had hoped. Postponing my 4K plan til Arctic Islands/Pascal. 980ti still a viable alternative but I've started hating 28nm with a passion over the past few weeks.

 

[exar333]

Disapointed.

If you know the card is being released, why don't your provide the correct overclocking tools to the reviewers.

That card is crap to my eyes for now. It would have been a win if it was close to the Titan x in most scenarios but it is not the case... It is below the GTX 980ti in most cases.

This definitely.

I don't remember cases in the past where the tools weren't available to OC. Unfortunately, the OC ability is a key purchasing decision because there are some great 980Ti custom models available now or coming soon.

 

[Flapdrol1337]

The frametime reports over at guru3d is very impressive. Of course nobody cares about it now lol, was only relevant when certain parties wanted it measured.

Haven't all single gpu's been perfect recently?