AMD Radeon R9 Fury X Reviews

[Rys]

As penance for deleting it earlier, I now have RSI and had to read the whole thread from start to finish again, to manually copy the posts from the backup to here. Sorry the posts don't have original attribution: the forum software doesn't support selectively restoring threads from a backup copy of the database.

 

[Ryan Smith]

Did something go wrong here? I'm seeing page after page of Rys posts, all made within minutes of each other.

 

[Rys]

Yes, I accidentally hard deleted the whole thread earlier today, while moderating. The forum software has no way to selectively restore content, so I had to manually copy the posts into a new thread by hand, from the backup server.

It was either that or lose all the posts made in the time between me deleting the thread while out and about, and being back at home in front of a computer.

 

[BRiT]

Did something go wrong here? I'm seeing page after page of Rys posts, all made within minutes of each other.

Here's the thread concerning this thread -- https://forum.beyond3d.com/threads/deleted-thread-radeon-r9-fury-x-reviews.57165/

 

[tabs]

Holy crap Rys. Thanks for the effort!

 

[UniversalTruth]

As penance for deleting it earlier, I now have RSI and had to read the whole thread from start to finish again, to manually copy the posts from the backup to here. Sorry the posts don't have original attribution: the forum software doesn't support selectively restoring threads from a backup copy of the database.

Rys, I am honestly sorry and apologise for causing this inconvenience.
I hope you will forgive me.

 

[UniversalTruth]

What do you guess is a necessry quantity for a 175 watt card with unknown characteristics that only has it's size to set it apart a bit?

Hmm, why do you say "unknown"? Do you have another information different to all the data from articles showing R9 Nano to feature the full Fiji die at lower frequencies?

And then, the R9 Nano will be the most important part. I expect from it rather decent quantities in order to saturate mainstream demand.

We will see, R9 Fury X might be too expensive, but R9 Nano could turn out the most appealing product.

 

[AlexV]

It is contradictory to claim that, on one hand, they are cherrypicking dies for a particular SKU and thus making their most expensive SKU scarce, while on the other the SKU they're doing the cherrypicking for will be cheap. You would be forgiven for assuming that AMD is not exactly a very business savvy entity, given their recent performance, but still that is rather unintuitive. Furthermore, historical evidence points to powerful mITX cards as being overpriced, and not underpriced, compared to their less petite siblings (see, e.g., the mITX 285 from Sapphire). I am willing to give AMD the benefit of doubt and assume that if they are indeed aggressively sieving dies, they are doing so because they'll have a product that at least does not erode their margin even further. Finally, strictly in terms of volume, it would be suprising if the Nano ends up as a very high volume affair.

 

[Kaarlisk]

It is contradictory to claim that, on one hand, they are cherrypicking dies for a particular SKU and thus making their most expensive SKU scarce, while on the other the SKU they're doing the cherrypicking for will be cheap.

It may not be cherrypicking. They might simply be keeping those few dies that do not clock high and have low power consumption.

 

[AlexV]

It may not be cherrypicking. They might simply be keeping those few dies that do not clock high and have low power consumption.

Which makes the theory that the Fury X is missing from stores because of the Nano somewhat flaky. Furthermore, there are multiple kinds of cherries to pick...

 

[Kaarlisk]

Which makes the theory that the Fury X is missing from stores because of the Nano somewhat flaky. Furthermore, there are multiple kinds of cherries to pick...

Whoops Yeah, I somehow concentrated on one part of the statement and missed the rest
So, to make up for it... maybe Nano will actually have higher margins than X, because while it's price will be lower, it will also get rid of the expensive cooler and the expensive power delivery circuity (just kidding).

In all seriousness, I do believe that there is a reason AMD is choosing to do this part, both on the supply and demand side. Maybe the 390X is getting hurt by the power disadvantage (demand) while AMD are getting enough dies that have all of the functional units that do not clock high (supply), and the Nano element is just an additional cherry. Maybe they simply are getting enough dies that do not clock high that it is worth the nonrecurring expenses to create a third SKU. Maybe the demand for the two existing SKUs is low enough that they want to create a third, cheaper or different SKU, but will only release it when they want to/have to move Hawaii lower on the stack.

 

[Alexko]

If Fiji were a "normal" GPU, as opposed to the very large one that it is, then AMD would take low-leakage parts and sell them as laptop boards. This is essentially what the Fury Nano is, except that even with low leakage and lower clocks, it's still too power-hungry for laptops, so it ends up being a small desktop board.

 

[AlexV]

Whoops Yeah, I somehow concentrated on one part of the statement and missed the rest
So, to make up for it... maybe Nano will actually have higher margins than X, because while it's price will be lower, it will also get rid of the expensive cooler and the expensive power delivery circuity (just kidding).

In all seriousness, I do believe that there is a reason AMD is choosing to do this part, both on the supply and demand side. Maybe the 390X is getting hurt by the power disadvantage (demand) while AMD are getting enough dies that have all of the functional units that do not clock high (supply), and the Nano element is just an additional cherry. Maybe they simply are getting enough dies that do not clock high that it is worth the nonrecurring expenses to create a third SKU. Maybe the demand for the two existing SKUs is low enough that they want to create a third, cheaper or different SKU, but will only release it when they want to/have to move Hawaii lower on the stack.

Don't get me wrong, Nano is interesting (perhaps the most interesting SKU in the Fury family, at least for someone like me), but I don't expect it to be a volume / cheap play. More along the lines of trying to carve / explore a niche, while exploiting one of the constructive advantages they hold (PCB compactness). If there is a sustained migration towards compact enclosures / rigs (a la the Steam machine idea), it could technically be a lucrative niche, since people in that space are ready to accomodate premiums (see pricing for mITX motherboards, SFX PSUs etc.).

 

[pjbliverpool]

Holy crap Rys, that's one long conversation with yourself!

Seriously though, thanks for the monumental effort there. At least it wasn't one of the uber threads!

 

[UniversalTruth]

http://www.techpowerup.com/reviews/ASUS/R9_Fury_Strix/7.html

All games where R9 Fury X is faster at 4K - in 12 out of 22 games:

Alien: Isolation
R9 Fury X 62.2 FPS
GTX 980 Ti 62.0 FPS

Assassin's Creed: Unity
R9 Fury X 31.6 FPS
GTX 980 Ti 29.2 FPS

Battlefield 3
R9 Fury X 52.8 FPS
GTX 980 Ti 49.5 FPS

Bioshock Infinite
R9 Fury X 63.2 FPS
GTX 980 Ti 60.2 FPS

Civilization: Beyond Earth
R9 Fury X 71.5 FPS
GTX 980 Ti 60.6 FPS

Crysis 3
R9 Fury X 28.2 FPS
GTX 980 Ti 26.6 FPS

Dead Rising 3
R9 Fury X 30.8 FPS
GTX 980 Ti 26.1 FPS

Far Cry 4
R9 Fury X 39.6 FPS
GTX 980 Ti 35.5 FPS

Ryse
R9 Fury X 41.5 FPS
GTX 980 Ti 40.5 FPS

Shadow of Mordor
R9 Fury X 52.7 FPS
GTX 980 Ti 49.9 FPS

The Witcher 3
R9 Fury X 32.8 FPS
GTX 980 Ti 32.6 FPS

Tomb Raider
R9 Fury X 58.7 FPS
GTX 980 Ti 53.4 FPS

 

[3dilettante]

I think this thread reveals the truth that we are all Rys. This is the forum trying to break through the illusion.

A lot of those wins are so close that I would consider them a draw, not that there's anything wrong with matching the competition. Going to the tenths place or single-digit percentages is such a small amount in an absolute sense that anyone running a somewhat different setup or choosing one slightly different setting is going to swamp the difference.
It might go to personal preference, but the sub-30 FPS or barely 30 examples verge on being more theoretical than tolerable for me.

 

[Razor1]

http://www.techpowerup.com/reviews/ASUS/R9_Fury_Strix/31.html

Same review when you tabulate the 4k performance of the entire game suite they used and there is a 1% difference in favor of the Ti what does that tell you about the Fury X? Same performance.at 4k but slower at lower resolutions. Its not faster, driver "may" yes "may" increase its performance in these games but right now Fury X is the same performance as Ti.

 

[Moloch]

May increase performance? I think it's pretty obvious over it's product life span the Fury's performance will improve with drivers, what would be truly surprising would be if they made it perform worse.

 

[Razor1]

true as does the competitors as well, waiting on drivers to improve performance is not the best way a company to be shown in good light when reviewers are reviewing cards now (well a few weeks ago). The only thing people can go by is what is the performance and features as of today

 

[Alexko]

Right, the question is whether the Fury X will gain more than the 980 Ti, but it's very difficult to predict.