Sapphire Radeon R9 Fury Tri-X OC 4GB Review

Performance Analysis

The first title of the bit-tech test suite, Alien: Isolation, starts the Sapphire R9 Fury Tri-X OC off in good standing. At all resolutions the R9 Fury is close behind the R9 Fury X and at 1440p and 4K there's a healthy margin of difference between the R9 Fury and the Nvidia GTX 980. Battlefield 4 reveals something similar though the R9 Fury is behind the GTX 980 at 1080p. At those higher resolutions the R9 Fury comes into its own with a growing lead over the GTX 980 and again the observed difference between R9 Fury and Fury X is very narrow.



Crysis 3 delivers a strong showing for the Sapphire R9 Fury Tri-X OC with a consistent performance advantage over the GTX 980 while the gap between the R9 Fury X is literally one or two frames at all resolutions. Grand Theft Auto V has similar results to Battlefield 4 in that the R9 Fury sits behind the GTX 980 at 1080p but charges off into the lead through 1440p and 4K.

Middle-earth: Shadow of Mordor loves the pixel-crunching power of AMD's GCN architecture and the R9 Fury maintains a healthy lead over the GTX 980 across the board. At 4K the Sapphire R9 Fury Tri-X OC is within touching distance of the Nvidia GTX 980 Ti though the large difference in overclocking headroom means the R9 Fury isn't quite in the same performance class. Witcher 3: Wild Hunt shows the R9 Fury to deliver a faster average framerate than the GTX 980 at all resolutions but, as is the repeating trend, that gap increases towards 4K.



Unigine Valley, the last of our 3D tests, demonstrates the closeness of the Sapphire R9 Fury Tri-X OC and stock Fury X once again. The gap is around 5 per cent, but taking into consideration Sapphire's factory overclock the stock vs. stock gap is more in the 10 per cent region. Sapphire's R9 Fury loses some of the power efficiency gains of the R9 Fury X, a result we believe stems from the use of air cooling. R9 Fury X was able to save around 20 to 30W of power by reducing leakage with a watercooling solution; the R9 Fury loses that benefit but temperatures are still respectable. Sapphire's cooling solution runs passively at idle and keeps temperatures in the low-to-mid 70s at gaming load.

Looking at some performance summary statistics we find the Sapphire R9 Fury Tri-X OC to be 2.1, 11.2 and 15.8 per cent faster than the GTX 980 at 1080p, 1440p and 4K, respectively, across the average framerates of our gaming benchmarks. The same comparison to the reference AMD R9 Fury X shows the Sapphire R9 Fury Tri-X is 3.4, 4.5 and 6.5 per cent slower than its bigger brother.

Conclusion

Sapphire will be coming to market with a duo of R9 Fury graphics cards of which one will sport a reference clock speed of 1,000MHz while the other will be an overclocked version running at 1,040MHz - the model we have tested today. There will be a small difference in price of $20 (£15) between the two models but aside from clock speeds they are identical in that they use the same PCB and cooling solutions.

In general terms the R9 Fury turns out to be an impressive graphics card for 1440p and 4K gaming. If the AMD R9 Fury X or Nvidia GTX 980 Ti are just out of budget then the R9 Fury is a lower-cost alternative that still delivers competitive performance. In terms of price-to-performance positioning in the marketplace the R9 Fury isn't too dissimilar to the Nvidia GTX 980, because it does generally outperform it though with a $70 higher MSRP, less overclocking headroom and considerably higher power consumption. In that regard the R9 Fury isn't as compelling as some may have hoped... and for all intents and purposes the GTX 980 is a more refined and well-rounded solution.

The R9 Fury would benefit from an MSRP of $499, as opposed to $549, to ensure there is a price-to-performance win for AMD against the GTX 980. That said, the R9 Fury is still a good buy at its current price point, particularly Sapphire's Tri-X OC variant which delivers a cool-and-quiet experience thanks to some design innovations in the cooling arena. Paying extra for the slightly overclocked variant doesn't seem overly worthwhile given that most Fiji silicon will top out at a similar frequency of 1,130~1,150MHz, thus prospective buyers are better off acquiring the stock clocked model and overclocking manually.