Pros

Great 1440p performance; efficient, especially compared to previous Radeon cards; ultra-quiet fans

Cons

Huge; coil whine is noticeable; limited overclocking headroom; not price-competitive for 1080p gaming

Star Rating

R9 Fury

Introduction

Entering June 2015, all eyes were on AMD, as rumors had been flying regarding its next big video card launch. It finally made a big splash on June 24th in announcing its liquid-cooled R9 Fury X model. In using a catchy premium naming scheme, it borrowed a page out of Nvidia's marketing handbook. Of course, Nvidia wasn't going to let AMD have all the fun. In early June, it preemptively launched its Fury fighter, the comparatively humble-sounding GTX 980 Ti. It set a bar that was a bit too high for the Fury X to match, and so enthusiasts held out hope that a card one step below the Fury X would pack more of a bang-for-the-buck punch, as is often the case in AMD's Radeon lineups. That card arrived in July 2015 in the form of the is air-cooled Radeon R9 Fury. It's far larger than the Fury X, but in foregoing the liquid hoses and radiator, it's a card that's a bit easier to handle overall. In today's review, we'll be looking at Sapphire's version of the Fury, the R9 Fury Tri-X.

We'd like to extend a special thanks to B&H Photo Video for providing a retail sample of the Sapphire Radeon Fury Tri-X 4GB for review. We'll be benchmarking this card against several worthy competitors, as follows:

  1. Sapphire Radeon R9 290 Tri-X 4GB at reference clocks of 947MHz/5000MHz
  2. Sapphire Radeon R9 290 Tri-X 4GB emulating a Radeon R9 390 8GB at clocks of 1000MHz/6000MHz
  3. EVGA GeForce GTX 980 ACX 4GB, at reference in-game clocks of 1265/7000MHz
  4. EVGA GeForce GTX 980 Ti ACX 2.0+ 6GB, at reference in-game clocks of 1202MHz/7000MHz

You can see the R9 Fury Tri-X pictured here, along with the comparatively puny EVGA GTX 980 Ti ACX (the 980 looks idential, by the way). Of course, the 980 Ti packs a punch that the Fury can't quite match, being a significantly more expensive card, after all. A more direct competitor might actually be the older GTX 980, which dropped in price from $550 to $500 not long before the Fury arrived, in part to differentiate itself from the competition.

We'll be providing benchmarks at both the typical 1920 x 1080 resolution, as well as the increasingly-popular 2560 x 1440 resolution. Note that for a few of our benchmarks, we also included data for an older high-end card, the GTX 780 Ti 3GB. This should be helpful to anyone looking to make an upgrade from an older Nvidia Kepler-based card.

The test bench used consisted of the following components:

We used Catalyst version 15.7.1 for our Radeon cards and GeForce version 353.62 for our Nvidia cards.

Description and Features

Rear View

 

The Radeon R9 Fury Tri-X is a massive video card, and in fact it's the largest card we've ever tested. It measures 12.2" long, 4.4" tall, and 2" wide. Its height and width both extend beyond the standard PCI slot dimensions, meaning it may be too tall for some cases, and more importantly doesn't qualify as a two-slot card. Not that you'd want to run two open-air cards like this on 4-slot micro ATX board, but with this particular model, that wouldn't even be possible. The Fury Tri-X tipped the scales at just over 2.5lb., beating even its fairly portly cousing, the R9 290 Tri-X, which came in at 2.25lb on our scale. You'll want to make sure the card is properly secured in your case, and moving the case around with the card mounted would most definitely not be a good idea.

Top View

Most of this card's mass results from its truly epic heatsink. As you'll see later in this review, Sapphire probably overdid it with this cooler, as its performance is simply outstanding. The card has a metal backplate covering the PCB, but as is quite obvious, the heatsink extends well beyond the compact circuit board, providing the unusual benefit of a direct exhaust through the back of the card. As is fitting for a card of this caliber, the Fury Tri-X is equipped with dual 8-pin PCIe power inputs, which together deliver 300W of power, in addition to the 75W provided via the motherboard PCIe slot.

We should note that one of the reasons that the circuit board is so small on this card is the use of stacked high-bandwidth memory, the crowning achievement in AMD's new "Fiji" GPU architecture. Operating at just 500MHz, this memory is hooked to an amazingly-wide 4096-bit bus, allowing for record-breaking memory bandwidth of 512GB/s. The core operates at 1000MHz, down slightly from the 1050MHz of the Fury X, and with 3,584 stream processors on board, its 12.5% behind the Fury X in that regard.

Enough with the specs, let's get to our benchmarks, shall we?

Next page