AMD R9 380X Review & Benchmark: Red Strides Forward

By Published November 19, 2015 at 7:58 am

Additional Info

  • Component: Video Card
  • Original MSRP: 240
  • Manufacturer: AMD

“Team Red” appears to have been invigorated lately, inspired by unknown forces to “take software very seriously” and improve timely driver roll-outs. The company, which went about half a year without a WHQL driver from 2H14-1H15, has recently boosted game-ready drivers near launch dates, refocused on software, and is marketing its GPU strengths.

The newest video card from AMD bears the R300 series mark, from which we previously reviewed the R9 380 & R9 390 GPUs. AMD's R9 380X 4GB GPU costs $230 MSRP, but retails closer to $240 through board partners, and hosts 13% more cores than the championed R9 380 graphics card (~$200 after MIRs). That places the R9 380X in direct competition with nVidia's GTX 960 4GB, priced at roughly $230, and 2GB alternative at $210.

Today, we're reviewing the Sapphire Nitro version of AMD's R9 380X graphics card, including benchmarks from Battlefront, Black Ops III, Fallout 4, Assassin's Creed Syndicate, and more. The head-to-head would pit the R9 380X 4GB vs. the GTX 960 4GB, something we've done in-depth below. We'll go into thermals, power consumption, and overclocking on the last page.

Sapphire R9 380X 4GB Video Card Review [Video]

AMD R9 380X Specs & Sapphire Nitro R9 380X AMD R9 380X AMD R9 380 AMD R9 390 AMD R9 390X
Process 28nm 28nm 28nm 28nm
Stream Processors 2048 1792 2560 2816
Boosted Clock 970MHz 970MHz 1000MHz 1050MHz
TMUs 128 112 160 176
Texture Fill-Rate 124.26GT/s 108.64GT/s 160GT/s 184.8GT/s
ROPs 32 32 64 64
Z/Stencil 128 128 256 256
Memory Configuration 4GB GDDR5 2 & 4GB GDDR5 8GB GDDR5 8GB GDDR5
Memory Interface 256-bit 256-bit 512-bit 512-bit
Memory Speed 5.7Gbps 5.5-5.7Gbps 6Gbps 6Gbps
Memory Bandwidth 182.4GB/s 182.4GB/s 384GB/s 384GB/s
Power 2x6-pin 2x6-pin 1x8-pin
TDP 190W 190W 275W 275W
API Support DX12, Vulkan,
DX12, Vulkan,
DX12, Vulkan,
DX12, Vulkan,
Launch Price 230-$240 $200.00 $330.00 $430.00

Note: Our R9 380X comes from Sapphire, who've overclocked the GPU core clock to 1040MHz from the reference 970MHz. The memory clock is pre-overclocked to 1500MHz against the 1425MHz reference. There is only a 4GB model of the 380X; a 2GB version does not exist.

TDP of the R9 380X rests at 190W, equivalent to the reference TDP of the R9 380, though we'll benchmark total system power draw in the final words of this review. The memory interface operates on a 256-bit wide bus and at 5.7Gbps reference speed, outputting a memory bandwidth of 182.4GB/s. Overclocked cards will slightly increase these numbers.

TMUs are the main increase over the R9 380, aside from a generally higher clockrate from AIBs. The R9 380X is outfitted with 128 texutre mapping units (+16 over the 380's 112 TMUs), but ROPs remains the same – 32 for each.

The same 28nm fab process we've come to know is still used in the R9 380X.

AMD R300 Series Architecture


AMD's using the same architecture we've seen a few times now, so no big news here. The only point that's worth reviving, though, is AMD's commitment to DirectX 12 and new APIs, like OpenGL Next (“Vulkan”). This is something we discussed with Chris Roberts of Cloud Imperium games recently, for followers of Star Citizen.

In its 380X press materials, AMD brought tessellation and tiled resource management to the forefront of discussion. With DirectX 12's takeover imminent, GCN architecture has the potential to make a stronger showing by way of its Asynchronous Compute Engines (ACE), designed for asynchronous shaders in DirectX 12. Asynchronous shaders overhaul the pipeline in a way that splices sizable workloads into multiple smaller parts. These are then processed (asynchronously, as the name would suggest) to accelerate the pipeline, ensuring that no single resource is holding-up the processing of others. Larger resources can be worked on in the background while the small stuff gets managed throughout operation.


Support of Dx12's asynchronous processing isn't new to the 380X, but is a strength to which that AMD has aptly called attention.

As stated in the initial R300 series launch, some overall changes have been made to the architecture to push this line beyond “rebrand” and into “refresh” territory. Primarily, those changes are summed-up in the form of thermal envelope reduction and a slightly bolstered clockrate over the R200 series equivalents. This makes for reasonable gains in performance when cards are coupled with stable drivers – something we didn't have for the R9 390 at launch. We'll talk about AMD's driver status further down.

Sapphire R9 380X Cooler Design


Sapphire's R9 380X uses the company's “Nitro” cooler design, introduced as a “gamer-targeted” brand alongside the R300 series. In Sapphire's vision, Nitro cards hit market close to MSRP and appeal most to gamers who don't care to do much tweaking or overclocking, not to say the cards can't overclock – we'll test that later – but they aren't targeted at overclockers.

The Sapphire R9 380X uses four copper heatpipes routed through an aluminum finned sink, fairly standard, and is fitted with two ~90mm fans for dissipation. The two fans have a sort of “cheap plastic” feel to them, but the cooler overall is relatively high-quality. This is something we discussed in the R9 380/390 Nitro reviews.

Sapphire's mounted a backplate to their 380X – with the words right-side up, even – and uses a futuristic white/black/gray color scheme. We like the overall styling of the backplate, but it is generally not of this website's interest to discuss aesthetics; we'll leave the rest to you, told through photos in the post.

Test Methodology

We tested using both of our GPU test benches, as we are in the process of merging into a single GPU test bench. Both are detailed in the tables below. Our thanks to supporting hardware vendors for supplying some of the test components.

The latest AMD Catalyst drivers (15.11.1) were used for testing, except in the case of the Fury X and R9 380/390, which were all on loan and have not been updated since ~15.7. NVidia's 359.00 drivers were used for testing the latest games. Game settings were manually controlled for the DUT. All games were run at presets defined in their respective charts. We disable brand-supported technologies in games, like The Witcher 3's HairWorks and HBAO. GRID: Autosport saw custom settings with all lighting enabled. All other game settings are defined in respective game benchmarks, which we publish separately from GPU reviews. Our test courses, in the event manual testing is executed, are also uploaded within that content. This allows others to replicate our results by studying our bench courses.

Each game was tested for 30 seconds in an identical scenario, then repeated three times for parity.

Z97 bench:

GN Test Bench 2015 Name Courtesy Of Cost
Video Card

This is what we're testing!

- -
CPU Intel i7-4790K CPU CyberPower
Memory 32GB 2133MHz HyperX Savage RAM Kingston Tech. $300
Motherboard Gigabyte Z97X Gaming G1 GamersNexus $285
Power Supply NZXT 1200W HALE90 V2 NZXT $300
SSD HyperX Predator PCI-e SSD Kingston Tech. TBD
Case Top Deck Tech Station GamersNexus $250
CPU Cooler Be Quiet! Dark Rock 3 Be Quiet! ~$60

X99 Bench:

GN Test Bench 2015 Name Courtesy Of Cost
Video Card

This is what we're testing!

- -
CPU Intel i7-5930K CPU iBUYPOWER
Memory Kingston 16GB DDR4 Predator Kingston Tech. $245
Motherboard EVGA X99 Classified GamersNexus $365
Power Supply NZXT 1200W HALE90 V2 NZXT $300
SSD HyperX Savage SSD Kingston Tech. $130
Case Top Deck Tech Station GamersNexus $250
CPU Cooler NZXT Kraken X41 CLC NZXT $110

Average FPS, 1% low, and 0.1% low times are measured. We do not measure maximum or minimum FPS results as we consider these numbers to be pure outliers. Instead, we take an average of the lowest 1% of results (1% low) to show real-world, noticeable dips; we then take an average of the lowest 0.1% of results for severe spikes. Anti-Aliasing was disabled in all tests except GRID: Autosport, which looks significantly better with its default 4xMSAA, and Black Ops III. HairWorks was disabled where prevalent. Manufacturer-specific technologies were used when present (CHS, PCSS).

The Assassin's Creed: Syndicate methodology will go live shortly after this post, but to recap, we tested these settings:

  • 1080p / ultra high: MSAA2X + FXAA enabled. SSAO (not HBAO+) and “High” shadows configured to ensure fair benchmarks.
  • 1080p / ultra custom: FXAA instead of MSAA. SSAO. “High” shadows.
  • 1440p / ultra custom: as above.
  • 4K / ultra custom: as above.

Overclocking was performed incrementally using MSI Afterburner. Parity of overclocks was checked using GPU-Z. Overclocks were applied and tested for five minutes at a time and, if the test passed, would be incremented to the next step. Once a failure was provoked or instability found -- either through flickering / artifacts or through a driver failure -- we stepped-down the OC and ran a 30-minute endurance test using 3DMark's FireStrike Extreme on loop (GFX test 2).

Thermals and power draw were both measured using our secondary test bench, which we reserve for this purpose. The bench uses the below components. Thermals are measured using AIDA64. We execute an in-house automated script to ensure identical start and end times for the test. 3DMark FireStrike Extreme is executed on loop for 25 minutes and logged. Parity is checked with GPU-Z.

Thermals, power, and overclocking were all conducted on the Z97 bench above.

Video Cards Tested

Prev Next »

Last modified on November 19, 2015 at 7:58 am
Steve Burke

Steve started GamersNexus back when it was just a cool name, and now it's grown into an expansive website with an overwhelming amount of features. He recalls his first difficult decision with GN's direction: "I didn't know whether or not I wanted 'Gamers' to have a possessive apostrophe -- I mean, grammatically it should, but I didn't like it in the name. It was ugly. I also had people who were typing apostrophes into the address bar - sigh. It made sense to just leave it as 'Gamers.'"

First world problems, Steve. First world problems.

We moderate comments on a ~24~48 hour cycle. There will be some delay after submitting a comment.

  VigLink badge