AMD Radeon R9 390 & R9 380 Specs

As discussed in the review, the R9 390 and R9 380 are effectively identical to their preceding components (the R9 290 and R9 285, respectively). The table above reflects this. Core counts, TMUs, ROPs, even the GPU itself are all the same – it's the memory capacity and clockrate that have changed, with some “tuning” to the power management engine. Astute readers will also notice that the pixel and texel rates have increased on the R9 300 series cards – this is a function of the 50MHz clock increment, as texel rate is calculated as a product of the clockrate and TMUs.

Beyond the memory and clock jumps, there are no changes of note for buyers. Owners of the 200 series have little business upgrading, though new buyers may be in the running for a 300 series device. The Fury devices, which will ship with the new Fiji architecture, are presently priced out of the range of these 300 devices, making them non-competitors to AMD's own lower-end products. The only direct competition to the R9 380 is the GTX 960 4GB ($239), and the R9 390 faces competition from the GTX 970 ($350).

AMD R9 300 Series Overclocking

Although AMD overclocking can be forced down through MSI Afterburner, the software doesn't have full access to all the tuning settings. AMD's Catalyst drivers have a built-in overclocking utility that allows for increases to power in percentages, similar to the Maxwell overclocking we've already discussed.

Through usage of AMD's targeting reticle, we're able to manipulate the power percentage over (or under) reference as supplied to the GPU. This directly impacts the amount of power the GPU will drink through its PSU connectors, and directly impacts the resulting clockrates and stability of the overclock. Attempting to OC just the core clocks and memory clocks without change to the power allotment, for instance, will result in a stifled OC that is throttled by power availability. Too little power, and the GPU will ignore your overclock frequency, instead pushing it as high as is sustainable with the power available. Power is, like on Maxwell, split between voltage and the clock.

Memory overclocking is done on the slider that rests beneath the green-red OC gradient. The Sapphire R9 390 we used is limited to 2000MHz on the memory clock (theoretical max) and +50% on the power target. Sapphire's loaned R9 380 is limited to ~1700-1800MHz on the memory clock (theoretical max) with a +20% power target limit.

For clarity, increasing the power by X% over 100% means that we're increasing the power over reference. Whenever overclocks, overvoltage attempts, and over-power attempts are applied to a GPU, there is an inherent risk of destroying the silicon. Overclocking increases thermals and threatens stability, and should only be performed to levels with which the user feels comfortable. In our case, we're generally working with review hardware, so it's our job to try and find a breaking point. We'd encourage greater caution when pushing upper limits for consumer devices.

Test Methodology

We tested using our updated 2015 GPU test bench, detailed in the table below. Our thanks to supporting hardware vendors for supplying some of the test components. Thanks to Jon Peddie Research for GTX 970 & R9 280X support.

The latest GeForce 353.06 drivers were used during testing. AMD Catalyst 15.15 was used for the 300 series cards, with 15.5 used for 200 series cards due to incompatibility. Game settings were manually controlled for the DUT.

VRAM utilization was measured using in-game tools and then validated with MSI's Afterburner and AIDA64, a custom version of the Riva Tuner software. Parity checking was performed with GPU-Z. FPS measurements were taken using FRAPS and then analyzed in a spreadsheet.

Each game was tested for 30 seconds in an identical scenario on the two cards, then repeated for parity.

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.

We conducted a large suite of real-world tests, logging VRAM consumption in most of them for comparative analysis. The games and software tested include:

• Far Cry 4 (Ultra 1080, Very High 1080).
• GRID: Autosport (Ultra 1440, Ultra 4K).
• Metro: Last Light (Very High + Very High tessellation at 1080; High / High at 1440).
• GTA V (Very High / Ultra at 1080p).
• Shadow of Mordor (Very High, 1080p).
• 3DMark Firestrike Benchmark
• GTA V
• The Witcher 3

We already know GTA V and Far Cry 4 consume massive amounts of video memory, often in excess of the 2GB limits of some cards. GRID: Autosport and Metro: Last Light provide highly-optimized benchmarking titles to ensure stability on the bench. Shadow of Mordor, GTA V, & The Witcher 3 are new enough that they heavily eat VRAM. 3DMark offers a synthetic benchmark that is predictable in its results, something of great importance in benchmarking.

Overclocked tests were conducted using EVGA Precision for application of settings to nVidia devices. AMD OverPower was used for AMD overclocking. All devices were tested for performance, stability, and thermals prior to overclocking to ensure clean results. On the OC bench, devices were set to maximize their voltage ceiling with incremental gains applied to the core clock (GPU) frequency. 3DMark Firestrike was used on infinite loop to determine stability in a relatively real-world scenario. Once stability was compromised -- either from crashing or artifacting -- we attempted to resolve the issue by fine-tuning other OC settings; if stability could not be achieved, we backed-down the core clock frequency until we were confident of stability. At this point, the device was placed on a burn-in test using 3DMark Firestrike for 25 minutes. If the settings survived this test without logged fault, we recorded the OC settings and logged them to our spreadsheet.

Final OCs were applied and tested on games for comparison.

Thermals were reported using Delta T over ambient throughout a 25-minute burn-in period using 3DMark FireStrike - Extreme on infinite loop, which renders graphics at 1440p resolution. This test loads the VRAM heavily, something Kombustor skips, and keeps the GPU under high load that is comparable to gaming demands. Temperatures were logged using AIDA64.

Continue to Page 2 for the AMD 300 overclocking results.